De Maat, Proefschrift.Indd

On the effectiveness of psychoanalytic therapy

Short if possible, long if necessary? The printing was financially supported by: JellinekMentrum Amsterdam

Cover design: Annelies Frölke, Amsterdam Printed by: Ridderprint Offsetdrukkerij BV.

ISBN: 978-90-5335-140-6 NUR: 770

All rights reserved. No part of this publication may be reproduced or transmit- ted in any form or by any means, electronic or mechanical, including photo- copying, recording or any information storage and retrieval system, without written permission of the copyright owner. VRIJE UNIVERSITEIT

THE EFFECTIVENESS OF PSYCHOANALYTIC THERAPY

Short if possible, long if necessary?

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad Doctor aan de Vrije Universiteit Amsterdam, op gezag van de rector magnificus prof.dr. L.M. Bouter, in het openbaar te verdedigen ten overstaan van de promotiecommissie van de faculteit der Psychologie en Pedagogiek op vrijdag 14 december 2007 om 10.45 uur, in de aula van de universiteit, De Boelelaan 1105

door

Saskia Cornelia Maria de Maat

geboren te Arnhem promotor: prof.dr. J.J.M. Dekker copromotoren: prof.dr. F.E.R.E.R. de Jonghe dr. R.A. Schoevers

Beoordelingscommissie:

prof.dr. A.T.F. Beekman prof.dr. P. Cuijpers prof.dr. J.A. Swinkels prof.dr. W. van Tilburg dr. M.H.M. de Wolf Als eine Spezialwissenschaft, ein Zweig der Psychologie, – Tiefenpsychologie oder Psychologie des Unbewussten, – ist sie ganz ungeeignet, eine eigene Weltanschauung zu bilden, sie muss die der Wissenschaft annehmen.

(S. Freud, 1933) Contents

Chapter 1 introduction 9

Chapter 2 relative efficacy of psychotherapy and pharmacotherapy in the treatment of depression: a meta-analysis 23

Chapter 3 relative efficacy of psychotherapy and combined therapy in the treatment of depression: a meta-analysis 45

Chapter 4 short-term Psychoanalytic Supportive Psychotherapy for depressed patients 63

Chapter 5 support and personality change: A psychoanalytic view 75

Chapter 6 short Psychodynamic Supportive Psychotherapy, antidepressants, and their combination in the treatment of major depression: a mega-analysis based on three randomized clinical trials 93

Chapter 7 comparison of Short Psychodynamic Supportive Psycho- therapy, pharmacotherapy and their combination on subdimensions of the HDRS and subscales of the SCL-90: a mega-analysis of three randomized controlled trials regarding depressed patients 109

Chapter 8 the effectiveness of long-term psychotherapy: methodological research issues 125

Chapter 9 the effectiveness of long-term psychoanalytic therapy: a sytematic review of empirical studies 137

Chapter 10 costs and benefits of long-term psychoanalytic therapy: changes in health care use and work impairment 169

Chapter 11 General discussion 191

Summary 221

Samenvatting 229

Curriculum Vitae 241

List of publications 243

Dankwoord 245

Chapter 1

Introduction

1.1 Background

Psychoanalytic therapies have existed for over a hundred years. From the very beginning, they have been contentious: applauded by some and reviled by others. The present situation is not essentially different. Gabbard (2004) indicates the continued popularity of psychoanalytic therapies and concludes: ‘The thirst to understand, to ‘know thyself’, persists despite managed care, the quick- fix mentality of our society and the remarkable progress in psychopharmacology.’ (p. 1). The popularity among clinicians and patients alike is underscored by studies of Wiener (1994) and Luborsky et al. (1993). They show that, at least in the United States, ‘psychoanalytic psychotherapy still seems to be the most frequent intervention practiced by the majority of psychiatrists’ (Wiener, 1994). yet, criticisms too persist. Since we live in the era of evidence based medicine, they are concentrating on the argument that psychoanalytic therapies are not ‘scientifically proven’. The term ‘Evidence based medicine’ (EBM) was dubbed in the eighties of the last century, initially to indicate a new teaching method. Since then, many definitions have emerged, of which the following illustrates its most important aspects: ‘EBM is the conscious, explicit and expert use of the best available evidence to guide decisions regarding individual patients. EBM integrates the individual clinical expertise with the best available evidence of systematic research. The preferences, wishes and expectations of the patient play a central role in the decision making process.’(Offringa et al., 2000, p. 2-3). In the English language, the term ‘evidence’ is to be distinguished from ‘proof’, the latter meaning that there is little to no doubt about the truth of a conclusion, the first referring to an indication lending more or less strength to a judgement. In EBM the strength of scientific evidence is indicated by ‘levels of evidence’. The first level consists of a) a systematic review of Randomized Controlled Trials (RCTs) and b) a single high quality RCT. The second level consists of a) a systematic review of cohort studies and b) a single cohort- or patiënt-control study or a lower quality RCT. After these two, further levels are distinghuished, consisting of a systematic review of case-control studies, a single case-control study, case series and study expert’s opinions or generally accepted therapeutic methods (Center of Evidence Based Medicine Oxford, Levels of evidence, 2001). rcts derive their name from their most important characteristic, the randomization of patients. This study design is also called the ‘confirmatory-deductive methodology’ and is applied in ‘efficacy research’. RCTs strive for maximum internal validity, so that it can be assumed, beyond reasonable doubt, that differences found between treatment and control groups (the dependent

variable) are explained by the therapeutic intervention (the independent variable). Cohort studies follow well-defined groups of people over a certain period of time in order to observe whether certain ‘outcomes’ occur. Relations between ‘determinants’ and outcomes are scrutinized. If the determinant is an intervention, this design is also called a quasi-experimental study: ‘experiments that lack random assignment (…) but that otherwise have similar purposes and structural attributes to randomized experiments.’ (Shadish et al., 2002). Quasi- experimental studies may or may not include comparison groups. the advantages of randomization are obvious and justify the high ranking of RCTs within the hierarchy of empirical evidence. However, RCTs too have been criticized. Leichsenring (2005) states: ‘The exclusive position of RCTs as method for demonstrating that a treatment works has been recently queried (Seligman, 1995, Roth and Parry, 1997, Beutler, 1998, Henry, 1998, Persons and Silberschatz, 1998, Fonagy, 1999, Leichsenring, 2004, Westen et al., 2004). The main argument is that it is questionable whether the results of RCTs are representative of clinical practice.’ In RCTs, many patients are excluded, or they refuse participation or drop out. The manuals and protocols applied, the many assessments, and so on, poorly connect to daily patient care. In short, this criticism applies to the external validity of RCTs, i.e., on the generalizability of their results to daily practice. cohort studies can be and have been criticized as well. Although their external validity is often higher than that of RCTs, their internal validity is evidently weaker. Many feel that cohort studies tend to overestimate the effects of treatments, as there is no correction for potential confounders. First, it cannot be ruled out that the changes found in a cohort might be effectuated over time and not (just) by the treatment. Second, the comparability of the treated and the untreated group, or of two treated groups, may be doubted as they lack randomization. Therefore, the conclusive power of cohort studies is still a matter of debate. contrary to what many people think, psychoanalytic therapies have been investigated intensively. In fact, a review of Beutler and Crago (1991) found that of the forty major international psychotherapy research programs at that time, eighteen were mainly psychoanalytic in orientation. Doidge (1997) concludes his review of the empirical evidence for psychoanalytic psychotherapies with the statement that: ‘(…) the psychoanalytic therapies (…) not only have been, but continue to be, among the most intensively studied treatments.’ Briefer forms of psychoanalytic therapies have been investigated by means of RCTs, of which many reviews and meta-analyses provide the integrated results (Crits-Christoph, 1992, Leichsenring, 2005, Leichsenring, 2001, Anderson and Lambert, 1995, Leichsenring, 2003, Doidge, 1997). The findings lead to the conclusion that there is strong evidence that short-term psychoanalytic therapies perform equal to other short-term psychotherapies in a variety of disorders. Because of serious practical and ethical problems, longer psychoanalytic therapy has been investigated mainly in cohort studies. This means that the best available evidence for long-term psychoanalytic therapy consists of the second strongest possible level of evidence. Several reviews have been

10 performed to summarize the evidence for long-term psychoanalytic therapy (Bachrach et al., 1991: Doidge, 1997: Fonagy, 1999; Leichsenring, 2005), showing that there is evidence for the effectiveness of long-term psychoanalytic therapy. however, despite the evidence, a striking discrepancy between the enduring ‘belief’ in the effectiveness of psychoanalytic therapies within the psychoanalytic community and the tenacious ‘disbelief’ outside remains. A telling example is the fact that, in the Netherlands, psychoanalytic therapy is not even mentioned in the present day interdisciplinary guideline for depressive disorders (2005) nor in that for anxiety (2003). This discrepancy is what motivated the studies in this thesis, researching existing evidence and participating in a project digging for more evidence in this controversial field of psychoanalytical therapy.

1.2 Psychoanalytic therapies

Psychoanalytic therapies are rooted in psychoanalytic theory. They show a great variety but have some basic principles in common. Gabbard (2004) summarizes them as follows: • Much of mental life is unconscious. • childhood experiences in concert with genetic factors shape the adult. • the patient’s transference to the therapist is a primary source of understanding. • the therapist’s countertransference provides valuable understanding about what the patient induces in others. • the patient’s resistance to the therapy process is a major focus of the therapy. • symptoms and behaviours serve multiple functions and are determined by complex and often unconscious forces. • a psychodynamic therapist assists the patient in achieving a sense of authen- ticity and uniqueness.

Psychoanalytic therapies are differentiated according to setting (vis-à-vis versus lying on the couch), frequency (twice a week maximum versus thrice a week minimum) and duration (short-term versus long-term (50 sessions maximum versus 51 minimum). These criteria are obviously arbitrary and superficial, but they are widely accepted among psychoanalysts. They are used in discriminat- ing between ‘psychoanalysis’ and ‘psychoanalytic psychotherapy’ (hereafter: ‘psychotherapy’), as figure 1.1 outlines. It should be clear from figure 1.1 that Long-term PsychoAnalytic Therapy (LPaT) consists of psychoanalysis and long-term psychotherapy while Short- term Psychoanalytic Therapy (SPT) consists of psychotherapy only. a more sophisticated and probably more valid criterion for differentiating between psychoanalytic therapies is the technique the therapist applies. Some psychoanalysts consider the technique applied in psychoanalysis primarily

11 Psychoanalysis Psychotherapy Setting lying on the couch vis-à-vis Frequency minimum thrice a week maximum twice a week Duration always long-term short- or long-term

Figure 1.1: Characteristics of psychoanalysis and psychotherapy.

Technique Psychoanalysis Psychotherapy Interpretive pole ...... overlapping area ...... Supportive pole

Figure 1.2: Interpretation and support in psychoanalysis and psychotherapy.

interpretive and in psychotherapy primarily supportive. The difference is qualitative, the classification is dichotomous. Others locate the various techniques, hence the different therapies they characterize, on a continuum extending from a “purely expressive” (which in this context means ‘purely interpretive’) to a “purely supportive” pole (Luborsky, 1984). The difference is quantitative, the classification is dimensional. Figure 1.2 tries to elucidate the issue. some distinguish between ‘psychoanalytic’ and ‘psychodynamic’ treatments, implying that the term psychoanalytic should be reserved for psychoanalysis and its interpretive technique. The team I work with and I favour the continuum view and consider the terms ‘psychoanalytic’ and ‘psychodynamic’ synonyms. In this thesis these terms will be used accordingly. the ultimate goals of psychoanalytic therapies are twofold. First, they aim at reducing symptoms and complaints, if possible down to their total and persistent disappearance. Second, they aim at preventing recurrence and at improving social functioning, life satisfaction and quality of life, preferably for a long time after treatment termination. These goals are by no means specific to psychoanalytic therapy. The specificity lies in the intermediate goal and in the means. the intermediate goal is more explicit in Long-term PsychoAnalytic Therapy (LPaT) than in Short-term Psychoanalytic Therapy (SPT). It aims at personality change, i.e., at achieving some change in some aspects of the patient’s personality. Improving the capacity of self-reflection is an example. These changes are meant to be lasting. They should enable the patient to make better use of his personal potentials and to meet the challenges and opportunities of life more successfully. In short, the intermediate goal is to reduce the patient’s vulnerability and reinforce his resilience.

12 The means consist of providing a blend of adequate, psychoanalytically understood interpretations and support, varying according to the specific type of therapy. In the classic views, personality change (or ‘growth’) is the result of self-insight, which in its turn results from interpretation, the hallmark of psychoanalytic technique (Bibring, 1954, Cooper 1987, Paniagua, 2004). Over the last decades, the classic views have increasingly been debated. Growth fostering mechanisms other than interpretation, mostly called “relational”, “interactional”, “intersubjective”, or “support” have increasingly become recognised as important (Stern, 1988, Andrade, 2005, Renik, 2003). However, the therapeutic action of support has not been clearly described yet.

1.3 JellinekMentrum Depression Research Project

The Depression Research Project of JellinekMentrum is long-standing. It started in 1993 and is still running. In concordance with the development towards evidence based treatments in mental health care (Dekker, 2003), it aims at comparing the effects of Short-term Psychodynamic Supportive Psychotherapy (SPSP) with other treatment modalities in the treatment of depression. So far, the project team is conducting its fifth randomized clinical trial (RCT). The first section of this thesis is in part based on the raw data of three of these studies. The patients participating in the Depression Research Project are outpatients suffering from depression, receiving ambulatory treatment for 6 months. Hereafter the first three trials are briefly described. the first trial (de Jonghe et al., 2001) compared antidepressants (n=84) with combined therapy (antidepressants plus SPSP) (n=83). It was concluded that patients found combined treatment significantly more acceptable, were significantly less likely to drop-out of combined therapy and significantly more likely to recover. The second trial (Dekker et al., 2005) investigated the comparative efficacy of two modalities of combined therapy (antidepressants and SPSP), one with 8 and the other with 16 sessions of SPSP. The results showed that the two treatment modalities were equally effective in reducing symptoms. The third trial (de Jonghe et al., 2004) assessed the relative efficacy of combined therapy (n=85) and SPSP (n=106). The results showed that neither the therapists nor the independent observers observed any differences between combined therapy and SPSP, but the patients did experience them clearly, in favour of combined therapy. It was concluded that the advantages of combining antidepressants with SPSP were equivocal. parts of the JellinekMentrum Depression Project were directed at the comorbidity of depression and personality disorders (Kool, 2005), at the prediction of therapy success (Van et al., 2007), at the influence of sex on therapy efficacy (Gijsbers-van Wijk et al., 2002) and at depression and social functioning (Molenaar et al., 2006).

13 1.4 Aims, questions and means

1.4.1 Aims This thesis has four main aims. These aims are: 1. to investigate the efficacy of short-term psychotherapy in the treatment of depression, relative to that of pharmacotherapy and combined therapy. 2. To present the characteristics of Short-term Psychoanalytic Supportive Psy- chotherapy (SPSP) and one of its main therapeutic agents: support. 3. To investigate the efficacy of SPSP in the treatment of depression, relative to that of pharmacotherapy and combined therapy. 4. To investigate the effectiveness of Long-term PsychoAnalytic Therapy (LPaT).

1.4.2 Research questions The four main aims of this thesis translate into the following specific research questions.

Aim 1 Q.01. What is the efficacy of short-term psychotherapy compared to that of pharmacotherapy in ambulatory psychiatric outpatients presenting a major depressive disorder? Q.02. What is the efficacy of short-term psychotherapy compared to that of short-term psychotherapy plus pharmacotherapy in ambulatory psychiatric outpatients presenting a major depressive disorder?

Aim 2 Q.03. What is Short-term Psychoanalytic Supportive Psychotherapy (SPSP)? Q.04. how may adequate, psychoanalytically understood support be a therapeutic factor?

Aim 3 In ambulatory psychiatric patients presenting a major depressive disorder, Q.05. What is the efficacy of SPSP compared to that of pharmacotherapy? Q.06. What is the efficacy of SPSP compared to that of SPSP plus pharmacotherapy? Q.07. What is the efficacy of SPSP plus pharmacotherapy compared to that of pharmacotherapy? Q.08. What is the answer to the questions Q.05-07. at the level of two HDRS factors and at the level of three SCL-90 subscales: Depression, Anxiety and Somatic complaints?

Aim 4 Q.09. What study type provides the best available evidence regarding the effectiveness of long-term psychotherapy? Q.10. What is the effectiveness of long-term psychoanalytic therapy in general terms?

14 Q.11. What is the effectiveness of long-term psychoanalytic therapy in costs and benefits terms?

1.4.3 Means Q.01-02 are investigated by performing two meta-analyses based on a literature search (chapters 2 and 3). Q.03-4 are addressed in two theoretical articles explaining the psychoanalytical roots and characteristics of SPSP (chapter 4) and reflecting on the putative mode of action of support (chapter 5). Q.05-08 are investigated by performing two mega-analyses based on the original data of three Mentrum RCT’s: one based on the sum scores of the HDRS (chapter 6) and one based on factor level scores of the HDRS and three subscales of the SCL-90 (chapter 7). Q.09 is addressed in a theoretical article reflecting on the relative values of randomized clinical trials and cohort studies (chapter 8). Q.10-11 are investigated by performing two systematic literature reviews (chapters 9 and 10).

1.5 Measures of effectiveness

Eight of the eleven research questions of this thesis are addressed in effectiveness studies. Treatment effectiveness can be measured with different instruments. Hereafter the effect measures used in this thesis are explained. As chapters 4, 5 and 8 consist of theoretical articles, they are not represented here.

Chapters 2 and 3 In the two meta-analyses based on literature, the efficacy of the investigated treatments is expressed in remission rates, measured by the HDRS (Hamilton, 1967). Remission was defined as a score on the HDRS below a certain cut-off score. There are different cut-off scores being used in literature, varying from 6 to 9 HDRS points. Remission rates were pooled, calculating the Relative Risk (RR) and the Odds Ratios (OR). The RR is the ratio between the risks of an event (e.g. remission) in group A and in group B. The OR is the ratio between two odds; the odds on an event in group A and the odds on the same event in group B. Both an OR and an RR amounting to 1 signify that there is no difference between the two treatments. Numbers Needed to Treat (NNT) were calculated, indicating the number of patients that would need to be treated with treatment A to produce one recovery from depression which would not have occurred had they been given treatment B. The dropout rates and relapse rates were pooled, calculating the Relative Risk. In one meta-analysis (chapter 2) the Relative Risk of relapse at follow-up was calculated. Patients did not relapse if they: a) were remitted after acute treatment and b) did not meet criteria for depression at follow-up.

15 Chapters 6 and 7 The two mega-analyses (comparing SPSP, pharmacotherapy and their combination) are based on three JellinekMentrum RCTs. They derived their efficacy data from three sources: the patients, the therapists and independent observers. the independent observers (main criterion) used the 17-item HDRS (Hamil- ton, 1967). This is a scale of 17 items that assesses the severity of depression. It regards questions about symptoms such as sleep, feelings of anxiety and guilt, eating and weight, hypochondrias and suicidal ideations. The therapists used the Clinical Global Impression of Severity (CGI-S) and Improvement (CGI-I) (Guy, 1976). This assessment instrument uses a 7-step Likert Scale to assess how depressed patients are (from normal (1) to extremely ill (7) on the CGI-S) and how much they have improved (from very much improved (1) to far worse on the CGI-I (7)). The CGI is widely used in clinical research (i.e. Zaider et al., 2003). In the combined therapy condition, the CGI was provided by the treating psychiatrist, not by the psychotherapist. the patients used the three subscales of the Ninety Symptom Checklist (SCL- 90, Arrindell & Ettema, 1986.): Depression, Anxiety and Somatic Complaints. The SCL-90 is a self-report questionnaire that consists of 90 items, clustered in 8 sub-scales, measuring a broad range of psychopathological symptoms. Illus- trative SCL-90 items are ‘feeling anxious’ and ‘having a feeling of emptiness’. Patients also used the Quality of Life Depression Scale (QLDS, Tuynman-Qua and de Jonghe, 1992). This questionnaire contains 34 items that patients have to answer with ‘true’ or ‘not true’. Examples of items are: ‘I just want time to pass’, ‘I feel hopeful about the future’, ‘I take good care of myself’, ‘I can’t be bothered about my friends’, ‘I enjoy my food’ and ‘My life has no meaning’. remission was defined as a final HDRS score of 7 points or less, a CGI-S or CGI-I score of 2 points or less and as an improvement of at least 1 standard deviation from base rate on the SCL subscales and the QLDS. Response was defined as a 50% reduction in the HDRS baseline score. As evidence-based medicine highly values the opinions of patients and clinicians, alongside those of independent observers, we consider the three sources of efficacy measurement to be an advantage. However, as the independent observer is considered the golden standard, this is the primary outcome, also used in the meta-analyses presented in chapters 2 and 3.

Chapter 9 The literature review regarding long-term psychoanalytic therapies inevitably used the various outcome measures found. Regarding symptom reduction and personality change, the primary outcome data were pre-post and pre-follow- up effect sizes (ES, Cohen, 1988). We considered these to be the most ‘robust’ outcome measures, firstly because they require pre- and post-treatment assessments using specific questionnaires and secondly because their standardization allows comparison. Effect sizes of less than 0.5 are considered to be small, from 0.5 to 0.8 to be medium and effect sizes exceeding 0.8 to be large (Cohen, 1988). Success rates were adopted as the secondary outcome measure. They consist of appraisals from patients and therapists of therapy success. We con-

16 sider these measures to be less ‘robust’, since they consist of more general subjective assessments (often based on five- or seven-step Likert scales like a Clinical Global Assessment (CGI, Guy, 1976)). We defined ‘success’ as at least moderate improvement. This means that ‘success’ varied from moderate to excellent improvement on a certain scale. Where possible, the outcomes were differentiated according to symptom reduction (such as SCL-90) and personality changes (such as capacity to enjoy, interpersonal capacities, capacity to deal with conflicts and interpersonal relationships).

Chapter 10 The literature review regarding financial costs and benefits of long-term psychoanalytic therapies used data regarding work impairment (assessed in days of sick leave) and health care use (assessing use of medication, medical consultations and hospital days). Costs reductions in these areas reflect pre-post treatment and pre-follow-up changes. Additionally, the costs of treatment were calculated and compared with the costs reductions.

1.6 Contents of the thesis

Chapters 2 and 3 consist of the articles ‘Relative efficacy of psychotherapy and pharmacotherapy in the treatment of depression’ (de Maat et al., 2006) and ‘Relative efficacy of psychotherapy and combined therapy in the treatment of depression’ (de Maat et al., 2007). In these chapters the results of two meta- analyses based on RCTs published between 1980 and 2005 are presented, comparing the efficacy of psychotherapy to that of pharmacotherapy and combined therapy (psychotherapy plus pharmacotherapy) respectively, in adult psychiatric outpatients with nonpsychotic, unipolar depression. An important limitation to most existing reviews and meta-analyses is their clinical heterogeneity. Clinical heterogeneity refers to differences between studies in patient samples, treatment protocols and clinical settings. It makes the pooling of data by means of meta-analysis hazardous. It may provide a partial explanation for the rather inconsistent conclusions reached by different reviews. In the present meta-analyses, clinical homogeneity is furthered by applying strict inclusion criteria for patient samples, diagnosis and treatment settings. chapter 4 consists of the article ‘Short-term Psychodynamic Supportive Psy- chotherapy for depressed patients’ (De Jonghe et al., submitted). It explains the theoretical roots and characteristics of SPSP, a treatment modality developed from 1992 on in the context of the Depression Research Project of JellinekMen- trum in Amsterdam. chapter 5 consists of the article ‘Support and personality change: A psychoanalytic view’ (De Jonghe et al., submitted). It proposes a psychoanalytic definition of support and differentiates between adequate and inadequate support. It presents some ideas about the putative mode of action of adequate support. It discusses a largely neglected process in psychoanalytic therapies: the relationship between support and personality change.

17 chapter 6 consists of the article ‘Short Psychodynamic Supportive Psycho- therapy, Antidepressants and their Combination in the Treatment of Major Depression: A mega-analysis based on three randomized clinical trials’ (de Maat et al., 2007). It presents the results of a mega-analysis based on the original data of three randomized clinical trials (de Jonghe et al., 2001; de Jonghe et al., 2004; Dekker et al., 2005). Mega-analysis is distinct from meta-analysis, which is performed on the outcome of trials (N= the numbers of studies reviewed), not on their original data. This mega-analysis compares the efficacy of SPSP, pharmacotherapy and combined therapy, taking into account the sum scores of the assessment instruments. Three hypotheses are tested: Combined therapy is more efficacious than pharmacotherapy; Combined therapy is more efficacious than SPSP; SPSP is more efficacious than pharmacotherapy. Symp- tom reduction is evaluated by patients, independent observers, and therapists. In addition, the effects of the three treatment modalities on quality of life, an aspect investigated but not reported on in the three original RCTs, is presented. It is evaluated by the patients. chapter 7 consists of the article ‘Comparison of Short Psychodynamic Supportive Psychotherapy, Pharmacotherapy and their Combination on sub- dimensions of the HDRS and subscales of the SCL-90: a mega-analysis of three randomized controlled trials regarding depressed patients’ (de Maat et al., submitted). It presents the results of a second mega-analysis, similar to the one presented in chapter 6, but using scores at HDRS factor level and three subscales of the SCL-90. Most depression scales present different symptom clusters discovered by factor analyses (Bagby, 2004; Arrindell & Ettema, 1986; Shafer, 2006). However, depression studies use the overall scores of these scales. With a multidimensional scale this can present problems for both assessment and treatment. The present mega-analysis tests three hypotheses: SPSP outperforms pharmacotherapy on the ‘mental’ factor of the HDRS, SCL-Depression and SCL-Anxiety; pharmacotherapy yields better results than SPSP on the ‘somatic’ factor of the HDRS and the SCL- Somatic complaints; combined therapy performs better than both monotherapies. chapter 8 consists of the article ‘The effectiveness of long-term psychotherapy: methodological research issues’ (de Maat et al., 2007). It compares Randomized Controlled Trails (RCTs), a method for investigating efficacy, with cohort studies, a method for investigating effectiveness. The article discusses the pros and cons of both methods. In evidence-based medicine hierarchy, randomized controlled trials (RCTs) rank higher than cohort studies. However, where cohort intervention studies are frequently used to investigate long-term psychotherapy, RCTs are used hardly ever. The article critically reviews the acceptability, feasibility and decisive power of RCTs in long-term psychotherapy. In addition, it presents the strengths and limitations of cohort studies. chapter 9 consists of the article ‘The effectiveness of long-term psychoanalytic therapy: a systematic review of empirical studies’ (de Maat et al., submitted). It presents a systematic review of studies investigating long-

18 term psychoanalytic therapies, published from 1970 onwards. In order to enhance clinical homogeneity, it focuses on individual, ambulatory LPT with adult patients with mostly ‘regular’ indications for psychoanalytic therapy. In addition, it distinguishes between psychotherapy and psychoanalysis. It performs a systematic assessment of study quality using an explicit quality criterion. Furthermore, it attempts to pool the data about the effectiveness of LPT, both for symptom reduction and personality change. chapter 10 consists of the article ‘Costs and benefits of long-term psychoanalytic therapy: changes in health care use and work impairment’ (de Maat et al., 2007). It is a systematic review comparable to the previous one, except for the fact that it focuses on the effects of LPT on health care use and work impairment. It presents a financial evaluation of the costs of LPT compared to its benefits in these two areas. in chapter 11 a general discussion, conclusions and implications of the research findings are presented.

References

Anderson, E.M., Lambert, M.J. (1995). Short-term dynamically orientated psychotherapy: A review and meta-analysis. Clinical Psychology Review, 15, 6: 503-514. Andrade VM (2005). Affect and the therapeutic action of psychoanalysis. International Journal of Psychoanalysis, 86: 677-97. Arrindell WA, Ettema JMM. SCL-D-90 (1986). Handleiding bij een multidimensionele psychopathologie-indicator. Lisse: Swets & Zeitlinger. Bachrach, H., Galatzer-Levy, R., Skolnikoff, A., Waldron, S. (1991). On the efficacy of psychoanalysis. Journal of the American Psychoanalytic Association, 39: 871- 916. Bagby RM, Ryder AG, Schuller DR, Marshall MB. (2004) The Hamilton Depression Rating Scale: Has the Gold Standard Become a Lead Weight? American Journal of Psychiatry; 161,2163-2177. Beutler, L.E.& Crago, M.,eds. (1991). Psychotherapy research. Washington, DC: Ameri- can Psychological Association. Beutler, L.E. (1998). Identifying empirically supported treatments: What if we didn’t? Journal of Consulting and Clinical Psychology, 66, 113-20. Bibring E (1954). Psychoanalysis and the Dynamic Psychotherapies. Journal of the American Psychoanalytic Association, 2: 745-70. Center of Evidence Based Medicine Oxford, Levels of evidence (2001). Available at: http://www.cebm.net. Accessed March, 2005. Cohen, J. (1988). Satistical power analysis for behavioural sciences. Hillsdale, NJ: Lawrence Erlbaum. Cooper AM (1987). Changes in psychoanalytic ideas: transference interpretation. Jour- nal of the American Psychoanalytic Association, 35: 77-98. Crits-Chritoph, P. (1992) The efficacy of brief dynamic psychotherapy: a meta-analysis. American Journal of Psychiatry, 149:2: 151-158. Dekker J, Molenaar PJ, Kool MA, van Aalst G, Peen J, de Jonghe F. (2005). Dose-effect relations in time-limited combined psycho-pharmacological treatment for depression. Psychological Medicine, 35(1): 47-58. Dekker, J. (2003). Evidence based psychological interventions in mental health care. VU, Amsterdam.

19 Doidge, N. (1997). Empirical evidence for the efficacy of psychoanalytic psychotherapies and psychoanalysis: An overview. Psychoanalytic Inquiry Supplemental, 102-150. Fonagy, P. (1999). Process and outcome in mental health care delivery: a model approach to treatment evaluation. Bulletin of the Menninger Clinic, 63: 288-304. Fonagy, P. (1999). An open door review of outcome studies in psychoanalysis. http:// www.ipa.org.uk/R-outcome.htm. Gabbard, G.O. (2004). Long-term Psychodynamic Psychotherapy, a basic text. Wash- ington, DC/London, England: American Psychiatric Publishing, inc. Guy W. (1976). ECDEU Assessment Manual for Psychopharmacology. Revised DHEW Pub. (ADM). Rockville, MD: National Institute for Mental Health. Gijsbers van Wijk, C.M.T., J. Dekker, J. Peen en F. de Jonghe (2002). Depressie bij mannen en vrouwen: sekseverschillen in behandeluitkomsten van farmacotherapie en gecombineerde therapie. Tijdschrift voor Psychiatrie,4, 301-312. Hamilton, M. (1967). Development of a rating scale for primary depressive illness. Brit- ish Journal of Social and Clinical Psychology, 6, 278-296. Henry, W.P. (1998). Science, politics, and the politics of science: The use and misuse of empirically validated treatment research. Psychotherapy Research, 8: 126-40. Jonghe, F., de, Kool, S., Aalst, G., Dekker, J., Peen, J. (2001). Combining psychotherapy and antidepressants in the treatment of depression, Journal of Affective Disorders, 64, 217-229. Jonghe, F., de, Hendriksen, M., Aalst, G., et al. (2004). Psychotherapy and Combined Therapy (pharmacotherapy plus psychotherapy) in the treatment of depression. Brit- ish Journal of Psychiatry, 185 (1), 37-45. Jonghe, F., de, Maat, S., de, Van, H., Hendriksen, M., Kool, S., Aalst, G., van, Schoevers, R., Dekker, J. Short-term Psychoanalytic Supportive Psychotherapy for depressed patients. Submitted. Jonghe, F., de, Maat, S., de, Van, H., Hendriksen, M., Kool, S., Aalst, G., van, Schoevers, R., Dekker, J. Support and personality change: A psychoanalytic view. Submitted. Kool, S. (2005). Comorbidity of Major Depression and Personality Disorders. Ridder- print Offsetdrukkerij BV, Amsterdam. Leichsenring, F. (2001). Comparative effects of short-term psychodynamic psychotherapy and cognitive-behavioral therapy in depression: a meta-analytic approach. Clinical Psychology Review, 21: 401-19. Leichsenring, F., Leibing, E., (2003). The effectiveness of psychodynamic therapy and cognitive behavior therapy in the treatment of personality disorders: A meta-analysis. American Journal of Psychiatry, 160: 1223-1232. Leichsenring, F. (2004). Randomized controlled versus naturalistic studies: a new research agenda. Bulletin of the Menninger Clinic, 68,2: 137-151. Leichsenring, F. (2005). Are psychodynamic and psychoanalytic therapies effective? A review of empirical data. International Journal of Psychoanalysis, 86: 841-68. Luborsky, L., Docherty, J.P., Berber, J.P., Miller, N.E. (1993). How this basic handbook helps the partnership of clinicians and clinical researchers: Preface. In: Psychody- namic treatment research: A handbook for clinical practice, ed. N.E. Miller, L. Luborsky, J.P. Barber, & J.P. Docherty. New York: Basic Books. Luborsky, L.(1984). Principles of Psychoanalytic Psychotherapy: A manual for supportive-expressive treatment. New York: Basic Books. Maat, S., de, Dekker, J., Schoevers, R., Jonghe, F., de. (2006). Relative efficacy of psychotherapy and pharmacotherapy in the treatment of depression: a meta-analysis. Psychotherapy Research, 16(5): 562-572. Maat, S., de, Dekker, J., Schoevers, R., Jonghe, F., de. (2007). Relative efficacy of psychotherapy and combined therapy in the treatment of depression: A meta-analysis. European Psychiatry, 22, 108. Maat, S., de, Dekker, J., Schoevers, R., Aalst, A., van, Gijsbers-van Wijk, C., Hendriksen, M., Kool, S., Peen, J., Van, H., Jonghe, F., de. (2007). Short Psychodynamic Sup- portive Psychotherapy, Antidepressants and their Combination in the Treatment

20 of Major Depression: a mega-analysis based on three randomized clinical trials. Depression and Anxiety (in press). Maat, S., de, Schoevers, R., Jonghe, F., de, Aalst, A., van, Gijsbers-van Wijk, C., Hen- driksen, M., Kool, S., Peen, J., Van, H., R., Dekker, J. Comparison of Short Psychody- namic Supportive Psychotherapy, Pharmacotherapy and their Combination on sub dimensions of the HDRS and sub scales of the SCL-90: a mega-analysis of three randomized controlled trials regarding depressed patients. Submitted. Maat, S., de, Dekker, J., Schoevers, R., Jonghe, F., de. (2007). The effectiveness of long- term psychotherapy: methodological issues. Psychotherapy Research, 17(1): 59-65. Maat, S., de, Jonghe, F., de, Schoevers, R., Dekker, J. The effectiveness of long-term psychoanalytic therapy: a systematic review of empirical studies. Submitted. Maat, S., de, Philipszoon, F., Dekker, J., Jonghe, F., de. (2007). Costs and benefits of long-term psychoanalytic therapy: changes in health care use and work impairment. Harvard Review of Psychiatry (in press).. Molenaar P, Dekker J, Van HL, Hendriksen M, Schoevers R. (2006). Does adding psychotherapy to pharmacotherapy improve social functioning in the treatment of depression? Depression and Anxiety; Nov 10;[Epub ahead of print] Offringa, M., Assendelft, W.J.J., Scholten, P.J.P.M. eds. (2000). Inleiding in the evidence based medicine. Klinisch handelen gebaseerd op bewijsmateriaal. Houten/ Diegem: Bohn Stafleu Van Loghum. Paniagua C (2003). Problems with the concept ‘interpretation’. International Journal of Psychoanalysis, 84: 1105-23. Persons, J.B. & Silberschatz, G. (1998). Are results of randomized trials useful to psychotherapists? Journal of Consulting and Clinical Psychology, 66: 126-35. Renik O (2004). Intersubjectivity in psychoanalysis. International Journal of Psychoa- nalysis, 85: 1053-64. Roth, A.D. & Parry, G. (1997). The implications of psychotherapy research for clinical practice and service development: lessons and limitations. Journal of Mental Health, 6: 367-80. Seligman, M.E.P. (1995). The effectiveness of psychotherapy. The Consumer Reports Study. American Psychologist, 50(12): 965-974. Shadish, W.R., Cook, T.D., Campbell, D.T. (2002) Experimental and quasi-experimental designs for generalized causal inference. Boston: Houghton Mifflin Company. Shafer, A. (2006). Meta-analysis of the factor structures of four depression questionnaires: Beck, CES-D, Hamilton, and Zung. Journal of Clinical Psychology, 62(1): 123-146. Stern DN, Sander LW, Nahum JP, Harrison AM, et al. (1998). Non-interpretive mechanisms in psychoanalytic therapy. The “something more” than interpretation (The Boston Change Process Study Group, Report No. 1) International Journal of Psy- choanalysis, 79: 903- 21. Tuynman-Qua, H., de Jonghe, F. (1992) Quality of Life Depressie Schaal. Ibero, Houten. Van HL, Schoevers RA, Kool S, Hendriksen M, Peen J, Dekker J. (2007). Does early response predict outcome in psychotherapy and combined therapy? Journal of Affec- tive Disorders, May 21; [Epub ahead of print]. Westen, D. Novotny, C.M., Thomson-Brenner, H. (2004). The empirical status of empirically supported psychotherapies: Assumptions, findings, and reporting in controlled clinical trials. Psychological Bulletin, 130: 631-63. Wiener, J. (1994). From the president: On psychotherapy. Psychiatry News, October 21, pp. 3-16. Zaider, T.I., Heimberg, R.G., Fresco, D.H., Scheier, F.R., Liebowitz, M.R. (2003). Evalu- ation of the Clinical Global Impression Scale among individuals with social anxiety disorder. Psychological Medicine, 33: 611-622.

Chapter 2 Relative efficacy of psychotherapy and pharmacotherapy in the treatment of depression: a meta-analysis

Saskia de Maat, Jack Dekker, Robert Schoevers, Frans de Jonghe Psychotherapy Research (2006) 16(5): 562-572

Abstract

We investigated the efficacy of pharmacotherapy and psychotherapy for depression by searching for RCT’s. Studies were classified according to chronicity and severity and a meta-analysis was applied. Ten studies were included. Remission did not differ between psychotherapy (38%) and pharmacotherapy (35%). No differences were found in chronic, or in non-chronic depression, and in mild or in moderate depression. Both treatments performed better in mild than in moderate depression. Drop-out was larger in pharmacotherapy (28%) than in psychotherapy (24%). At follow-up relapse in pharmacotherapy (57%) was higher than in psychotherapy (27%). Psychotherapy and pharmacotherapy appear equally efficacious in depression. Both treatments have larger effects in mild than in moderate depression, but similar effects in chronic and non-chronic depression and at follow-up psychotherapy outperforms pharmacotherapy.

2.1 Introduction

In the past twenty-five years a number of reviews and meta-analyses comparing the efficacy of psychotherapy and pharmacotherapy in depression have been conducted (e.g. Hollon et al., 2005a; Casacalenda et al., 2002; Hol- lon et al., 2002; DeRubeis et al., 1999; Gloaguen et al., 1998; Jarrett, 1995; Wexler & Cicchetti, 1992; Hollon et al., 1991; Robinson et al., 1990; Dobson, 1989; Weissman et al., 1987; Steinbrueck et al., 1983; Royal Australian and New Zealand College of Psychiatrists, 1983). It has been argued that many of these reviews and meta-analyses present methodological limitations. They often do not provide Intention-To-Treat (ITT) analyses, they present effect sizes from which obviously no remission rates can be deduced, they include flawed studies (e.g. studies that did not use standardized diagnostic criteria) or present response rates instead of remission rates (Casacalenda et al., 2002). an even more important limitation may be the striking methodological and clinical heterogeneity of the studies included in most reviews and

23 meta-analyses. Clinical heterogeneity refers to differences in patient samples, treatment protocols and treatment settings across studies. We mention three examples. In Casacalenda’s et al. meta-analysis (2002), three trials regard primary care patients, whereas the other three trials consider psychiatric outpatients. Treatment duration varies from 10 to 34 weeks. Psychotherapy conditions include cognitive therapy and interpersonal psychotherapy, as well as problem solving therapy and social work counselling. In the meta- analysis of Gloaguen et al. (1998) settings vary even more, including hospital patients, outpatients, volunteers, students, adolescents and geriatric patients. Treatment duration varies from 4 to 79 weeks. Not surprisingly, the authors frequently report that the hypothesis of inter trial homogeneity was rejected. The review of Hollon et al. (2005) considers primary care, geriatric and adult in and outpatients, suffering from dysthymia or major depressive disorder (MDD). Although some of the reviewers (e.g. Gloaguen et al., 1998) do address the issue of heterogeneity, most of the reviews and meta-analyses mentioned above do not include statistical analyses assessing the influence of the clinical heterogeneity on the review outcome. Clinical heterogeneity among studies included in reviews or meta-analyses makes data pooling hazardous (see Cochrane Reviewers Handbook 4.2.2, Cochrane Collaboration, 2004). It certainly does not allow specific conclusions regarding particular patients groups or settings. Heterogeneity may provide a partial explanation for the rather inconsistent conclusions reached by different reviews. Many of them conclude that psychotherapy and pharmacotherapy are equally efficacious, but some deduce that psychotherapy outperforms pharmacotherapy (Gloaguen et al., 1998; Dobson, 1989; Weissman et al., 1987; Steinbrueck et al., 1983; The Quality Assurance Project, for neurotic depression, 1983). in this article we present the results of a meta-analysis based on RCT’s published between 1980 and 2005, comparing psychotherapy to pharmacotherapy in adult psychiatric outpatients with nonpsychotic unipolar depression. We increased clinical homogeneity among studies by applying rather strict inclusion criteria regarding patient samples, diagnoses and treatment settings (see Search strategy). Subsequently, we statistically tested the heterogeneity among the included studies in order to assess to what extent we had achieved clinical homogeneity. Thus, studies were selected on the basis of clinical criteria only. Statistical heterogeneity analysis was not used as a selection criterion but as a test run afterwards. The primary research question regards the relative efficacy of pharmacotherapy and psychotherapy in the acute treatment of depression, assessed at treatment termination and at follow-up. The secondary question regards possible differences in dropout rates during treatment. We took into consideration two variables known to influence treatment prognosis: chronicity and severity. To that end we differentiated between mild, moderate and severe depression, and between chronic and non-chronic depression.

24 2.2 Method

Search strategy A systematic search for Randomized Controlled Trials (RCT’s) was performed in MEDLINE, EMBASE, the Cochrane Controlled Trials Register, the Cochrane Database of Reviews and Protocols, and PsychInfo. Search-headings were: DEPRESSION, MAJOR DEPRESSIVE DISORDER, PSYCHOTHERAPY, PHAR- MACOTHERAPY, ANTIDEPRESSANTS. Limits were: (randomized controlled trial[Publication Type] OR controlled clinical trial[Publication Type] OR randomized controlled trials OR random allocation OR double-blind method OR single-blind method OR clinical trial[Publication Type] OR clinical trials OR (clinical AND trial*) OR ((singl* OR doubl* OR trebl* OR tripl*) AND (blind* OR mask*)) OR placebos OR placebo* OR random* OR research design OR comparative study OR evaluation studies OR follow up studies OR prospective studies OR control OR controlled OR prospective* OR volunteer*) NOT (Animal[MESH] NOT (Human[MESH] AND Animal[MESH])) and a time-limit of 1980 (the year DSM-III was published) until 2005. Titles and abstracts were screened. References of the retrieved articles were searched. Book chapters on treatment of depression were retrieved. No special efforts were made to discover unpublished data. Figure A.1 (Appendix) is a quorum flow diagram of the process and results of literature search. to obtain a clinically rather homogeneous sample, several selection criteria were applied. To be included, the study should compare psychotherapy with pharmacotherapy and focus on efficacy of acute treatment (no maintenance treatment, nor sequential treatment). The study sample should consist of psychiatric outpatients (no primary care patients or inpatients), aged between 19 and 65 years (no geriatric patients, no children), diagnosed with unipolar major depression according to DSM-III-(R) (APA, 1980), DSM-IV-(R) (APA, 1994) or Research Diagnostic Criteria (Spitzer et al., 1978). Treatment protocols in the studies should apply a formal (according to behavioural, cognitive, psychodynamic or client centred theories and techniques) time-limited (maximum 6 months) individual psychotherapy and an adequate treatment with regular antidepressants. (i.e., an adequate dose (different per antidepressant) administered during an adequate time period (at least 4 weeks) by a registered clinician). A regular antidepressant is approved as such by national authorities. Method sections were checked for the specifics of the treatment regime, but no efforts were made to obtain additional information. The study should report remission rates and drop out rates. Methodological quality was judged according to four criteria of Cochrane Collaboration.

1. the study should have a randomized design to minimize selection bias. 2. apart from the treatments, the two study groups must have been treated equally to minimize performance bias. 3. the study should report on selective drop out in the treatment conditions (e.g. have ITT analyses or specify differences in drop out). 4. Detection bias should be minimised by blind assessment of outcomes.

25 Two reviewers, who needed to agree on all criteria in order to include a study, judged all selection criteria independently. No studies were excluded due to disagreement between the reviewers. the main outcome of the meta-analysis was efficacy at treatment termination, expressed in remission rates, and at follow-up, expressed in relapse rates. Remission rates were pooled calculating the Relative Risk (RR) and the Odds Ratios (OR). The RR is the ratio between the risks of an event (e.g. remission) in group A and in group B. The OR is the ratio between two odds; the odds on an event in group A and the odds on the same event in group B. Both an OR and a RR amounting to 1 signify that there is no difference between the two treatments. Numbers Needed to Treat (NNT) were calculated, indicating the number of patients that would need to be treated with treatment A to produce one recovery from depression which would not have occurred had they been given treatment B. The dropout rates and relapse rates were pooled calculating the Relative Risk. Remission rates at treatment termination were pooled in an Intention To Treat sample (ITT, i.e., a sample consisting of all randomized patients). Relapse rates at follow-up were pooled against all patients remitted at treatment termination. all data were analysed using the Review Manager 4.2 software of the Cochrane Collaboration. Dichotomous data (RR and OR) were analysed using the Mantel-Haenszel fixed effects model with 95% confidence intervals. Our analyses included a formal test of statistical heterogeneity. Statistical heterogeneity is the variability in the treatment effects in the different studies. It is a consequence of clinical and/or methodological diversity among these studies. Statistical heterogeneity occurs when the observed treatment effects are more different from each other than one would expect based on random chance alone. Significant heterogeneity suggests that the studies are not estimating the same quantity. The heterogeneity test we used was the natural approximate chi-square test; non-significant results (using p=0.10 as a limit) indicate a lack of evidence for heterogeneity in the results. The test also describes the percentage of the variability in effect estimates (I²) due to heterogeneity rather than to sampling error. An I² of more than 50% indicates notable heterogeneity (Cochrane Reviewers Handbook 8.7.2, Cochrane Collaboration, 2004). all analyses were also performed in sub-samples regarding chronicity and severity of the depression. First, studies regarding chronic depression (the majority of patients were diagnosed as presenting a depression lasting at least 2 years) were differentiated from studies regarding non-chronic depression. Second, using the mean baseline scores on the 17-item Hamilton Depression Rating scale (HDRS, Hamilton, 1967), studies regarding mild (12-19 points) moderate (20-24 points) and severe (25 points or more) depression were differentiated.

Outcome of the literature search The Quorum flow diagram in the Appendix shows the process and results of the literature search. Table 2.1 lists the studies considered suitable for our review.

26 As can be seen in table 1 our meta-analysis is based on 10 studies that, taken together, include 1,233 patients, 640 treated with pharmacotherapy and 593 with psychotherapy. In the study of Elkin et al. (1989) there were two psychotherapy conditions (CBT and IPT). In the study of Blackburn & Moore (1997) there were two antidepressants groups. We decided to combine the similar treatment groups in these two studies. To enter two comparisons for each study in the meta-analysis would violate the assumption that all comparisons in a meta-analysis should be independent (Cooper & Hegdes, 1994).

Chronic versus non-chronic depression We found seven studies of non-chronic depression (Blackburn et al., 1981a and b; Murphy et al, 1984; Elkin et al., 1989; Hollon et al., 1992; Murphy et al., 1995; Hautzinger et al., 1996; Blackburn & Moore, 1997) and three studies of chronic depression (Jarrett et al., 1999; Keller et al., 2000 and DeRubeis et al., 2005). In the Keller study 35% of the patients suffered from chronic major depression, 42% from MDD plus dysthymia, and 23% from recurrent depression without complete remission between episodes, which in our opinion signifies that all patients suffered from chronic depression. In the DeRubeis study 90% of the patients had chronic or recurrent depression. Mean duration of the last episode was 7,5 years in the Keller study, 46 months in the DeRubeis study and respectively 73 months (in the psychotherapy group) and 50 months (in the pharmacotherapy group) in the Jarrett study.

Mild versus moderate depression Eight studies provided 17-item HDRS mean baseline scores. The Jarret study, however, used the 21-item and the Keller study the 24-item HDRS version. We used the O’Sullivan et al. (1997) report to translate the 21- and 24-item scores into 17-item scores. The authors found a ratio of 1.098 between the 21-item and 17 item HDRS and a ratio of 1.25 between the 24-item and 17 item HDRS. We calculated that the mean baseline scores in the Jarret study, 21.1 points (psychotherapy) and 20.3 points (pharmacotherapy), correspond to respectively 19 (21.1/1.098) and 18 (20.3/1.098) 17-item HDRS points. The mean baseline score of 27 points in the Keller study corresponds to 22 (27/1.25) 17-item HDRS points. in all, we found five studies of (on average) mild depression (Blackburn et al., 1981, Murphy et al., 1984, Elkin et al., 1989; Murphy et al., 1995 and Jarrett et al., 1999) and five studies of (on average) moderate depression (Hollon et al., 1992; Hautzinger et al., 1996; Blackburn & Moore, 1997; Keller et al., 2000 and DeRubeis et al., 2005). We did not find suitable studies regarding severe depression.

Follow-up studies Six publications reported follow-up data. Hollon et al. (2005b) add follow- up data to the publication of DeRubeis et al. (2005), Evans et al. (1992) to Hollon et al. (1992), Shea et al. (1992) to Elkin et al. (1989), Simons et al. (1986) to Murphy et al. (1984) and Blackburn et al. (1986) to Blackburn et al.

in in

6. 9. 8. 7.

not

has not not not not ≤ ≤ ≤ ≤

scores need are

definition: score

months MDD. Study

Two

1986)

phase

Cognitive

depression. HDRS HDRS HDRS HDRS

weeks DBI that physicians included included

definition: definition: definition: definition: definition:

meeting condition week. conditions depression. for

Research

Relapse: patients

BDI 2 2

outpatients

groups condition 1 considered by condition condition condition and condition HDRS>13.

At 9.

not not of

review. review.

≤ by outpatients 1986) or

Only Remission weeks

29.

Study 90% chronic treatment this this Criteria recurrent

(Blackburn,

meeting definition: definition: definition:

definition:

Relapse: Relapse: Relapse: >

and

symptoms defined

Remission Remission Remission Remission Only Remission

in in 2005b) Relapse: HDRS

two

9 7. 6. 6. 7. or

depression.

MDD

1992)

8. 1992) ≤ ≤ ≤ ≤ ≤

(Simons, acute

separated ≤ and score review. review. included. included.

15.

therapy plus placebo Placebo treatment Combined therapy Relaxation therapy Combination therapy Placebo condition Combined therapy Placebo condition Combined therapy Combined therapy 0%)

(24%) (16%) Remission (26%) Remission (%)

(47%) BDI (42%) (18%) (25%) Remission (14%) (21%) (25%) (25%) (44%) >15 (40%) symptoms (33%) HDRS (19%) HDRS (11%) Only ( (41%) (21%) pooled. (36%)

9/60 (15%) 5/12 3/17 9/36 5/36 5/24 8/24 3/16 3/27 0/11 9/25

(%) (n/N)

(32%) 55/228 (46%) 19/120 (29%) 59/226

(40%) (33%) (47%) (24%) 18/38 (58%) (58%) (50%) (33%) (63%) (35%) 10/40 (41%) 16/63 (33%) 25/57 (34%) 25/62 (30%) (100%) (38%) 26/63 (19%) 10/48

(32%)

4/12 8/17 9/38 8/24 8/27 9/48 (n/N) HDRS not included. (Hollon,

relapse indicating > included. included. included. this included. Relapse: criteria

treatment.

BDI

HDRS atypical included. considered HDRS

55/120 (Evans, Diagnostic (Shea, included pharmacotherapy

ITT) ITT) included

post Remission Drop-out Comment

4.79 6.8 8.8 8.6* 21/36 7.7 12/24 8.3 10/16 8.5 14/40 9.8 26/63 9.8 24/63

10.9 15.1** 72/228 10.2* 21/36 10.7 14.2 19/57 10.7 21/62 14.7** 64/226 2.27 11/11 11.4 13.3 8/25 13.3 Remission sample: this

baseline 24/60 17-item

– – – – – – – – – – – – – – – – – – – – and

(completers) IDS (completers) not

(modified

mean HDRS

23.4 chronic

20.8 19.6

weeks 16.4 weeks/ 18.9 weeks 20.9 weeks/ 26.4 weeks/ 21.1 weeks/ Mean weeks/ 19.9 weeks 20.3 weeks/ 18.8 weeks 23.8 weeks/ weeks 17.4 weeks/ 19.6 weeks weeks 26.8 weeks/ 15.7 weeks 19.5 weeks 20.2 weeks/ 24.1

weeks 25.1 weeks/ 22.9

Duration/ Pre-

sessions treatment 20

20 20 20 20-24 whole 23

63) 16-20 62) 16-20 57)

(n= (n=17) 20 (n=36) 10 (n=60) 16 (n=27) 16 (n=24) 12 16 (n=11) 12 (n=25) 12 (n=48) 16 8

16 12 16 (at

(n=228) 16-20 (modified (n= (n=16) (n=24) 12

(n=38)

(n=12) 12

or (n=226)

(n=57) 12 (n=

(n=120) (n=36) 10

therapy therapy behavioral- 12 therapy therapy therapy therapy behaviour 8 therapy therapy doctor)

(n=40) 24 system

tudies included in the meta-analysis.

(desipramine) (amitriptyline) Pharmacotherapy

(nefazodone) (paroxetine)

clomipramine) (nortryptiline) Pharmacotherapy Pharmacotherapy Pharmacotherapy (phenelzine)

(imipramine) Cognitive Pharmacotherapy

Pharmacotherapy Pharmacotherapy

Pharmacotherapy choice (imipramine)

S HDRS HDRS

Interpersonal Cognitive Cognitive Pharmacotherapy Cognitive Cognitive

Cognitive

et et al.,

and et

1996 therapy 2005

1984 1992 1995 2000 analysis 1999 24-item al., al., al.,

Moore,

21-item

et **

Table 2.1: Study Treatments/N al.,

Blackburn

1989 psychotherapy al., al., al.,

al., et et Jarrett Keller DeRubeis

Blackburn

1981a

Murphy Elkin

19981b (amitriptyline Hautzinger Cognitive

Murphy &

1997 Pharmacotherapy

Hollon

in in

6. 9. 8. 7.

not

has not not not not ≤ ≤ ≤ ≤

scores need are

definition: score months MDD. Study

Two

1986)

phase

Cognitive

depression. HDRS HDRS HDRS HDRS weeks DBI that physicians included included

definition: definition: definition: definition: definition:

meeting condition week. conditions depression. for

Research

Relapse: patients

BDI 2 2

outpatients

groups condition 1 considered by condition condition condition and condition HDRS>13.

At 9.

not not of

review. review.

≤ by outpatients 1986) or

Only Remission weeks

29.

Study 90% chronic treatment this this Criteria recurrent

(Blackburn, meeting definition: definition: definition:

definition:

Relapse: Relapse: Relapse: >

and

symptoms defined

Remission Remission Remission Remission Only Remission

in in 2005b) Relapse: HDRS

two

9 7. 6. 6. 7. or

depression.

MDD

1992)

8. 1992) ≤ ≤ ≤ ≤ ≤

(Simons, acute

separated ≤ and score review. review. included. included.

15. therapy plus placebo Placebo treatment Combined therapy Relaxation therapy Combination therapy Placebo condition Combined therapy Placebo condition Combined therapy Combined therapy 0%)

(24%) (16%) Remission (26%) Remission (%)

(47%) BDI (42%) (18%) (25%) Remission (14%) (21%) (25%) (25%) (44%) >15 (40%) symptoms (33%) HDRS (19%) HDRS (11%) Only ( (41%) (21%) pooled. (36%)

9/60 (15%) 5/12 3/17 9/36 5/36 5/24 8/24 3/16 3/27 0/11 9/25

(%) (n/N)

(32%) 55/228 (46%) 19/120 (29%) 59/226

(40%) (33%) (47%) (24%) 18/38 (58%) (58%) (50%) (33%) (63%) (35%) 10/40 (41%) 16/63 (33%) 25/57 (34%) 25/62 (30%) (100%) (38%) 26/63 (19%) 10/48

(32%)

4/12 8/17 9/38 8/24 8/27 9/48 (n/N) HDRS not included. (Hollon,

relapse indicating > included. included. included. this included. Relapse: criteria

treatment.

BDI

HDRS atypical included. considered HDRS

55/120 (Evans, Diagnostic (Shea, included pharmacotherapy

ITT) ITT) included

post Remission Drop-out Comment

4.79 6.8 8.8 8.6* 21/36 7.7 12/24 8.3 10/16 8.5 14/40 9.8 26/63 9.8 24/63

10.9 15.1** 72/228 10.2* 21/36 10.7 14.2 19/57 10.7 21/62 14.7** 64/226 2.27 11/11 11.4 13.3 8/25 13.3 Remission sample: this

baseline 24/60 17-item

– – – – – – – – – – – – – – – – – – – – and

(completers) IDS (completers) not

(modified mean HDRS

23.4 chronic

20.8 19.6

weeks 25.1 weeks/ 22.9

Duration/ Pre-

sessions treatment 20

20 20 20 20-24 whole 23

63) 16-20 62) 16-20 57)

(n= (n=17) 20 (n=36) 10 (n=60) 16 (n=27) 16 (n=24) 12 16 (n=11) 12 (n=25) 12 (n=48) 16 8

16 12 16 (at

(n=228) 16-20 (modified (n= (n=16) (n=24) 12

(n=38)

(n=12) 12

or (n=226)

(n=57) 12 (n=

(n=120) (n=36) 10

therapy therapy behavioral- 12 therapy therapy therapy therapy behaviour 8 therapy therapy doctor)

(n=40) 24 system

of tudies included in the meta-analysis.

(desipramine) (amitriptyline) Pharmacotherapy

(nefazodone) (paroxetine)

clomipramine) (nortryptiline) Pharmacotherapy Pharmacotherapy Pharmacotherapy (phenelzine)

(imipramine) Cognitive Pharmacotherapy

Pharmacotherapy Pharmacotherapy

Pharmacotherapy choice (imipramine)

S HDRS HDRS

Interpersonal Cognitive Cognitive Pharmacotherapy Cognitive Cognitive

Cognitive

et et al.,

et and et

1996 therapy 2005 et

1984 1992 1995 2000 analysis 1999 24-item al., al., al.,

Moore,

21-item

et **

Table 2.1: Study Treatments/N al.,

Blackburn

1989 psychotherapy al., al., al.,

al., et et Jarrett Keller DeRubeis

Blackburn

1981a

Murphy Elkin

19981b (amitriptyline Hautzinger Cognitive

Murphy &

1997 Pharmacotherapy

Hollon

29 164.77] 1.57] 2.19] 1.49] 4.85] 6.23] 3.57] 2.22] 1.85] 4.19] 4.52]

(fixed)

95% CI 95% [0.63, [1.07, [0.51, [0.79, [0.67, [0.56, [1.01, [0.67, [0.85, [0.61, [0.25, 95%

RR (fixed) RR 10.15

RR 1.29

1.08 1.26

1.06 1.80 1.88

1.89 1.22 1.60 1.06

1.60 [0.61, 4.19] [0.61, 1.60 1.85] [0.85, 1.26 [0.51, 1.06 2.19] 1.06 [0.25, 4.52] [0.25, 1.06 2.22] [0.67, 1.22 164.77] [0.63, 10.15 3.57] [1.01, 1.89 6.23] [0.56, 1.88 4.85] [0.67, 1.80 1.49] [0.79, 1.08 1.57] [1.07, 1.29 % 8.97 3.74 2.87 0.39 7.28 9.35 3.74 2.17 40.94 20.54 100.00 Weight % 2.17 9.35 0.39 7.28 2.87 3.74 40.94 00.00 Weight 3.74 1 20.54 8.97 10 psychotherapy

5 Favours

Favours psychotherapy Favours CI

(fixed) 1

95% 95% CI 95%

RR RR (fixed) RR 0.5 30 0.2

pharmacotherapy 0.1

I²

Favours 6 593 n/N 9/60 5/36 3/27 0/11 9/25 5/24 3/17 10/40 3 55/228 41/125

Favours pharmacotherapy Favours 0.73),

n/N =

(P (Psychotherapy)

Psychotherapy 9

Psychotherapy =

140

0.009)

df I² = 0% I² =

= ),

6.11,

= 640

n/N 9/36 5/12 8/24 2.61 25/57 26/63

59/226 P = 0.73 = P = (

Chi²

Z 640 593 640

Pharmacotherapy n/N (Pharmacotherapy), 18/38 df = 9 9 = df

, 10/48 3/16 19/120 effect:

Pharmacotherap

182

1996 CI)

1997 1981 26/63 41/125 26/63 25/57 9/25 25/57 8/24 5/24 8/24 0/11 5/12 10/40 18/38 3/27 10/48 5/ 9/36 55/228 59/226 9/60 19/120 3/16 3/17 3/16

2005

1995 1984 overall heterogeneity:

1992 1999

2000

events: 1989 (95%

for for

: Chi² = 6.11 = : Chi² y

Test Test Total Total DeRubeis, Keller, Jarrett, Blackburn, Hautzinger, Figure 2.1: Relative Risk of Drop out in Psychotherapy versus Pharmacotherapy. Murphy, Hollon, Elkin, Murphy, Blackburn,

Study eneit g

30 1999 Test for hetero for Test Test for overall effect: Z = 2.61 (P = 0.009) = (P 2.61 = Z effect: overall for Test Elkin, 1989 1989 Elkin, Hollon, 1992 1992 Hollon, CI) (95% Total Murphy, 1984 1984 Murphy, 1995 Murphy, 1996 Hautzinger, 1997 Blackburn, Jarrett, 2000 Keller, 2005 DeRubeis, (Psychotherapy) 140 (Pharmacotherapy), 182 events: Total Study 1981 Blackburn, Figure 2.1 Relative Risk of Drop out in Psychotherapy versus Pharmacotherapy (1981). Hautzinger et al. (1999) report follow-up data in their own publication. Patients did not relapse if they: a) were remitted after acute treatment and b) did not meet criteria for depression at follow-up. There were differences across studies in the definition of relapse (see table 1 for the definitions of remission and relapse per study). In most studies follow-up was naturalistic (i.e. there was no control for receiving treatment during follow-up). In three studies (Simons et al., 1986, Evans et al., 1992 and Shea et al., 1992) the authors provided data on re-entering treatment during follow-up. However, we based our analyses on the relapse data defined by cut-off scores or depression criteria and not on definitions that included ‘re-entering treatment’. There were considerable differences between follow-up phases across studies. Follow-up durations varied from one year (the Simons, the Hautzinger and the Hollon studies), to 1.5 year (the Shea study) and 2 years (the Evans and the Blackburn studies). In the Blackburn study treatment was continued for 6 months in the so-called follow-up period (antidepressants at a normal regimen, psychotherapy at a 6 weekly booster session regimen). In the Shea study both pharmacotherapy and psychotherapy were gradually reduced in 4 to 6 weeks after termination. Furthermore, in this study we combined the results of both psychotherapy conditions, because entering two comparisons for each study in the meta-analysis would violate the assumption that all comparisons in a meta-analysis should be independent (Cooper & Hegdes, 1994). In the follow-up data of the Haut- zinger study no differentiation was made between inpatients and outpatients. In the Simons study medication was tapered off before being discontinued at treatment termination. In the Hollon study, one medication group received a placebo after treatment termination, the other group continued medication. We included only the first group in our analysis. Patients who had been treated with psychotherapy in the Hollon study received three booster sessions. The Evans study had a medication continuation and a non-continuation group. We included only the latter in our meta-analysis.

2.3 Results

Dropout rates The dropout rates are shown in figure 2.1. as can be seen in figure 2.1, the pooled dropout rate in pharmacotherapy (28.43%) is larger than that in psychotherapy (23.60%). The difference (4.83%) is statistically significant (RR=1.29, p=0.009). The chi-square test of heterogeneity indicates a lack of evidence for heterogeneity (p=0.73 and I²=0%).

Efficacy at treatment termination Relative Risk of Remission Figure 2.2 shows the remission rates and Relative Risk (RR) for remission. as can be seen in figure 2.2 the pooled remission rate for psychotherapy (37.94%) is somewhat larger than that of pharmacotherapy (34.84%), but the difference (3.1%) is not statistically significant (pooled RR =0.91, p=0.24). The

31 1.06] 1.65] 1.19] 1.45] 1.48] 0.74] 1.38] 2.05] 1.49] 1.33] 2.50] 79, 1.65] 79, 95% CI 95%

. CI RR (fixed) RR

(fixed)

[0.79, [0.79, [0.68, [0.28, [0.68, [0.15, [0.33, [0.53, [0.69, [0.33, [0.71, 95%

1.33 [0.71, 2.50] [0.71, 1.33 1.33] [0.33, 0.67 1.49] [0.69, 1.01 2.05] [0.53, 1.04 0.74] [0.15, 0.33 1.38] [0.33, 0.68 1.45] [0.28, 0.63 1.48] [0.68, 1.00 1.19] [0.68, 0.90 [0 1.15 1.06] [0.79, 0.91 RR 0.91

% % 4.60 0.63 9.44 1.00 5.16 6.13 0.33 0.68 5.00 1.04 5.40 0.67 3.49 1.33 14.39 1.15 32.23 0.90 14.16 1.01 3.49 5.40 14.16 5.00 5.16 6.13 4.60 9.44 32.23 14.39 00.00 100.00 Weight Weight 1 5 pharmacotherapy

Favours

Favours pharmacotherapy Favours CI

1 (fixed)

95%

95% CI 95% RR

RR (fixed) RR 0.5 32 psychothotherapy 0.2

Favours psychothotherapy Favours

23.7%

Favours =

I²

n/N 593 n/N 8/27 8/25 8/17 24/60 21/36 11/11 14/40 12/24 72/228 47/125 0.23),

Psychotherapy =

(Psychotherapy) 9

Psychotherapy

225 df

0.24)

11.80,

640 593 640 = 640

n/N 4/12 8/24 1.18 21/36 19/57 24/63 n/N

64/226 =

Chi²

Pharmacotherapy (Pharmacotherapy),

8/24 12/24 8/24 47/125 24/63 8/25 19/57 11/11 4/12 14/40 9/38 8/27 9/48 21/36 21/36 72/228 64/226 24/60 55/120 10/16 8/17 10/16 effect:

223

1996 9/38 CI)

1997 9/48 1981 10/16

2005 55/120

1995 1984 overall heterogeneity:

1992 1999

2000

Pharmacotherapy 1989 events: (95%

for for

Figure 2.2: Relative Risk of Remission in Psychotherapy versus Pharmacotherapy.

Test Test Total Total DeRubeis, Jarrett, Keller, Hautzinger, Blackburn, Murphy, Hollon,

Elkin, Murphy, Study Blackburn, 1999 32 Murphy, 1984 1984 Murphy, 1989 Elkin, 1992 Hollon, 1995 Murphy, 1996 Hautzinger, 1997 Blackburn, Jarrett, 2000 Keller, 2005 DeRubeis, (Psychotherapy) 225 23.7% = I² 0.23), (Pharmacotherapy), = (P 223 9 = df events: 11.80, = Total Chi² 0.24) heterogeneity: = for (P Test 1.18 = Z effect: overall for Test Total (95% CI) (95% Total Study 1981 Blackburn, Figure 2.2 Relative Risk of Remission in Psychotherapy versus Pharmacotherapy Table 2.2: ITT remission rates in studies considering chronic¹ depression and studies considering non-chronic² depression. Psychotherapy Pharmacotherapy Significance Studies considering chronic depression 36.11% 36.64% RR=0.98, p=0.83

Studies considering non-chronic depression 41.14% 32.17% RR=0.83, p=0.12

Significance RR=0.90, p=0.31 RR=1.14, p=0.25 ¹ Jarret et al., 1999, Keller et al., 2000, DeRubeis et al., 2005. ² Blackburn et al., 1981, Murphy et al., 1984, Elkin et al, 1989, Hollon et al., 1992, Murphy et al., 1995, Hautzinger et al., 1996, Blackburn&Moore, 1997. chi-square test of heterogeneity indicates a lack of evidence for heterogeneity (p=0.23 and I²=23.7%).

Chronicity Table 2.2 separately shows the Relative Risk for remission in the three studies of chronic depression and in the seven studies of non-chronic depression. as can be seen in table 2.2, the pooled remission rates of psychotherapy and pharmacotherapy do not differ significantly in chronic depression (36.11% and 36.64%, respectively, p=0.83) and in non-chronic depression (41.14% and 32.17%, respectively, p=0.12). In both analyses the chi-square test of heterogeneity indicates that there is no evidence for heterogeneity (non-chronic: p=0.14/ I²=38% and chronic: p=0.58/ I²=0%). It also appears that the pooled remission rates of chronic and non-chronic depression do not differ significantly for psychotherapy (36.11% and 41.14%, respectively, p=0.31) and not for pharmacotherapy (36.64% and 32.17%, respectively, p=0.25). In the last two analyses, heterogeneity was not an issue because we made only one comparison between two groups of studies.

Severity Table 2.3 separately shows the Relative Risk for remission in the five studies of mild depression and in the five studies of moderate depression. the pooled remission rates of psychotherapy and pharmacotherapy do not differ significantly in mild depression (46.47% and 44.37%, respectively, p=0.34) and not in moderate depression (33.15% and 31.90%, respectively, p=0.44). In the analysis of moderate depression no evidence for heterogeneity was found (p=0.55/ I²=0%) but in the analysis of mild depression the chi-square test of heterogeneity indicated moderate heterogeneity (p=0.07/ I²=54.3%). This is possibly due to the outlying results of Murphy et al. (1984) which, compared to the other studies, show a larger difference in remission between psychotherapy (100%) versus pharmacotherapy (33%). The pooled remission rates of mild and

33 Table 2.3: ITT remission rates in studies considering mild¹ depression and studies considering moderate² depression. Psychotherapy Pharmacotherapy Significance Studies considering mild depression 46.47% 44.37% RR=0.90, p=0.34

Studies considering moderate depression 33.15% 31.90% RR=0.92, p=0.44

Significance RR=1.40, p=0.001 RR=1.39, p=0.003 ¹ Blackburn et al., 1981, Murphy et al., 1984, Murphy et al.,1995, Elkin et al, 1989, Jarrett et al., 1999. ² Hollon et al., 1992, Hautzinger et al., 1996, Blackburn&Moore, 1997, Keller et al., 2000 DeRubeis et al., 2005. moderate depression do differ significantly, both for psychotherapy (46.47% and 33.15%, respectively, p=0.001) and pharmacotherapy (44.37% and 31.90%, respectively, p=0.003). In the last two analyses, heterogeneity was not an issue beacuse we made only one comparison between two groups of studies.

Odds Ratio of Remission Figure 2.3 shows the Odds Ratios (OR) for remission. as can be seen in figure 2.3, the pooled OR is 0.87, and the difference between pharmacotherapy and psychotherapy is not statistically significant (p=0.24). The chi-square test of heterogeneity indicates no evidence for heterogeneity (p=0.30 and I²=16%). the odds ratios in subanalyses regarding chronicity and severity of depression do not indicate any statistically significant differences between the two treatments in chronic (p=0.82) and non-chronic (p=0.12) depression nor in mild (p=0.35) and moderate (p=0.44) depression. All chi-square tests indicate a lack of evidence for heterogeneity (p=0.58, p=0.17, p=0.10, p=0.58, respectively).

Numbers needed to treat Pooled data show that 32 patients would need to be treated with psychotherapy to produce one recovery from depression which would not have occurred had they been given antidepressants (NNT = 32) (1/0.031).

Efficacy at follow up Figure 2.4 shows the relative risk of relapse during follow up. there is a statistically significant difference (RR=0.46, p<0.0001) between the pooled relapse rate of pharmacotherapy (56.56%) and that of psychotherapy (26.51%). The chi-square test of heterogeneity indicates that the results lack evidence of heterogeneity (p=0.68, I²=0%). As there was considerable clinical heterogeneity in the follow-up phases across studies, we performed various

34 95% CI 95% 1.10] 2.38] 1.28] 1.64] 2.55] 0.49] 1.55] 2.90] 1.91] 1.61] 7.53] OR (fixed)

(fixed)

[0.68, [0.68, [0.57, [0.18, [0.39, [0.00, [0.21, [0.39, [0.55, [0.16, [0.47, 95%

1.88 [0.47, 7.53] 0.50 [0.16, 1.61] 1.02 [0.55, 1.91] 1.06 [0.39, 2.90] 0.02 [0.00, 0.49] 0.58 [0.21, 1.55] 0.55 [0.18, 1.64] 1.00 [0.39, 2.55] 0.86 [0.57, 1.28] 1.27 [0.68, 2.38] 0.87 [0.68, 1.10] OR 0.87

% % 2.05 5.64 13.75 5.23 5.51 7.34 5.87 6.17 36.23 12.22 00.00 5.87 0.55 6.17 1.00 5.51 7.34 0.02 0.58 5.23 1.06 5.64 0.50 2.05 1.88 12.22 1.27 36.23 0.86 13.75 1.02 100.00 Weight 1 Weight 5 pharmacotherapy

Favours

Favours pharmacotherapy Favours CI

(fixed) 1

95% 95% CI 95% OR

OR (fixed) 0.5 38 psychotherapy

0.2

Favours 16.0%

Favours psychotherapy Favours

I²

n/N 593 n/N 8/27 8/25 8/17 24/60 21/36 11/11 14/40 12/24 72/228 47/125 0.30),

1/36 2 =

Psychotherapy (P

(Psychotherapy) 9

Psychotherapy

225 df

0.24)

10.72,

= 640 593 640

n/N 4/12 8/24 1.18 21/36 19/57 24/63

64/226 n/N =

Chi²

Pharmacotherapy (Pharmacotherapy), Pharmacotherapy

effect: 10/16 8/17 8/24 12/24 24/63 47/125 19/57 8/25 4/12 11/11 9/38 14/40 9/48 8/27 21/36 64/226 72/228 55/120 24/60

223

1996 9/38 CI)

1997 9/48 1981 10/16

2005 55/120

1995 1984 overall heterogeneity:

1992 1999

2000

1989 events: (95%

for for

Test Test Total Total DeRubeis, Jarrett, Keller, Hautzinger, Blackburn, Murphy, Hollon, Elkin, Murphy, Blackburn,

Study Figure 2.3: Odds Ratio of Remission Rates of Psychotherapy versus Pharmacotherapy.

35 : Study 1981 Blackburn, 1984 Murphy, 1989 Elkin, 1992 Hollon, 1995 Murphy, 1996 Hautzinger, 1997 Blackburn, 1999 Jarrett, 2000 Keller, 2005 DeRubeis, CI) (95% Total Total events: 223 (Pharmacotherapy), 225 (Psychotherapy) 225 (Pharmacotherapy), 223 events: Total 16.0% = I² 0.30), = (P 9 df = 10.72, = Chi² heterogeneity: for Test 0.24) = (P 1.18 = Z effect: overall for Test Figure 2.3 Odds Ratio of Remission Rates Psychotherapy versus Pharmacotherapy 0.65] 0.69] 1.39] 1.14] 1.60] 6.62] 1.33]

95% CI 95% (fixed)

RR (fixed) RR [0.33, [0.24, [0.04, [0.17, [0.10, [0.01, [0.40, 95%

RR 0.46

0.73 [0.40, 1.33] 1.33] [0.40, 0.73 1.60] [0.10, 0.40 6.62] [0.01, 0.30 1.14] [0.17, 0.44 1.39] [0.04, 0.22 0.69] [0.24, 0.41 0.65] [0.33, 0.46

% 8.45 0.22 8.21 0.40 2.58 0.30 44.35 0.41 15.43 0.44 20.98 0.73 100.00 Weight % 20.98 8.21 2.58 15.43 8.45 44.35 00.00 Weight 1 10 pharmacotherapy

2 Favours

Favours pharmacotherapy Favours CI

1 (fixed)

95%

95% CI 95%

RR RR (fixed) RR 0.5 40 0.2 psychotherapy

0.1 Favours psychotherapy Favours

Favours 0%

8 I²

n/N 99 6/8 1/9 n/N 8/19 5/10 9/18 27/35 0.68),

= Pharmacotherapy

Pharmacotherapy (Pharmacotherapy)

0.0001)

3.13,

132 99 132

= 132

n/N 0/10 2/10 n/N 4.38 11/35 16/44

Chi²

Psychotherapy 16/44 9/ 16/44 2/10 5/10 5/10 2/10 1/9 0/10 8/19 5/27 6/8 1/6 27/35 11/35 (Psychotherapy), effect:

1996 5/27 CI)

1986 1/6

overall heterogeneity: 1986

2005

1992 Psychotherapy

events: (95% 1992

for for

Test Test Total Total Hollon, Blackburn, Simons, Hautzinger, Shea, Evans,

Study Figure 2.4: Relative Risk of Relapse Rates of Psychotherapy versus Pharmacotherapy.

36 Study 1992 Shea, Evans, 1992 1992 Evans, 1986 Simons, 1996 Hautzinger, 1986 Blackburn, 2005 Hollon, CI) (Pharmacotherapy) (95% 56 Total 0% = I² (Psychotherapy), 35 0.68), = (P 5 events: = df Total 3.13, = Chi² heterogeneity: for 0.0001) < Test (P 4.38 = Z effect: overall for Test Figure 2.4 Relative Risk of Relapse Rates analyses on subgroups of studies. First, we excluded the study of Hollon et al. (2005). We consider it an outlier because the patients treated with medication received placebo throughout the follow-up period. Second, we discriminated between follow-up durations (combining the studies with one year follow-up, and combining those with 1.5 to 2 years follow-up). All sub analyses showed results similar to those of the overall analysis, i.e., a significant difference in favour of psychotherapy. In none of these analyses the homogeneity hypothesis was rejected.

2.4 Discussion

We performed a meta-analysis comparing psychotherapy and pharmacotherapy in the treatment of adult psychiatric outpatients suffering from mild to moderate major depression. In contrast to existing reviews, our meta-analysis furthered homogeneity of the included studies by applying strict clinical inclusion and exclusion criteria. We performed statistical tests a posteriori supporting our argument that the included studies were indeed sufficiently homogeneous. In addition, we took into account two potential determinants of treatment prognosis by performing subanalyses on chronicity and severity of depression. psychotherapy and pharmacotherapy appeared equally effective at treatment termination. This means that in the long-standing controversy regarding the relative effectiveness of both treatment modalities, our results support the ‘no difference’ point of view. according to clinical lore, chronicity and severity influence the relative effectiveness of the two therapeutic modalities. However, we found no differences in efficacy between both treatments in chronic and non-chronic and in mild and moderate depression. Understandably, but unfortunately, we found no data regarding severe depression. Our results show that severity, in contrast to chronicity, affects the efficacy of both treatments. They have superior results in mild depression compared to moderate depression. This may indicate that monotherapies are not the first choice in moderate depression (HDRS>20). This hypothesis is supported by the findings of Thase et al. (1997), who report a statistically significant and clinically relevant difference in favour of combined therapy over psychotherapy in ‘more severe’ (HDRS>19 points) but not in less severe (HDRS<20) depression. Several reviews and meta-analyses (Hollon et al., 2005a, Pampallona et al., 2004; Friedman et al., 2004, Hegerl et al., 2004) report superior results of combined treatment compared to medication alone, especially for more severe depressed patients. the parity in efficacy found at treatment termination does not seem to last beyond actual treatment. Our follow up data show that twice as many patients relapse after pharmacotherapy termination than after psychotherapy termination. According to our results, the idea that short-term therapies yield short- lived effects applies more to pharmacotherapy than to psychotherapy. The difference might even be bigger than presented here, for in 2 of the 6 follow-up

37 studies we included (Blackburn et al., 1986; Hollon et al., 1992) the study design seems to favor pharmacotherapy above psychotherapy. In the Blackburn study pharmacotherapy was continued for 6 months while psychotherapy was provided only at a 6 weekly booster session regimen. In the Hollon study medication was substituted by placebo during the follow-up period while patients treated with psychotherapy received only three booster sessions. Furthermore, in the Evans and Simons studies, more pharmacotherapy patients than psychotherapy patients sought treatment during follow-up, possibly indicating a relapse that was not accounted for in our relapse data. Our findings regarding relapse are comparable to those reported in the reviews of Hollon et al. (2005a) and Gloaguen et al. (1998). In addition, the follow-up studies of Hollon et al. (2005b) and Evans et al. (1992) show that psychotherapy patients are no more likely to relapse than pharmacotherapy patients who keep taking medication. In our opinion, our relapse data, apart from obvious clinical implications, are highly relevant for establishing cost-benefit ratios, a topic that is not addressed in this review, nor in the included RCT’s. We found that dropout rates in pharmacotherapy are significantly higher than in psychotherapy, although the difference (5%) is not impressive. As researchers and clinicians alike know, medication nonadherence is a major problem in pharmacotherapy. Still, psychotherapy too is beset with the problem of non-compliance, because 20% to 25% of patients drops out. Our review has several limitations. First, conclusions based only on the results of RCT’s have wellknown limitations. An obvious one is selection bias. RCT’s leave patients with serious comorbidity, such as drug dependence, suicide intentions or severe personality disorders out of scope. In fact, the majority of suitable patients do not end up in RCT’s due to all in- and exclusion criteria that have to be met. Keitner et al. (2003), for example, mention that only 14.5 % of eligible depressed patients eventually took part in an RCT. Second, our meta- analysis only compares psychotherapy with pharmacotherapy leaving comparisons with combined therapy out of scope. Third, efficacy was measured with the HDRS only. Most of the studies we found did not assess social functioning or quality of life, which are the ultimate goals of therapy. Some studies measured depression with other scales, such as the Beck Depression Inventory (BDI, Beck et al., 1961). Because it was not our aim to compare scales, and we used the HDRS as inclusion criterion, we did not perform a meta-analysis on the BDI. However, it certainly would be interesting to perform meta-analyses based on both scales. Fourth, the methodological quality of the included studies varied. Some studies of acute treatment (Blackburn et al., 1981 and Murphy et al., 1995) and all follow-up studies were characterised by small sample sizes, lacking statistical power to detect differences. Although some studies (e.g. the Keller study) controlled for medication compliance, most did not. In short, only the two more recent studies correspond well to actual research criteria (Sackett, 1998; Hegerl et al., 2004). Fifth, allegiance effects (Gaffan et al., 1995) cannot be excluded. perhaps more important than these limitations is the fact that we based our conclusions concerning severity of depression on mean baseline scores of the studies, not on individual patient data. We are aware of the fact that

38 this is a rather rough division of a spectrum. Nevertheless, our results do not seem to diverge from findings based on individual patients. Blackburn & Moore (1997), Hollon et al. (1992), Hautzinger et al. (1999) and Elkin et al. (1989) performed sub-analyses on severity. They too did not find significant differences between psychotherapy and pharmacotherapy in less severe (HDRS<20) and more severe (HDRS>19) depressed patients. Our HDRS cut-off scores for the distinction between mild, moderate and severe depression are in accordance with what is mostly found in literature. Unfortunately, Hamilton himself did not define cutting scores for his scale. The result is that there is no generally accepted definition of mild, moderate and severe depression. According to clinical lore, mild depression and moderate depression range respectively from 12-14 to 18-20, and from 18-20 to 24-26 17-item HDRS points. finally, although our approach explicitly aimed to further homogeneity, it cannot be denied that the included studies still present some heterogeneity. Psychotherapy includes cognitive therapy and interpersonal therapy; pharmacotherapy includes TCA’s and SSRI’s. The importance of this point, however, may be limited as meta-analytic studies found no significant differences between TCA’s and SSRI’s (Anderson, 2000) or between psychodynamic and cognitive-behavioural therapies (Leichsenring, 2001) in the treatment of depressed outpatients. Only one study in our meta-analysis considered IPT; the remaining studies applied cognitive behaviour therapy. Therefore it may be argued that our conclusions do not merit psychotherapy per se, but apply mainly to cognitive therapy. Treatment durations varied, and, although the differences in psychotherapy sessions across studies are limited, the efficacy of pharmacotherapy might be somewhat weighted down by studies with relatively short treatment periods (e.g. Hautzinger et al., 1996). The definition of remission differs per study, with cut-off scores varying from 6 to 9 HDRS points. This is quite a disparity as, according to Jonghe and Swinkels (2005), most researchers agree that a difference of 3 points borders on clinical significance. The follow-up studies are obviously heterogeneous in various clinical aspects, so much so that we were somewhat surprised to find all the tests in this area indicating strong homogeneity of the results. There are no generally accepted definitions of relapse. Most authors applied more or less strict HDRS or MDD criteria, but some, e.g. Shea et al. (1992), Evans et al. (1992) and Simons at al. (1986), included ‘re-entering treatment’ in one of their relapse definitions. It appears that more pharmacotherapy patients than psychotherapy patients re- enter treatment during follow-up. This might have underestimated the relapse rates for pharmacotherapy in our study. All in all, pooling these data is debatable and interpretation of the relapse rates should be done cautiously. Our results, however, are corroborated by the studies of Hollon et al. (2005a) and of Gloaguen et al. (1998), who simply listed relapse rates of individual trials, not pooling the data, and came to comparable conclusions. We conclude that depressed patients equally profit from psychotherapy and pharmacotherapy after short-term treatment. Furthermore, it may be concluded that they seem to benefit more from psychotherapy than from pharmacotherapy during the one to two year follow-up period.

39 References

American Psychiatric Association (1980). Diagnostic and Statistical Manual of Mental Disorders (3rd edn) (DSM-III). Washington DC: APA. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders. Fourth edition (DSM-IV). Washington: APA. Anderson, I.M. (2000). Selective serotonin reuptake inhibitors versus tricyclic antidepressants: a meta-analysis of efficacy and tolerability. Journal of Affective Disorders, 58, 19-36. Beck, A.T., Ward, C.H., Mendelson, M., Mock, J., Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561-571. Blackburn, I.M., Bishop, S., Glen, A.I.M., Whalley, L.J., Christie, J.E. (1981a). The efficacy of cognitive therapy in depression: A treatment trial using cognitive therapy and pharmacotherapy, each alone and in combination. British Journal of Psychiatry, 139, 181-189. Blackburn, I.M., Bishop, S. (1981b). Is there an alternative to drugs in the treatment of depressed ambulatory patients? Behavioural Psychotherapy, 9, 96-104. Blackburn, I.M., Eunson, K.M., Bishop, S. (1986) A two-year naturalistic follow-up of depressed patients treated with cognitive therapy, pharmacotherapy and a combination of both. Journal of Affective disorders, 10, 67-75. Blackburn, I.M., & Moore, R.G. (1997). Controlled acute and follow-up trial of cognitive therapy and pharmacotherapy in out patients with recurrent depression. British Journal of Psychiatry, 171, 328-334. Casacalenda, N., Perry, J.C., Looper, K. (2002). Remission in Major Depressive Dis- order: A comparison of pharmacotherapy, psychotherapy, and control conditions. American Journal of Psychiatry, 159: 1354-1360. Cochrane Collaboration, Cochrane Reviewers Handbook. Available at: http://cochrane. com. Accessed May, 2004. Cooper, H., & Hedges, L.V. (Eds.)(1994). The Handbook of Research Synthesis. NY: Sage. DeRubeis, R.J., Gelfand, L.A., Tang, T.Z., Simons, A.D. (1999). Medication versus cognitive behavior therapy for severely depressed outpatients: mega-analysis of four randomized comparisons. American Journal of Psychiatry, 156: 1007-1013. DeRubeis, R.J., Hollon, S.D., Amsterdam, J.D., Shelton, R.C., Young P.R., Salomon, R.M., O’Reardon, J.P., Lovett, M.L., Gladis, M.M., Brown, L.L., Gallop, R. (2005). Cognitive therapy vs medication in the treatment of moderate to severe depression. Archives of General Psychiatry, 62: 409-416. Dobson, K.S. (1989). A meta-analysis of the efficacy of cognitive therapy for depression. Journal of Consulting and Clinical Psychology, 57: 414-419. Elkin, I., Shea, M.T., Watkins, J.T., Imber, S.D., Sotsky, S.M., Collins, J.F., Glass, D.R., Pilkonis, P.A., Leber, W.R., Docherty, J.P., Fiester, S.J., Parloff, M.B. (1989). National Institute of Mental Health Treatment of Depression Collaborative Research Program. Archives of General Psychiatry, 46, 971-982. Evans, M.D., Hollon, S.D., DeRubeis, R.J. et al (1992). Differential relapse following cognitive therapy and pharmacotherapy for depression. Archives of General Psy- chiatry, 49, 802-808. Friedman, M.A., Detweiler-Bedell, J.B., Leventhal. H.E., Horne, R., Keitner, G.I., Miller, I.W. (2004). Combined psychotherapy and pharmacotherapy for the treatment of major depressive disorder. Clinical Psychology: Science and Practice, 11, 47-68. Gaffan, E.A., Tsaousis, I., Kemp-Wheeler, S.M. (1995). Researcher allegiance and meta- analysis: the case of cognitive therapy for depression. Journal of Consulting and Clinical Psychology, 6, 966-980 Gloaguen, V., Cottraux, J., Cucherat, M., Blackburn, I. (1998). A meta-analysis of the effects of cognitive therapy in depressed patients. Journal of Affective Disorders, 49: 59-72.

40 Hamilton, M. (1967). Development of a rating scale for primary depressive illness. Brit- ish Journal of Social and Clinical Psychology, 6, 278-296. Hegerl, U., Plattner, A., Möller, H.J. (2004). Should combined Pharmaco- and psychotherapy be offered to depressed patients? A qualitative review of randomized clinical trials from the 1990s. European Archives Psychiatry Clinical Neurosci, 254: 99-107. Hautzinger, M., de Jong-Meyer, R., Treiber, R.,Rudolf, G.A.E., Thien, U. (1996). Wirk- samkeit Kognitiever Verhaltenstherapie, Pharmakotherapie und deren Kombination bei nicht-endogenen, unipolaren Depressionen. Zeitschrift für Klinische psychologie, 25(2): 130-145. Hollon, S.D., Shelton, R.C., Loosen, P.T. (1991). Cognitive therapy and pharmacotherapy for depression. Journal of Consulting and Clinical Psychology, 59: 88-99. Hollon, S.D., DeRubeis, R.J., Evans, M.D., Wiemer, M.J., Garvey, M.J., Grove, W.M., Tuason, V.B. (1992). Cognitive therapy and Pharmacotherapy for Depression. Archives of General Psychiatry, vol. 49, 774-781. Hollon, S.D., Thase, M.E., Markowitz, J.C. (2002). Treatment and prevention of depression. Psychological science in the public interest, 3: 39-76. Hollon, S.D., Jarrett, R.B., Nierenberg, A.A., Thase, M.E., Trivedi, M., Rush, A.J. (2005a). Psychotherapy and medication in the treatment of adult and geriatric depression: which monotherapy or combined treatment? Journal of Clinical Psychiatry, 66: 455-468. Hollon, S.D., DeRubeis, R-J., Shelton, R.C., Amsterdam, J.D., Salomon, R.M., O’Reardon, J.P., Lovett, M.L., Young, P.R., Haman, K.L., Freeman, B.B., Gallop, R. (2005b). Pre- vention of Relapse following cognitive therapy vs medications in moderate to severe depression. Archives of General Psychiatry, 62: 417-422. Jarret, R.B. (1995). Comparing and combining short-term psychotherapy and pharmacotherapy for depression. In E.E. Beckman & W.R. Leber (red.), Handbook of depression, New York: Guilford Press. Jarret, R.B., Schaffer, M., McIntire, D., Witt-Browder, A., Kraft, D., & Risser, R.C. (1999). Treatment of atypical depression with cognitive therapy of phenelzine. Archives of General Psychiatry, vol. 56, 431-437. Jonghe, F., de, Swinkels, J. (2005). Antidepressiva. Een leidraad voor het rationeel omgaan met antidepressiva bij de behandeling van patienten met een depressie. Benecke N.I., Amsterdam. Keitner, G.I., Posternak, M.A., Ryan, C.E. (2003). How many subjects with major depressive disorder meet eligibility requirements of an antidepressant efficacy trial? Jour- nal of Clinical psychiatry, 59, 598-607. Keller,M.B., McCullough, J.P., Klein D.N., Arnow, B., Dunner, D.L., Geleberg, A.J., Markowitz, J.C., Nemeroff, C.B., Russel, J.M., Thase, M.E., Treivedi, M.H., Zajecka, J. (2000). A comparison of Nefazodone, the Cognitive Behavioral-Analysis System of Psychotherapy and their combination for the treatment of chronic depression. The New England Journal of Medicine, 342, 1462-1470. Leichsenring, F. (2001) Comparative effects of short-term psychodynamic psychotherapy and cognitive-behavioral therapy in depression: a meta-analytical approach. Clinical psychology Review, 21 (3), 401-419. Murphy, G.E., Simons, A.D., Wetzel, R.D., Lustman, P.J. (1984). Cognitive therapy and pharmacotherapy. Archives of General Psychiatry, 41, 33-41. Murphy, G.E., Carney, R.M., Knesevich, M.A., Wetzel, R.D., Withworth, P. (1995). Cog- nitive behavior therapy, relaxation training, and tricyclic antidepressant medication in the treatment of depression. Psychological Reports, 77, 403-420. O’Sullivan, R.L., Fava, M., Agustin, C., et al. (1997). Sensitivity of the six-item Hamilton Depression Rating Scale. Acta Psychiatrica Scandinavcia, 95, 379-384. Pampallona, S., Bollini, P., Tibaldi, G., Kupelnick, B., Munizza, C. (2004). Combined pharmacotherapy and psychological treatment for depression. A systematic review. Archives of General psychiatry; 61: 714-719.

41 Robinson, L.A., Berman, J.S., Neimeyer, R.A. (1990). Psychotherapy for the treatment of depression: a comprehensive review of controlled outcome research. Psychological Bulletin, 108: 30-49. Royal Australian and New Zealand College of Psychiatrists (1983). A treatment outline for depressive disorders, The Quality Assurance Project. Australian and New Zea- land Journal of Psychiatry, 17: 129-146. Sackett (1998). Evidence Based Medicine. Spine, 23(10):1085-6. Shea, M.T., Elkine, I., Imber, S.D., Sotsky, S.M., Watkins, J.T., Collines, J.F., Pilkonis, P.A., Bechham, E., Glass, D.R., Dolan, R.T., Parloff, M.B. (1992). Course of depressive symptoms over follow-up. Findings from the national Institute of Mental Health Treatment of depression collaborative research program. Archives of General Psy- chiatry, 49: 782-787. Simons, A.D., Murphy, G.E., Levine, J.L., Wetzel, R.D.l (1986). Cognitive therapy and pharmacotherapy for depression. Sustained improvement over one year. Archives of General Psychiatry, 43, 43-48. Spitzer, R.L.Robins, E. (1978). Research Diagnostic Criteria: Rationale and reliability. Archives of General Psychiatry, 35, 773-782. Steinbrueck, S.M., Maxwell, S.E., Howard, G.S. (1983). A meta-analysis of psychotherapy and drug therapy in the treatment of unipolar depression with adults. Journal of Consulting and Clinical Psychology, 51: 856-863. Thase, M.E., Greenhouse, J.B., Frank, E., Reynolds III, C.F., Pilkonis, P.A., Hurley, K., Grochocinsky, V., Kupfer, D.J. (1997). Treatment of major depression with psychotherapy or psychotherapy-pharmacotherapy combinations. Archives of General Psychiatry, 54, 1009-1015. Weissman, M.M., Jarrett, R.B., Rush, J.A. (1987). Psychotherapy and its relevance to the pharmacotherapy of major depression: a decade later (1976-1985). In: H.Y. Meltzer (ed.), Psychopharmacology: The third generation of progress. New York: Raven Press. Wexler, B.E., & Cicchetti, D.V. (1992). The outpatient treatment of depression. Implica- tions of outcome research for clinical practice. The Journal of nervous en mental disease, 180, 277-286.

42 Appendix

Figure A.1: QUOROM flow diagram.

Potentially relevant RCT’s identified in literature search and screened for retrieval (n=1.601)

RCT’s excluded based on screening of titles and abstracts (n =1.526)

RCT’s relevant for more detailed evaluation (n =75)

Publications excluded based on primary care (n=5), age (n=5), design (n=13), therapy (n=12), inpatients (n=6), other (n=8). Total: 49 Second screening of publications retrieved and screened (n =26)

Publications excluded based on therapy (n=6), inpatients (n=2), double publication (n=2), methodological (n=6). Total: 16 RCT’s included in meta-analysis (n = 10)

Chapter 3 Relative efficacy of psychotherapy and combined therapy in the treatment of depression: a meta-analysis

Saskia de Maat, Jack Dekker, Robert Schoevers, Frans de Jonghe European Psychiatry (2007) 22, 108

Abstract

Background: Reviews of the relative efficacy of psychotherapy and combined therapy (psychotherapy with pharmacotherapy) for depression have yielded contradicting conclusions. This may be explained by the clinical heterogeneity of the studies reviewed. Aims: To conduct a meta-analysis with an acceptable level of homogeneity in order to investigate the relative efficacy of psychotherapy and combined therapy in the acute treatment of depression. Method: A systematic search was performed for RCTs published between 1980 and 2005 comparing psychotherapy and combined therapy in adult psychiatric outpatients with non-psychotic unipolar major depressive disorder. The studies were classified according to the chronicity and severity of the depression. Data were pooled by means of meta-analysis and statistical tests were conducted to measure heterogeneity. Results: The meta-analysis included seven studies looking at a total of 903 patients. None of the heterogeneity tests established significance. This indicates a lack of evidence for the heterogeneity of the results. The dropout rates did not differ significantly between the two treatment modalities (25% in combined therapy and 24% in psychotherapy, p=0.77). At treatment termination, the intention-to-treat remission rate for combined therapy (46%) was better than for psychotherapy (34%)(p= 0.0007); Relative Risk 1.32 (95% CI: 1.12-1.56), Odds Ratio 1.59 (95% CI: 1.22-2.09). In moderate depression, the difference between the remission rates for combined therapy and psychotherapy was statistically significant (47% compared to 34% respectively, p=0.001). This was not the case in mild major depression (42% compared to 37% respectively, p=0.29). The difference was also statistically significant in chronic major depression (48% compared to 32%, p<0.001), but not in non-chronic major depression (43% compared to 37%, p=0.22). On a more specific level, no differences were found in the remission rates for the treatment modalities in mild or moderate non-chronic depression. Combined therapy led to significantly better results than psy-

45 chotherapy in moderate chronic depression only (48% compared to 32%, p<0.001). Conclusions: In the acute treatment of adult psychiatric outpatients with major depressive disorder, patient compliance with combined therapy matches compliance with psychotherapy alone. Combined therapy is more efficacious than psychotherapy alone. However, these results depend on severity and chronicity. Combined therapy outperformed psychotherapy in moderate chronic depression only. No differences were found in mild and moderate non-chronic depression. No data were found for mild chronic depression and for severe depression.

3.1 Introduction

There have been only a few head-to-head comparisons of psychotherapy and combined therapy in the treatment of depression, and the existing evidence is contradictory. It has been argued that few trials have had enough statistical power to detect differences between the two treatment modalities (Jindal & Thase, 2003). There have therefore been several attempts to pool the data in meta-analyses. This has not settled the issue, however, as some reviewers did find significant differences in favour of combined treatment (Jindal & Thase, 2003, Segal et al., 2002, Friedman et al., 2004) while others did not (Conte et al., 1986, Robinson et al., 1990, Wexler & Cicchetti, 1992, Jarret, 1995, Blom et al., 2000). the clinical heterogeneity of the studies included is a striking characteristic of most of the reviews and meta-analyses and this may partially explain their conflicting conclusions. Clinical heterogeneity refers to differences between studies in patient samples, treatment protocols and clinical settings. the recent meta-analysis and qualitative review performed by Friedman et al. (2004) is an example. The authors included twenty studies comparing psychotherapy and pharmacotherapy with combined therapy. Twelve of these compared psychotherapy with combined therapy. While most of these studies dealt with acute treatment, three (Blackburn et al., 1986, Evans et al., 1992, Simons et al., 1986) looked at follow-up. Studies examining psychiatric outpatients were combined with studies examining primary care patients (Mynor-Wallis et al., 2000). One study used a sequential treatment design (Frank et al., 1990). Most studies involved formal psychotherapies, but one involved a problem-solving therapy. Friedman et al. expressed their findings in weighted average comparative effect sizes (Cohen, 1988) and found effect sizes ranging from 0.10 to 0.69 in favour of combined therapy. Another example of clinical heterogeneity is the review of Blom et al. (2000). The authors included studies of psychiatric outpatients and studies of primary care patients. Some of the studies dealt with specific age groups, for instance elderly patients (Beutler et al., 1987, Reynolds et al., 1999). A wide array of therapies was included that ranged beyond formal psychotherapy, e.g. relaxation training (Rush & Watkins, 1981), self-control

46 therapy (Roth et al., 1982) and social skills training (Bellack et al., 1981). Some studies did not compare combined therapy with psychotherapy alone, but with psychotherapy plus placebo (Hersen et al., 1984), or treatment as usual (Teas- dale et al., 1984, Scott & Stradling, 1990). The reviewers concluded on the basis of all the studies that there was no difference between single and combined treatment. clinical heterogeneity makes the pooling of data by means of meta-analysis hazardous. The Cochrane Collaboration (Cochrane Reviewers Handbook, Cochrane Collaboration, 2004) states that ‘the use of meta-analysis to describe the size of effect may not be meaningful if the implementations are so diverse than an effect estimate cannot be interpreted in any specific context. (…) If studies are clinically diverse then a meta-analysis may be meaningless and genuine differences in effects may be obscured.’ They add: ‘Decisions concerning what should and should not be combined are inevitably subjective and are not amenable to statistical solutions, but require discussion and clinical judg- ment.’ the clinical heterogeneity of reviews looking at combined therapy and psychotherapy for depression could provide a partial explanation of the inconsistent conclusions and does not allow specific conclusions regarding certain groups of patients in certain settings. There is evidence, for example, that depressed primary care patients are significantly different from depressed psychiatric outpatients in clinical terms. A large survey (Spijker et al., 2002) in the Netherlands showed that 50% of the participants recovered within 3 months and 63% within 6 months. These patients are likely to be primary care patients. As primary care is the first step in the health care system, only patients who do not recover spontaneously or who do not recover with help from their GP will end up in psychiatric care. In general, psychiatric outpatients constitute a more severely and persistently ill group than primary care patients. in this article we present the results of a meta-analysis based on RCTs published between 1980 and 2005, comparing psychotherapy to combined therapy in adult psychiatric outpatients with nonpsychotic unipolar depression. We tried to establish a high level of clinical homogeneity across the studies we selected by applying quite strict inclusion criteria for patient samples, diagnosis and treatment settings (see Appendix). We then carried out statistical tests of the heterogeneity of the studies included to assess the extent to which we had been successful. Studies were therefore selected on the basis of clinical criteria only. Statistical heterogeneity analysis was not used as a selection criterion but for retrospective testing. The primary research question relates to the relative efficacy of psychotherapy and combined therapy in the acute treatment of depression. The secondary question relates to possible differences in dropout rates during treatment. We took into consideration two variables known to influence treatment prognosis – chronicity and severity – and therefore differentiated between mild, moderate and severe depression, and between chronic and non-chronic depression.

47 3.2 Methods

The method, selection criteria and data used for the literature search are described in the Appendix. The main outcome of the meta-analysis was efficacy at treatment termination, expressed in remission rates based on the intention-to-treat sample (ITT, i.e., a sample consisting of all randomized patients). Remission rates were pooled by calculating the Relative Risk (RR) and the Odds Ratios (OR). The RR is the ratio between the risk of an event (e.g. remission) in group A and in group B. The OR is the ratio between two odds; the odds on an event in group A and the odds on the same event in group B. Both an OR and a RR amounting to 1 signify that there is no difference between the two treatments. Numbers needed to treat (NNT) were calculated, indicating the number of patients who would have to receive treatment A to produce one recovery from depression which would not have occurred with treatment B. The dropout rates were pooled by calculating the Relative Risk. all the data were analysed using the Review Manager 4.2 software from the Cochrane Collaboration. Dichotomous data (RR and OR) were analysed using the Mantel-Haenszel fixed effects model with 95% confidence intervals. Our analyses included a formal test of statistical heterogeneity. Statistical heterogeneity is the variability in the treatment effects in the different studies. It is a consequence of clinical and/or methodological diversity between these studies. Statistical heterogeneity occurs when the observed treatment effects are more varied than one would expect in a random situation. Significant heterogeneity suggests that the studies are not looking at the same quantity. The heterogeneity test we used was the natural approximate chi-square test; non-significant results (using p=0.10 as a limit) indicate a lack of evidence for heterogeneity across the studies. The test also describes the percentage of the variability in effect estimates (I²) due to heterogeneity rather than to sampling error. An I² of more than 50% indicates notable heterogeneity (Cochrane Reviewers Hand- book 8.7.2, Cochrane Collaboration, 2004). all analyses were also performed in subsamples consisting of studies broken down according to the chronicity and severity of the depression. First of all, a distinction was made between studies of chronic depression (the majority of patients in the study suffered from depression for at least 2 years) and studies of non-chronic depression. Secondly, the mean baseline scores on the 17-item Hamilton Depression Rating scale (HDRS, Hamilton, 1967) were used to distinguish between studies of mainly mild (12-19.9 points) moderate (20-24.9 points) and severe (25 points or more) depression.

3.3 Outcome of the literature search

Table 3.1 lists the seven studies that were included in the meta-analysis. The studies deal with a total of 903 patients, of whom 459 received psychotherapy and 444 combined therapy.

48 in in in

de- 7. ≤ 8.

≤

this

included. in

HDRS included included included included

8. HDRS chronic included.

HDRS

10. <

not not not not

and <

definition: included

BDI BDI HDRS Cognitive therapy plus 8.

definition:

outpatients

not ≤ definition: considered

outpatients

condition condition condition condition and

Only BDI

Only Remission Study

definition: definition: definition: definition: and

Remission

6. 9

Remission conditions

≤ review.

≤

review. review. review.

this

placebo Pharmacotherapy Pharmacotherapy Pharmacotherapy Pharmacotherapy Pharmacotherapy (32%) (21%) pression. (24%) (24%) Remission

(19%) HDRS (32%) Remission (22%) Remission (18%) (21%) (36%) HDRS (25%) (18%) review. (27%) (36%)

3/16 4/18 3/17 5/24 9/25 4/22 4/15 9/25

(48%) 48/227

(36%) 32/101 (32%) 55/228 (32%) 26/107

(63%) (45%) 12/38 (56%) (47%) (50%) (52%) (35%) 10/40 (59%) (27%) (32%) (%) (n/N) (%)

8/17 4/15 8/25

(n/N)

10/16 36/101 17,3 Remission 109/227 17/38 13/25 13/22

weeks/ weeks/ 10/18 weeks/ 12/24 weeks/ weeks/ 72/228 weeks/ 34/107

weeks/ 14/40

20 23,9 this 23 in 20 19,7 20 20,6 sessions HDRS (n=101) (n=16) (n=227) (n=38) (n=25) (n=22) (n=15)

(n=25) 12 (n=18) 12 (n=24) 12 (n=17) 20

(n=228) 16-20 26,9 this (n=107) 16 18,1

therapy therapy therapy therapy therapy therapy therapy

therapy therapy therapy behavioral- 12 behaviour 8 therapy

(n=40) 24 23 this system

Psychodynamic 16

Combined Combined Combined Combined Combined Combined tudies included in the meta-analysis. S

Cognitive Cognitive Short Cognitive

Combined Cognitive

Cognitive

et al.,

et and 2004 Psychotherapy 1996 therapy

Jonghe 1992 2000 analysis 1984

al., al., al.,

1981b

1981a

Hollon

Murphy al., et al., et Keller Beck al., Hautzinger

1985 Table 3.1: Study Treatments/N Duration/ Mean Remission Drop-out Comment Blackburn De

49 Chronic versus non-chronic depression We found six studies of non-chronic depression (Blackburn et al., 1981a and b; Murphy et al., 1984; Beck et al., 1985; Hollon et al., 1992; Hautzinger et al., 1996; De Jonghe et al., 2004) and one study of chronic depression (Keller et al., 2000). In the Keller study, 35% of the patients suffered from chronic major depression, 42% from MDD plus dysthymia, and 23% from recurrent depression without complete remission between episodes, which in our opinion means that all patients suffered from chronic depression. The mean duration of the last episode was 7.5 years.

Mild versus moderate depression We found six studies with 17-item HDRS mean baseline scores. The Keller study, however, used the 24-item HDRS version. We used an algorithm from the publication of O’Sullivan et al. (1997) to translate the 24-item scores into 17- item scores. The authors found a ratio of 1.25 between the 24-item and 17-item HDRS. We calculated that the mean baseline scores of 27 points in the Keller study corresponded to 22 (27/1.25) 17-item HDRS points. in all, we found three studies of average mild depression (Blackburn et al., 1981, Murphy et al., 1984, and De Jonghe et al., 2004) and four studies of average moderate depression (Beck et al., 1985, Hollon et al., 1992; Hautzinger et al., 1996; and Keller et al., 2000). We did not find suitable studies of severe depression.

3.4 Results of the meta-analysis

Dropout Figure 3.1 shows that the pooled dropout rates for combined therapy (25%) and for psychotherapy (24%) do not differ statistically (RR=1.03, p=0.77). The chi-square test of heterogeneity indicates a lack of evidence for heterogeneity (p=0.88 and I²=0%).

Efficacy Relative Risk of Remission (RR) It can be seen in figure 3.2 that the pooled remission rate for combined therapy (46%) is significantly higher than for psychotherapy (34%)(RR=1.32, p=0.0007). The chi-square test indicates a lack of evidence for heterogeneity (p=0.31 and I²=15.8%).

Odds Ratio for Remission (OR) Figure 3.3 shows the Odds Ratio for remission in combined therapy and psychotherapy. The OR is 1.59 (p=0.0007). The chi-square test indicates a lack of evidence for heterogeneity (p=0.24 and I²=24.6%).

Number Needed to Treat (NNT) Pooled data show that 8 patients would need to be treated with combined therapy to produce one recovery from depression that would not have occurred had they received psychotherapy alone (NNT = 8) (1/0.12).

50 95% CI 95% RR(fixed) 1.30] 1.23] 2.03] 2.09] 2.57] 4.00] 2.84] 4.52]

(fixed)

[0.82, [0.62, [0.84, [0.48, [0.62, [0.36, [0.27, [0.25, 95%

1.06 [0.25, 4.52] 1.20 [0.36, 4.00] 1.30 [0.84, 2.03] 1.03 [0.82, 1.30] 0.87 [0.27, 2.84] 1.00 [0.48, 2.09] 1.26 [0.62, 2.57] 0.88 [0.62, 1.23] RR 1.03

% % 8.17 1.00 8.84 1.26 3.30 1.20 4.34 0.87 2.64 1.06 49.80 0.88 22.91 1.30 2.64 3.30 22.91 00.00 100.00 Weight Weight 4.34 8.17 8.84 49.80 1 10 5 psychotherapy

Favours

Favours psychotherapy Favours CI

(fixed) 1

95%

RR 95% CI 95% RR(fixed) 0.5 therapy

55 0.2 combined

0.1

Favours =

I²

459 n/N 9/25 4/18 5/24 3/17 10/40 55/228 26/107 n/N 0.88),

Favours combined therapy combined Favours =

(Psychotherapy)

Psychotherapy 6

Psychotherapy 112

0.77)

therapy

2.42,

therapy),

= 444

n/N 9/25 4/15 4/22 0.29

48/227 =

Chi²

444 459 Combined n/N (Combined

effect:

112

1996 12/38 2004 32/101 CI)

1981 3/16

Combined therapy Combined

1984 4/15 4/18 12/38 10/40 32/101 26/107 3/16 3/17 4/22 5/24 9/25 9/25 48/227 55/228 overall heterogeneity:

1992

2000

1985 events: (95%

for for

Jonghe,

Test Test Total Total Keller, Hautzinger, De Hollon, Beck, Murphy, Blackburn,

Study Figure 3.1: Relative Risk of Dropout in Combined therapy and Psychotherapy.

51 Beck, 1985 1985 Beck, 1996 Hautzinger, 2004 Jonghe, De CI) (95% Total Blackburn, 1981 1981 Blackburn, Murphy, 1984 1984 Murphy, 1992 Hollon, 2000 Keller, (Psychotherapy) 112 therapy), (Combined 112 events: Total 0% = I² 0.88), = (P 6 df = 2.42, = Chi² heterogeneity: for Test 0.77) = (P 0.29 = Z effect: overall for Test Study Figure 3.1 Relative Risk of Dropout in Combined therapy and Psychotherapy , 1.56] 2 95% CI 95% RR(fixed) 1.56] 1.64] 3.22] 2.22] 1.92] 1.22] 2.01] 2.50] [0.19, 1.22]

8 CI

(fixed)

[1.12, [0.77, [0.82, [0.74, [1.20, [0.19, [0.70, [0.71, 95%

1.33 [0.71, 2.50] 0.4 1.52 [1.20, 1.92] 1.32 [1.1 1.18 [0.70, 2.01] 1.63 [0.82, 3.22] 1.28 [0.74, 2.22] 1.12 [0.77, 1.64] RR 1.32

% 5.01 5.87 46.40 00.00 % 5.17 1.63 8.81 1.28 5.87 0.48 7.41 1.18 5.01 1.33 Weight 7.41 5.17 8.81 21.33 1 21.33 1.12 46.40 1.52 100.00 therapy Weight

10 5 combined

Favours

Favours combined therapy combined Favours CI

(fixed) 1

95%

RR 95% CI 95%

RR(fixed) 0.5 57 0.2 psychotherapy

0.1

Favours

15.8%

I²

459 n/N Favours psychotherapy Favours 8/25 8/17 14/40 10/18 12/24 34/107 72/228 n/N 0.31),

(Psychotherapy)

Psychotherapy 6

Psychotherapy 158

0.0007)

therapy

7.12,

therapy),

= 444

n/N 4/15 3.38 13/25 13/22

109/227 =

Chi²

444 459 Combined n/N (Combined

effect:

202

1996 17/38 2004 36/101 CI)

1981 10/16

Combined therapy Combined 1984 overall heterogeneity:

1992 4/15 10/18 17/38 14/40 109/227 72/228 10/16 8/17 13/22 12/24 36/101 34/107 13/25 8/25

2000

1985 events: (95%

for for

Jonghe,

Test Test Total Total Hautzinger, Keller, De Hollon, Beck, Murphy, Blackburn, Study Figure 3.2: Relative Risk of Remission in Combined therapy and Psychotherapy.

52 Beck, 1985 1985 Beck, 1996 Hautzinger, 2000 Keller, CI) (95% Total Study 1981 Blackburn, Total events: 202 (Combined therapy), 158 (Psychotherapy) 158 therapy), (Combined 202 events: Total 15.8% = I² 0.31), = (P 6 df = 7.12, = Chi² heterogeneity: for Test 0.0007) = (P 3.38 = Z effect: overall for Test Murphy, 1984 1984 Murphy, 2004 Jonghe, De Hollon, 1992 1992 Hollon, Figure 3.2. Relative Risk of Remission in Combined therapy and Psychotherapy 2, 2.09] 2 95% CI 95% OR (fixed) [0.07, 1.27] 2.09] 2.11] 7.27] 3.74] 2.93] 1.27] 4.64] 7.53]

9 CI

(fixed)

[1.22, [0.67, [0.73, [0.60, [1.37, [0.07, [0.45, [0.47, 95%

1.88 [0.47, 7.53] 0.2 2.00 [1.37, 2.93] 1.59 [1. 1.44 [0.45, 4.64] 2.30 [0.73, 7.27] 1.50 [0.60, 3.74] 1.19 [0.67, 2.11] OR 1.59

3.45 7.91 44.33 00.00 % Weight 5.57 4.56 8.95 25.22 1 4.56 2.30 8.95 1.50 7.91 0.29 5.57 1.44 3.45 1.88 25.22 1.19 44.33 2.00 100.00 therapy Weight

10 5 combined

Favours Favours combined therapy combined Favours CI

(fixed) 1

95%

95% CI 95% OR

OR (fixed) 0.5 59 0.2 psychotherapy

0.1 Favours

24.6%

Favours psychotherapy Favours I²

459 n/N 8/25 8/17 14/40 10/18 12/24 34/107 72/228 n/N 0.24),

(Psychotherapy)

Psychotherapy 6

Psychotherapy 158

0.0007)

therapy

7.96,

therapy),

= 444

n/N 4/15 3.40 13/25 13/22

109/227 =

Chi²

Z 444 459

n/N (Combined

effect:

202

1996 17/38 2004 36/101 CI)

1981 10/16

Combined therapy Combined

1984 overall heterogeneity: 4/15 10/18 17/38 14/40 109/227 72/228 10/16 8/17 13/22 12/24 36/101 34/107 13/25 8/25

1992

2000

Combined 1985 events: (95%

for for

Jonghe,

Test Test Total Total Hautzinger, Keller, De Hollon, Beck, Murphy, Blackburn,

Study Figure 3.3: Odds Ratio of Remission Rates of Combined therapy and Psychotherapy.

53 Beck, 1985 1985 Beck, 1996 Hautzinger, 2000 Keller, CI) (95% Total Study 1981 Blackburn, Murphy, 1984 1984 Murphy, 2004 Jonghe, De Hollon, 1992 1992 Hollon, Total events: 202 (Combined therapy), 158 (Psychotherapy) 158 therapy), (Combined 202 events: Total 24.6% = I² 0.24), = (P 6 df = 7.96, = Chi² heterogeneity: for Test 0.0007) = (P 3.40 = Z effect: overall for Test Figure 3.3 Odds Ratio of Remission Rates Combined therapy and Psychotherapy Table 3.2: ITT remission rates in studies examining chronic¹ depression and studies examining non-chronic² depression. Psychotherapy Combined therapy Significance Chronic depression 32% 48% p<0.001

Non-chronic depression 37% 43% RR=1.15, p=0.22

Significance RR=0.85, p=0.20 RR=1.12, p=0.28 ¹ Keller et al., 2000. ² Blackburn et al., 1981, Murphy et al., 1984, Beck et al., 1985, Hollon et al., 1992, Hautzinger et al., 1996, De Jonghe et al., 2004.

Table 3.3: ITT remission rates in studies examining mild¹ depression and studies examining moderate² depression. Psychotherapy Combined therapy Significance Mild depression 37% 42% RR=1.17, p=0.29

Moderate depression 34% 47% RR=1.39, p=0.001

Significance RR=0.93, p=0.57 RR=1.10, p=0.39 ¹ Blackburn et al., 1981, Murphy et al., 1984, De Jonghe et al., 2004. ² Beck et al., 1985, Hautzinger et al., 1996, Hollon et al., 1992, Keller et al., 2000.

Sub-analyses of chronicity Table 3.2 shows separately the pooled remission rates and RR for the six studies of non-chronic depression and the remission rates in the single study of chronic depression. in non-chronic depression, the pooled remission rates of psychotherapy and combined therapy do not differ significantly (37% and 43% respectively, p=0.22). The chi-square test indicates a lack of evidence for heterogeneity (p=0.45 and I²=0%). in chronic depression (Keller et al., 2000), remission rates of psychotherapy and combined therapy differ significantly (32% and 48% respectively, p<0.001). the efficacy of combined therapy does not differ significantly for non- chronic and chronic depression (43% compared to 48%, p=0.28). Similarly, the efficacy of psychotherapy does not differ significantly in non-chronic and chronic depression (37% compared to 32%, p=0.20). In the last two analyses, heterogeneity was not an issue since we made only one comparison between two groups of studies.

54 Table 3.4: ITT remission rates associated with chronicity and severity. Non-chronic Chronic Total Mild CT (42%) vs Pt (37%) No data CT (42%) vs Pt (37%) p=0.29 p=0.29 Moderate CT (44%) vs Pt (39%) CT (48%) vs Pt (32%) CT(47%) vs Pt (34%) p=0.52 p<0.001 p=0.001 Severe No data No data No data Total CT (43%) vs Pt (37%) CT (48%) vs Pt (32%) CT (46%) vs Pt (34%) p=0.22 p<0.001 p=0.0007

Non-chronic Chronic Total Mild De Jonghe et al. n=208 No data De Jonghe et al. n=208 Blackburn et al. n=33 Blackburn et al. n=33 Murphy et al. n=46 Murphy et al. n=46 Moderate Beck et al. n=33 Keller et al.n=455 Beck et al. n=33 Hollon et al. n=50 Hollon et al. n=50 Hautzinger et al. n=78 Keller et al. n=455 Hautzinger et al. n=78 Severe No data No data No data Total De Jonghe et al. n=208 Keller et al. n=455 De Jonghe et al. n=208 Blackburn et al. n=33 Blackburn et al. n=33 Murphy et al. n=46 Murphy et al. n=46 Beck et al. n=33 Beck et al. n=33 Hollon et al. n=50 Hollon et al. n=50 Hautzinger et al. n=78 Keller et al. n=455 Hautzinger et al. n=78

Sub-analyses of severity Table 3.3 shows the remission rates and RR for three studies of mild depression and four studies of moderate depression. in mild depression, the pooled remission rates of psychotherapy and combined therapy do not differ significantly (37% and 42% respectively, p=0.29). The chi-square test indicates a lack of evidence for heterogeneity (p=0.90 and I²=0%). in moderate depression the pooled remission rates of psychotherapy and combined therapy differ significantly (34% and 47% respectively, p=0.001). The chi-square test for heterogeneity borders on significance (p=0.13 and I²=46.7%). This is probably due to the outlying results from the study of Beck et al. (1985).

55 the efficacy of combined therapy does not differ significantly for mild and moderate depression (42% compared to 47%, p=0.39). Similarly, the efficacy of psychotherapy does not differ significantly for mild and moderate depression (37% compared to 34%, p=0.57). In the last two analyses, heterogeneity was not an issue since we made only one comparison between two groups of studies.

Sub-analyses of chronicity and severity Table 3.4 shows the remission rates for all comparisons, including the separate remission rates for non-chronic mild depression and non-chronic moderate depression. No data were available for mild chronic depression. in moderate depression, combined therapy outperforms psychotherapy. However, there is no difference in remission for moderate non-chronic depression (44% for combined therapy and 39% for psychotherapy, p=0.52). The chi-square test in this analysis indicates moderate heterogeneity (p=0.11 and I²=55.5%). This is probably due to the outlying results from the study of Beck et al. (1985).

3.5 Discussion

Existing reviews and meta-analyses comparing psychotherapy and combined therapy in the treatment of major depression are often characterised by the clinical heterogeneity of the studies included. This may partially explain why the results are contradicting. In this meta-analysis, we furthered the homogeneity of the studies included by applying strict clinical inclusion and exclusion criteria. We performed statistical tests a posteriori which supported our belief that the studies included were indeed sufficiently homogeneous. In addition, we took into account two potential determinants of treatment prognosis by performing sub-analyses of the chronicity and severity of depression. a combination of psychotherapy and pharmacotherapy appeared to be more efficacious than psychotherapy alone at treatment termination. The difference – 12% in remission rate – is both statistically and clinically significant. This means that our findings are in line with the results of Friedman et al. (2004), Jindal & Thase (2003) and Segal et al. (2002). however, our results must be read in the light of the chronicity and severity of depression. Chronicity and severity both seem to influence the relative efficacy of the two treatments. The superiority of combined therapy was not demonstrated for non-chronic depression or mild depression, which is in line with the main findings of our own trial (de Jonghe et al., 2004) considering for the most part mild, non-chronic depression. In moderate versus mild depression, our results are corroborated by the study of Thase et al. (1997). The authors reported a statistically significant and clinically relevant difference in favour of combined therapy in a ‘more severe’ (HDRS > 19 points) subgroup, but found no differences in the ‘less severe’(HDRS<20) subgroup. However, they did not break down the more severe subgroup according to chronicity. Our results indicate that combined therapy yields better results in moderate chronic

56 depression but not in moderate non-chronic depression. A tentative methodological explanation for this finding may be that the variability in outcome is decreased in studies on chronic depression (due to the depression becoming more chronic), as a result of which small differences in benefit of therapies would become more significant. finally, it must be added that psychotherapy never achieved better results than combined therapy. turning to our secondary research question, we found that the dropout rates for combined therapy were comparable to those for psychotherapy. In all but one study, however, the dropout in combined therapy was not specified as pharmacotherapy and/or psychotherapy dropout, and this precludes a more specific interpretation of these findings. Our review has several limitations. It only compares psychotherapy with combined therapy, and does not include comparisons with pharmacotherapy. That topic has been dealt with in other publications (for example, Pampallona et al., 2004, de Jonghe et al., 2001, de Maat et al., 2006). We only address the Pampallona study here. In the meta-analysis of Pampallona et al. (2004) sixteen trials (932 patients) were included comparing combined therapy with pharmacotherapy alone. Pooled remission rate was 33% in pharmacotherapy and 44% in combined therapy. The authors reported a statistically significant odds ratio of 1.86 in favour of combined therapy, raising to 2.21 in studies longer than 12 weeks. We found only seven studies that were suitable for our meta-analysis and their methodological qualities vary. Some of them present small patient numbers, therefore limiting statistical power. This means that some of our own sub-analyses involve small sample sizes and caution is required when making interpretations. There was no strict monitoring of medication compliance in any of the studies, except that of Keller et al. The two most recent studies are the only ones that comply satisfactorily with actual research criteria. Allegiance effects (Gaffan et al., 1995) cannot be excluded. With regard to chronic depression, the evidence is limited to only one study; e.g. Keller et al. (2000). This study considers a large sample of patients and is clearly an important paper in this area. However it is highly recommendable that it be replicated in order to further the evidence on the treatment of chronic depression. efficacy was measured with the HDRS only. Hamilton designed this observer-rated scale for the assessment of symptomatic improvement in severely depressed inpatients treated with antidepressants. It may ignore important aspects of depression. The studies we found did not assess social functioning or quality of life, which are after all the ultimate goal of therapy. Although we included only studies with formal psychotherapy and pharmacotherapy in our meta-analysis, it could be argued that there was still some heterogeneity in the treatments included. The de Jonghe study, regarding 208 patients, examined a time limited psychodynamic psychotherapy. The Keller study (455 patients) considered an integrative psychotherapy with an emphasis on psychodynamic principles. The other studies applied cognitive behaviour therapies. This means that 73% of the patients included in this meta-analysis received a more or less psychoanalytic oriented psychotherapy. It may therefore be argued that our con-

57 clusions do not apply to psychotherapy in general, but mainly to psychodynamic therapy. There were also differences between studies in pharmacotherapies. However, this factor may be of only limited importance. Meta-analytic studies found no significant differences for the treatment of depressed outpatients between TCAs and SSRIs (e.g. Anderson, 2000) or between psychodynamic and cognitive-behavioural therapies (Leichsenring, 2001). finally, conclusions based on the results of only RCTs suffer from limitations. An important one is that RCTs contain systematic biases in patient selection, since they exclude patients with serious comorbidity such as drug dependence, suicide intentions or severe personality disorders. Keitner et al. (2003), e.g., mention that only 14.5% of eligible depressed patients eventually took part in an RCT. another important limitation is the fact that we based our conclusions about the severity of depression on mean baseline scores from the studies and not on individual patient data. We are aware that this is a rather coarse way to break down a spectrum. Nevertheless, our results do not seem to diverge from findings based on individual patients. Hollon et al. (1992) and Hautzinger et al. (1999) performed sub-analyses of severity. They also found no differences between combined therapy and psychotherapy in mild and moderate non- chronic depression. Our HDRS cut-off scores for the distinction between mild, moderate and severe depression correspond to those usually found in the literature. Unfortunately, Hamilton himself did not define cut-off scores for his scale. The result is that there is no generally accepted definition of mild, moderate and severe depression. In clinical practice, mild depression and moderate depression are thought to range respectively from 12-14 to 18-20, and from 18-20 to 24-26 17-item HDRS points. We chose a cut-off score of 20 in accordance with the NIMH study (Elkin et al., 1989) and other authors (e.g. Thase et al.1997, Hollon et al., 1992). to end our review on a sombre note, we cannot but confirm the fact known to clinicians and researchers alike that the efficacy of time-limited treatments, be they combined therapy or psychotherapy, are modest, with an average of less than half of treated patients achieving remission.

3.6 Conclusions

In the ongoing debate about whether combined therapy is more efficacious than psychotherapy (leaving aside the pharmacotherapy option), the results of our study support the view that combined therapy yields better results than psychotherapy alone as far as the time-limited treatment of outpatients with major depression is concerned. However, severity and chronicity seem to be important modifiers in the efficacy of both treatments. Although combined therapy consistently achieved better results than psychotherapy alone, the superiority of the first treatment is in fact only demonstrated in moderate, chronic depression.

58 References

American Psychiatric Association (1980). Diagnostic and Statistical Manual of Mental Disorders (3rd edn) (DSM-III). Washington DC: APA. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders. Fourth edition (DSM-IV). Washington: APA. Anderson, I.M. (2000). Selective serotonin reuptake inhibitors versus tricyclic antidepressants: a meta-analysis of efficacy and tolerability. Journal of Affective Disorders, 58, 19-36. Beck, A.T., Hollon, S.D., Young, J.E., Bedrosian, R.C., Budenz, D. (1985). Treatment of Depression with Cognitive Therapy and Amitriptyline. Archives of General Psychia- try, 42, 142-148. Bellack, A.S., Hersen, M., & Himmelhoch, J. (1981). Social skills training compared with pharmacotherapy and psychotherapy in the treatment of unipolar depression. American Journal of Psychiatry, 138, 1562-1567. Beutler, L.E., Scogin, F., Kirkish et al. (1987). Group cognitive therapy and alprazolam in the treatment of depression in older adults. Journal of Consulting and Clinical Psycholoy, 55, 550-556. Blackburn, I.M., Bishop, S., Glen, A.I.M., Whalley, L.J., Christie, J.E. (1981a). The efficacy of cognitive therapy in depression: A treatment trial using cognitive therapy and pharmacotherapy, each alone and in combination. British Journal of Psychiatry, 139, 181-189. Blackburn, I.M., Bishop, S. (1981b). Is there an alternative to drugs in the treatment of depressed ambulatory patients? Behavioural Psychotherapy, 9, 96-104. Blackburn, I.M., Eunson, K.M., Bishop, S. (1986). A two-year naturalistic follow-up of depressed patients treated with cognitive therapy, pharmacotherapy and a combination of both. Journal of Affective disorders, 10, 67-75. Blom, M.B.J., Jonker, K, Haffmans, P.M.J., Spinhoven, Ph. (2000). Combinatie van medicatie. en psychotherapie bij depressie: een overzicht. Tijdschrift voor Psychi- atrie, 42, 99-407. Cochrane Collaboration, Cochrane Reviewers Handbook. Available at: http://cochrane. com. Accessed May, 2004. Cohen, J. (1988). Statistical power analysis for the behavioural sciences (2nd ed.). Hillsdale, NJ: Lawrence Earlbaum Asssociates. Conte, H.R., Plutchik, R., Wild, K., Karasu, T.B. (1986). Combined psychotherapy and pharmacotherapy for depression. Archives of General Psychiatry, 43, 471-479. Evans, M.D., Hollon, S.D., DeRubeis, R.J., Piasecki, J.M., Grove, W.M., Garvey, M.J., Tuason, V.B. (1992). Differential relapse following cognitive therapy and pharmacotherapy for depression. Archives of General Psychiatry, 49, 802-808. Frank, E., Kupfer, D.J., Perel, J.M., Cornes, C., Jarret, D.B., Malinger, A.G., Thase, M.E., McEachran, A.B., Grocinski, V.J. (1990). Three-year outcomes for maintenance therapies in recurrent depression. Archives of General Psychiatry, 47, 1093-1099. Friedman, M.A., Detweiler-Bedell, J.B., Leventhal, H.E., Horne, R,m Keitner, G.I., Miller, I.W. (2004). Combined psychotherapy and pharmacotherapy for the treatment of major depressive disorder. Clinical Psychology; Science and Practice, 11, 47-68. Gaffan, E.A., Tsaousis, I., Kemp-Wheeler, S.M. (1995). Researcher allegiance and meta- analysis: the case of cognitive therapy for depression. Journal of Consulting and Clinical Psychology, 6, 966-980. Hamilton, M. (1967). Development of a rating scale for primary depressive illness. Brit- ish Journal of Social and Clinical Psychology, 6, 278-296. Hautzinger, M., de Jong-Meyer, R., Treiber, R.,Rudolf, G.A.E., Thien, U. (1996). Wirk- samkeit Kognitiever Verhaltenstherapie, Pharmakotherapie und deren Kombination bei nicht-endogenen, unipolaren Depressionen. Zeitschrift für Klinische Psycholo- gie, 25(2): 130-145.

59 Hersen, M., Bellack, A.S., Himmelhoch, J.M., et al. (1984). Effects of social skills training, amitriptyline and psychotherapy in unipolar depressed women. Behavioral Therapy, 15, 21-40. Hollon, S.D., DeRubeis, R.J., Evans, M.D., Wiemer, M.J., Garvey, M.J., Grove, W.M., Tuason, V.B. (1992). Cognitive therapy and Pharmacotherapy for Depression. Singly and in combination. Archives of General Psychiatry, vol. 49, 774-781. Jarrett, R.B. (1995). Comparing and combining short-term psychotherapy and pharmacotherapy for depression. In E.E. Beckman & W.R. Leber (red.), Handbook of depression, New York: Guilford Press. Jindal, R.D., Thase, M.E. (2003). Integrating psychotherapy and pharmacotherapy to improve outcomes among patients with mood disorders. Psychiatry Services, 54 (11), 1484-1490. Jonghe, F., Kool, S., van Aalst, G., Dekker, J., Peen, J. (2001). Combining psychotherapy and antidepressants in the treatment of depression. Journal of Affective disorders, 64, 217-229. Jonghe, F., de, Hendriksen, M., Aalst, G., Kool, S., Peen, J., Van, R., Eijnden, van den, E., Dekker, J. (2004). Psychotherapy and Combined Therapy (pharmacotherapy plus psychotherapy) in the treatment of depression. British Journal of Psychiatry, 185 (1), 37-45. Keitner, G.I., Posternak, M.A., Ryan, C.E. (2003). How many subjects with major depressive disorder meet eligibility requirements of an antidepressant efficacy trial? Jour- nal of Clinical psychiatry, 59, 598-607. Keller,M.B., McCullough, J.P., Klein D.N., Arnow, B., Dunner, D.L., Geleberg, A.J., Markowitz, J.C., Nemeroff, C.B., Russel, J.M., Thase, M.E., Treivedi, M.H., Zajecka, J. (2000). A comparison of Nefazodone, the Cognitive Behavioral-Analysis System of Psychotherapy and their combination for the treatment of chronic depression. The New England Journal of Medicine, 342, 1462-1470. Leichsenring, F. (2001) Comparative effects of short-term psychodynamic psychotherapy and cognitive-behavioral therapy in depression: a meta-analytical approach. Clinical psychology Review, 21 (3), 401-419. Maat, S. de, Dekker, J., Schoevers, R., Jonghe, F., de (2006). Relative efficacy of psychotherapy and pharmacotherapy in the treatment of depression; a meta-analysis. Psychotherapy Research, 16(5): 562-572. Murphy, G.E., Simons, A.D., Wetzel, R.D., Lustman, P.J. (1984). Cognitive therapy and pharmacotherapy. Archives of General Psychiatry, 41, 33-41. Mynor-Wallis, L.M., Gath, D.H., Day, A. et al. (2000). Randomized controlled trial of problem-solving treatment, anti-depressant medication, and combined treatment for major depression in primary care. British Medical Journal, 320, 26-30. O’Sullivan, R.L., Fava, M., Agustin, C., et al. (1997). Sensitivity of the six-item Hamilton Depression Rating Scale. Acta Psychiatrica Scandinavcia, 95, 379-384. Pampallona, S., Bollini, P., Tibaldi, G., Kupelnick, B., Munizza, C. (2004). Combined pharmacotherapy and psychological treatment for depression. A systematic review. Archives of General psychiatry; 61: 714-719. Reynolds, C.F., Miller, M.D., Pasternak, R.E., et al. (1999). Treatment of bereavement related major depressive episodes in later life: A controlled study of acute and continuation treatment with nortryptiline and interpersonal psychotherapy. American Journal of Psychiatry, 156, 202-208. Robinson, L.A., Berman, J.S., Neimeyer, R.A. (1990) Psychotherapy for the treatment of depression: a comprehensive review of controlled outcome research. Psychological Bulletin, 108: 30-49. Roth, D., Bielski, R., Jones, M., et al. (1982). A comparison of self-control therapy and combined self-control therapy and antidepressant medication in the treatment of depression. Behavioural Psychotherapy, 13, 133-144. Rush, A.J., & Watkins, J.T. (1981). Group versus individual therapy: A pilot study. Cog- nitive Therapy Research, 5, 95-103.

60 Scott, C., & Stradling, S.G. (1990). Group cognitive therapy for depression produces clinically significant reliable change in community based settings. Behavioural Psy- chotherapy, 18, 1-19. Segal, Z, Vincent, P., Levitt, A. (2002). Efficacy of combined, sequential and crossover psychotherapy and pharmacotherapy in improving outcomes in depression. Journal of Psychiatry Neurosci, 27 (4), 281-290. Simons, A.D., Murphy, G.E., Levine, J.L., Wetzel, R.D.l (1986). Cognitive therapy and pharmacotherapy for depression. Sustained improvement over one year. Archives of General Psychiatry, 43, 43-48. Spijker, J., Graaf, de, J., Bijl, R.V., Beekman, A.T.F., Ormel, J., Nolen, W. (2002). Dura- tion of major depressive episodes in the general population: results from the Neth- erlands Mental Health Survey and Incidence Study (NEMESIS). British Journal of Psychiatry, 181, 208-213. Spitzer, R.L., Endicott, J., Robins, E. (1979). Research Diagnostic Criteria: Rationale and Reliability. Archives of General Psychiatry, 35, 773-785. Teasdale, T.D., Fennel, M.J.V., Hibbert, G.A., et al. (1984). Cognitive therapy for major depressive disorder in primary care. British Journal of Psychiatry, 144, 400-406. Thase, M.E., Greenhouse, J.B., Frank, E., Reynolds III, C.F., Pilkonis, P.A., Hurley, K., Grochocinsky, V., Kupfer, D.J. (1997). Treatment of major depression with psychotherapy or psychotherapy-pharmacotherapy combinations. Archives of General Psychiatry, 54, 1009-1015. Wexler, B.E., Cicchetti, D.V. (1992). The outpatient treatment of depression. Implica- tions of outcome research for clinical practice. The Journal of Nervous and Mental disease, 180, 277-286.

61 Appendix

Search strategy A systematic search for Randomized Controlled Trials (RCTs) was performed in MEDLINE, EMBASE, the Cochrane Controlled Trials Register, the Cochrane Database of Reviews and Protocols, and PsychInfo. The search headings were: DEPRESSION, MAJOR DEPRESSIVE DISORDER, PSYCHOTHERAPY, COM- BINED THERAPY, PHARMACOTHERAPY, ANTIDEPRESSANTS. Limits were: (randomized controlled trial[Publication Type] OR controlled clinical trial[Publication Type] OR randomized controlled trials OR random allocation OR double-blind method OR single-blind method OR clinical trial[Publication Type] OR clinical trials OR (clinical AND trial*) OR ((singl* OR doubl* OR trebl* OR tripl*) AND (blind* OR mask*)) OR placebos OR placebo* OR random* OR research design OR comparative study OR evaluation studies OR follow up studies OR prospective studies OR control OR controlled OR prospective* OR volunteer*) NOT (Animal[MESH] NOT (Human[MESH] AND Animal[MESH])) and a time period of 1980 (the year DSM-III was published) to 2005. Titles and abstracts were screened. References in the articles retrieved were searched. Book chapters about the treatment of depression were retrieved. No special efforts were made to discover unpublished data. in order to obtain a relatively homogeneous sample in clinical terms, the following selection criteria were applied. The studies had to compare psychotherapy with a combination of psychotherapy and pharmacotherapy (combined therapy) and focus on the efficacy of acute treatment (no maintenance treatment or sequential treatment). The study sample had to consist of psychiatric outpatients (no primary care patients or inpatients) aged between 19 and 65 years (no geriatric patients, no children), diagnosed with unipolar major depression according to DSM-III-(R) (APA, 1980), DSM-IV-(R) (APA, 1994) or Research Diagnostic Criteria (Spitzer et al., 1978). Treatment protocols in the studies had to involve formal (in other words, using behavioural, cognitive, psychodynamic or client-centred theories and techniques) time-limited (maximum 6 months) individual psychotherapy and adequate treatment with regular antidepressants. By ‘adequate treatment’ we mean an adequate dose (different per antidepressant) administered during an adequate time period (at least 4 weeks) by a registered clinician. A regular antidepressant is approved as such by national authorities. Method sections were checked for the specifics of the treatment regime, but no efforts were made to obtain additional information. The studies had to report remission rates and dropout rates. Methodological quality was judged using the four criteria of the Cochrane Collaboration. Firstly, the studies had to have a randomized design to minimise selection bias. Secondly, apart from the treatments, the two study groups had to have been treated equally to minimise performance bias. Thirdly, the study had to report on selective dropout in the treatment conditions (e.g. have ITT analyses or specify differences in dropout) and, fourthly, detection bias had to be minimised by the blind assessment of outcomes. Two reviewers, who needed to agree on all criteria in order to include a study, judged all selection criteria independently. No studies were excluded as a result of disagreement between the reviewers.

62 Chapter 4 Short-term Psychoanalytic Supportive Psychotherapy for depressed patients

Frans de Jonghe, Saskia de Maat, Rien Van, Marielle Hendriksen, Simone Kool, Gerda van Aalst, Robert Schoevers, Jack Dekker Submitted

Abstract

Short-term Psychoanalytic Supportive Psychotherapy (SPSP) is a face-to-face, individual psychotherapy, consisting of sixteen sessions in six months (first eight weekly, then eight fortnightly sessions). It is rooted in psychoanalytic theory. Its primary aim is to cure depression. To reduce patient’s vulnerability to depression is a secondary goal. The emphasis is on supportive techniques that counter regression and foster psychological growth. The putative process consists in experiencing ‘relational dissonance’, i.e., feeling two contradictory relationships in the therapeutic situation simultaneously. One is determined by the past, the other by the present. Mostly unconsciously, the patient experiences adequate gratification of his unmet early infantile needs, as they manifest themselves in the archaic aspects of the therapeutic relationship, which is assumed to be an important curative factor. SPSP unfolds as a discourse in which nine levels can be distinguished. Each regards a specific subject, which is the focus of the interaction between patient and therapist at that level. The results of a mega-analysis pooling the original data of three randomized clinical trials suggest that SPSP and pharmacotherapy are equally efficacious, in treating outpatients with mild to moderate major depressive disorder. Combined they are as efficacious as SPSP alone and more efficacious than pharmacotherapy alone. Therefore, we consider Short-term Psychoanalytic Supportive Psycho- therapy (SPSP) a valuable extension to the existing options for the treatment of depressed patients.

4.1 Introduction

Several options, both pharmacological and psychological, are available for the treatment of depressed patients. Their effectiveness is undeniable but unfortunately a considerable amount of patients do not (fully) respond. In addition, as depressed patients constitute a heterogeneous group, it seems probable that different patients need different treatments. Therefore, the search for new alternatives that could potentially benefit patients with poor outcome of other-

63 wise effective therapies, is useful (Fava, 2006). We think Short-term Psycho- analytic Supportive Psychotherapy (SPSP) could constitute a valuable extension to the existing options. SPSP is a short-term cure tailored to the conditions of depressed patients. It combines a psychoanalytic with supportive stance. We are not aware of any other psychoanalytic therapy presenting these characteristics. Firsty, we summarize the theoretical roots of SPSP in this article. Secondly, we portray its principal characteristics. Thirdly, we try to elucidate its position among other therapies and finally we report on the efficacy studies providing empirical validation of its efficacy.

4.2 Theoretical roots of SPSP

SPSP is rooted in Freud’s drive theory (Freud, 1957) and Ego-psychology (Freud, 1961). In addition, we value theories that focus on ‘developmental needs’, i.e., innate, basic, social needs, which must be met adequately in early infancy in order to allow the first stages of growth to unfold. Prominent among them are the need to engage in relationships, to be loved, to be protected and to be esteemed, brought to the fore by Klein’s object-relations theory (Klein, 1975), Balint’s primary love theory (Balint, 1952), Bowlby’s attachment theory (Bowlby, 1969) and Kohut’s self-psychology (Kohut, 1971; Kohut, 1977), respectively. We consider the vicissitudes of these developmental needs particularly relevant for the treatment of depressed people. In short, we conceptualise patients’ problems as consisting of both conflict and developmental pathology. We look at both types of pathology from a relational perspective. We distinguish between interactions with others present in the outer world (‘external-interpersonal’), with others only present in the inner world (‘internal-interpersonal’) and between I and Me. The latter we call the ‘IntraPersonal Relationship’ (IPR). Second and third type relationships are ‘internal relationships’. Internal-interpersonal relationships are seen as internalised external-interpersonal relationships. The IPR results from the subject identifying with (some aspects) of (some) internal-interpersonal relationships. Carl’s case may help elucidate these concepts. He thinks that the relation with his mother, deceased years ago has gradually evaporated. She does not play a role in his life any longer. Much to his surprise, however, he dreamt about her last night. She was being reproachful, about what is unclear. Apparently she is part of his past that lives on in his present as an internal- interpersonal relationship. In therapy, it gradually occurs to him, that he as an adult is no less disappointed in himself than she was in him when he was a child. He has come to view himself, in his perception that is, as he used to be seen by her. It has become a lasting, unconscious aspect of his IPR. He did not seek therapy for his malignant IPR itself but for its natural, logical consequences: he desperately craves recognition and consequently “buys” appreciation with pleasing behaviour; he is exceedingly naïve in contacts with sycophants: he fears and expects rejection and consequently shuns contacts and withdraws into grandiosity; he is exceedingly suspicious of his loving partner and behaves in a self-defeating way, making his darkest prophecies come true. The consequence

64 is that he feels chronically rejected, lonely, inferior, depressed, and that is what he is asking help for. If inadequately met in early infancy, developmental needs persist in adults as ongoing, malignant, early infantile aspects of internal (interpersonal and intrapersonal) relationships. They manifest themselves in the archaic aspects of relationships, where they act as moulds on potentially new relationships, thus stimulating repetition instead of growth. We suggest the therapeutic action of SPSP resides mainly in its power to evoke, in the patient, an experience of relational dissonance or friction between two contradictory, external-interpersonal relationships, simultaneously felt in the present. Simplifying matters, we hereafter sketch what may happen in Carl’s therapy. On one hand, Carl is driven to externalise his internal, rather malignant relationships, especially his IPR. He projects them on to and into the therapist and experiences him as a rather debasing person. That relationship is dominated by Carl’s past. In that sense it is ‘old’ and ‘unrealistic’. On the other hand, due to the interaction between the therapist’s genuine behaviour and Carl’s capacities to test reality, he experiences the therapist as a rather valuing person. That relationship is dominated by actual reality. In this sense, it is ‘new’ and ‘realistic’. Both relationships regard the same developmental need, to be esteemed, manifesting itself in the therapeutic relationship. The new relationship must be new enough but not too new. If it bears too much novelty there is no dissonance. The two feeling states pass each other without any wringing, jarring contact, hence without mutual influence. Dissonance is not a persisting phenomenon. It tends to evolve towards consonance. In the end, either the new or the old relationship will prevail. If the new, benevolent relationship is not new enough, the old relationship will prevail. Carl’s malevolent internal relationships are strengthened and his old feelings confirmed. Retraumatiza- tion is the result and the process is anti -therapeutic. On the other hand, if the new, benevolent relationship differs neither too little nor too much from the old one, it may prevail. Carl will experience adequate gratification of his unmet early-infantile need. This gratification forms the core of ‘psychoanalytic informed support’, in our view the most important curative factor in SPSP. In so far Carl internalises the new relationship and identifies with it, the old, malevolent, internal relationships (the old templates) will be corrected by it. Thus personality is structured or restructured. At its deepest level, this process is neither reflexive, explicit, verbal, symbolic, declarative nor repressed. It unfolds at the ‘procedural knowledge level’. It is unconscious but not dynamically so. It nevertheless may result in emotional insight, as Carl, feeling adequately supported, may abandon his resistances and defences and shift to self- exploration.

4.3 Principal characteristics of SPSP

SPSP is a face-to-face, six-month, individual psychotherapy consisting of sixteen sessions (first eight weekly, then eight fortnightly sessions). Prior to the start of treatment, the therapeutic frame is discussed and settled by agreement.

65 The therapist will stick to the arrangements firmly but not rigidly. He bears in mind and regularly discusses with the patient that the therapeutic encounter will last only a limited period of time. SPSP’s primary goal is to cure depression. The secondary goal is to reduce patient’s vulnerability to depression. We con- ceive the latter as the shaping or altering of the internal relationships, especially the IPR. Taking into account SPSP’s restricted number of sessions, it is all too obvious that personality change will be limited. Setting, frame and contract are ingredients of the supportive approach. They are supposed to evoke peace and quiet, reliability, predictability, carefulness, professionalism, clarity, trans- parency, unhurried activity and realistic optimism. The therapist’s attitude is expected to be basically interested, accepting, affirmative, empathic, concerned, helpful, patient, actively expectant, perseverant, honest, transparent and flexible. The therapist has to adapt to the personality and actual state of mind of the patient. He certainly has to reckon with the role enactment he is invited to by the patient’s projective identification. The therapist is supposed to promote, maintain and, if necessary, restore a supportive therapeutic relationship, to the best of his abilities. We understand the therapeutic relationship to be an interaction, a bilateral phenomenon constituted by the active roles of both contribu- tors. However, this ‘two-person approach’ does not downplay the processes taking place within patient and therapist. In that sense, it includes a ‘one-person approach’ to both participants. The therapist will have to be flexible, adapting his technique to the temporary and structural level of his patient’s functioning. Admittedly, up to now the supportive technique has been less well articulated than the interpretive one. Important aspects of it are containing (Bion, 1962), holding (Winnicott, 1965), mirroring (Kohut, 1977), developmental help (Fonagy and Target, 1996) and the judicious blending of homeostatic (‘maternal’) and disruptive (‘paternal’) attunement. Supportive interventions can consist of practical action, e.g. making a telephone call on behalf of the patient. We exercise restraint in these matters though, because this type of support easily furthers regression. ‘Helping the patient to help himself’ remains the leading idea. To this purpose verbal interventions mostly suffice. We mention some examples: showing interest by providing and asking information, expressing understanding and empathy, expressing acceptance and esteem, reducing guilt feelings, shame and isolation, reassuring, normalising, instilling hope, motivat- ing, facilitating verbalisation, exploring by listening actively and by asking open and closed questions, encouraging to dwell upon feelings, furthering acknowl- edgment and acceptation of painful or frightening affects, fostering controlled abreaction (catharsis) while discouraging all too heavy affects, evoking new and advisable feelings or furthering more adequate coping with feelings, clarifying, confronting, suggesting new ideas or possibilities, if necessary not shunning to give advice, if it needs be persuading, countering dysfunctional cognitions and behaviours, and encouraging autonomy. The aforementioned interventions are not adequate or inadequate in themselves. It is their (non) integrated and (non) judicious application and the patient’s perception of it that determine whether they are adequate or not.

66 4.4 The typical course

SPSP presents important nonverbal communicative aspects but is essentially a ‘talking cure’. It unfolds as a discourse in which we distinguish nine levels. We illustrate our views with vignettes regarding the SPSP treatment of Catherine, a 28-year-old head nurse. She is a psychiatric outpatient suffering from DSM- IV defined, moderate major depressive disorder. The case has been modified for discretion’s sake; still we trust the vignettes convey the essence of the message. 4.1. At the first level the focus is on the patient’s physical and psychological complaints and symptoms. For at least ten months, Catherine’s life has been miserable. Most importantly, she is tired all the time for no good reason. She awakes exhausted and drags herself through the day (she still works at the hospital). Besides that, her mood is low, her interest has vanished and nothing in life is pleasurable anymore. She cries a lot without even knowing why: ‘Isn’t that silly?’ Catherine understands there is something wrong with her. In this sense she gives evidence of some insight. The therapist asks her how she explains her depressed state herself. There is no explanation. ‘I may have a creeping physical illness the doctors can’t find.’ She is here on the advice of her friend but she doesn’t believe in it. The therapist registers her demand to be rescued. He knows he is unable to do it. He feels powerless, which he conceives as a token of her own powerlessness. He listens with an attentive ear and shows empathy, not pity. He is careful, not worried. He considers the situation serious, not hopeless. He provides some psycho-education on depression, discusses how to cope with her symptoms but refrains from giving advice beyond very simple, day-to-day issues. Meanwhile, he cautiously inquires after suicidality, which in Catherine’s case is not a matter for concern. Apart from that, he simply bides his time. 4.2. At the second level the focus shifts from symptoms and complaints to life circumstances, which somehow might influence or have influenced the depression. It is a first step in looking for an explanation. The therapist enquires about Catherine’s actual life situation and maps out its taxing and supporting aspects. He asks whether Catherine’s mood reacts to circumstances. Indeed, it does. Reluctantly, she tells that she went on a holiday with her boyfriend the other day and had indeed felt better then, only to sink back as soon as she resumed work. She had also felt very depressed when recently visiting a rather busy restaurant, but walking her dog had lightened her up. The therapist comments that it is a positive fact that her depression still is sensitive to circumstances. What happened in her life about a year ago? The hospital ward had to move temporarily. The patients were upset, there was a lot to do and she worked even harder than she used to. ‘But somehow I was not “there” at work, it was as if I was constantly somewhere else in my thoughts. I was criticised for that and it upset me a great deal’. She was not angry with the manager, on the contrary, she sympathized with him. After all, he was right, she was absent- minded. Catherine is able to relate her symptoms to circumstances. She realizes that they are at least contributory to the development and continuation of her symptoms. Catherine shows growing insight. The therapist says it is important

67 that she realizes her depression did not come out of the blue. In some sense it is understandable. She objects that it did not depress her colleagues. He proposes to jointly explore the impact of life circumstances on her mood. Everybody’s life, he says, is highly determined by what happens and even more by the way events are perceived. He does not miss how remarkably understanding Cather- ine is towards her critical boss but he does not bring the issue to the fore. 4.3. At the third level the focus shifts from life circumstances to one or more external, interpersonal relationships. This level specifies the former one by translating ‘circumstances’ into ‘relationships’. The therapist wonders how ‘being overloaded with work’ relates to ‘people in the work situation’ and where Catherine was ‘in her thoughts’ a year ago, while working at the hospital. Was there something the matter with important people in her life? Yes, there was her grandfather on mother’s side. He died around that time. ‘I loved my grandfather very much; in fact he was more like a father to me than my father was. But I think I am over it now, it happened such a long time ago.’ The therapist asks how grandfather died. The story brings tears to Catherine’s eyes. In those days she was a great comfort to her grandmother and her mother. She arranged the funeral almost all by herself. The therapist says: “You were there for your grandmother and for your mother but who was there for you?” Now she is silent and cries. The therapist wonders where her father was. He didn’t come to the funeral. Her parents separated when Catherine was eleven, after long years of unhappy marriage. The relationship between them never normalized. ‘Nobody asked me how I felt and how I was coping with grandfather’s death.’ ’After- wards, my mother had no time to ask about or listen to my feelings. Every time I tried to tell her, she said: ‘Stop crying, I’m getting too sad’. Catherine was not angry with her mother, on the contrary, she empathized with her. Her mother was the one having a hard time. After all, it was about her father and not, as in Catherine’s case, about her grandfather. The therapist keeps exploring facts and experiences. Meanwhile, he is moved by Catherine’s account. He sees a brave, forlorn child and wonders how he could reach it while Catherine herself is out of touch with it. He also feels outraged about what was done to her but just in time realizes that Catherine, who is totally unaware of it in herself, has aroused this feeling in him. Catherine’s insight increases. She realizes the importance of strained relationships in what happened with her. Still, she conceives her problem as ‘external’ and muses upon external solutions, e.g. looking for another job. 4.4. At the fourth level the focus shifts from one or more relational problems to one or more relational patterns in patient’s life. The therapist wonders about the situations at the hospital and the funeral. Different as they may be, do they present a common feature? ‘Yes, I felt overlooked’. Her colleagues, and particularly the head of the ward, never asked her how she herself was doing during the hectic period at the hospital. At the funeral the family was glad she managed the situation quite efficiently (‘I’m good at that’), for sure, but everybody seemed too much absorbed in their own grief to show any interest in hers. Catherine gradually recognizes the pattern, even in the relationship with her boyfriend. ‘Work has to be done, problems have to be tackled and it all falls

68 on my shoulders. I’m the worker ant of the ward and the help and stay of the family. Besides that, nobody seems interested in me. Sometimes I even doubt whether my boyfriend knows me.’ She is amazed, even frightened, by the reali- zation how deeply these feelings are entrenched in her life. The therapist feels less powerless. Clearly, mutual acceptance and cooperation begin to prevail in the therapeutic relationship. Catherine feels better and has started self-exploration. The therapist values her growing awareness of an interfering pattern or theme in her life (‘being overlooked’). Wonder is the beginning of wisdom, he says. Catherine shows growing insight but she conceives her problems and their potential solutions still external. 4.5. At the fifth level the focus shift from external problems and patterns to an internal issue: the patient’s attitude in life. The therapist does not challenge Catherine’s views on the selfishness and ingratitude of people. After a while Catherine increasingly puts her ideas into perspective. ‘Of course, I don’t make things easy for others. I get myself lumbered with responsibilities, I volunteer for work I am not supposed to do, and I resent asking for help. I’m good at putting on a happy face. Small wonder they see me as the uncompli- cated, funny one who needs no help at turning problems into challenges and opportunities.’ Catherine is quite knowledgeable on dog issues but at the dog- club she seldom speaks her mind from fear she might ‘give the wrong answer’. ‘I am always afraid they’ll think I am a stupid, fat girl’. Catherine is quite upset by the discovery that the problem is not ‘out there’ but inside her. ‘Doesn’t that make the problem irremediable?’ The therapist feels rewarded. Catherine is able to discuss her own contribution in maintaining the interpersonal pattern she has become aware of at level four. She now exhibits psychoanalytically defined self-insight. Apparently she trusts him and feels sufficiently safe to explore uncharted waters. He is well aware she has taken quite a step and no interpretations were required to get her to do so. He realizes she is doing what he wants her to do: question herself instead of others. He admits to her that this seems to him the right path to follow. He understands her new insight frightens her and comments that the chances on changing herself are considerably higher than those on changing the world. 4.6. At the sixth level the focus shifts to an epigenetic explanation of level five. It is about how past relationships persist in the patient’s actual life. The therapist says he doubts Catherine’s attitudes are just inborn and suggests looking for their sources. The aim, he explains, is not ‘digging up the past’ but ‘understanding the present better’. This is not a difficult task for Catherine. ‘I can still hear my father say to me: ‘You are a stupid, fat girl. I’d rather had the girl next door for a daughter.’ It hurts, she cries. When Catherine cried at father’s comments, he would reply she was ‘a moron to cry for this’. She fears the therapist would think the same without saying. He says he doesn’t, on the contrary, he understands she couldn’t possibly feel safe with her father. Father was not always that derogative though. Each time she came home with good school marks, merited by hard work, he would praise her. Her mother was a master in covering up. She frequently told Catherine not to cry or to stop crying, and cleared the sky by making funny remarks. ‘Together we laughed a lot’.

69 Meanwhile mother suffered from various ailments. She didn’t speak about it, but Catherine knew. It still happens that her mother’s suffering face haunts her. Catherine was often to be found at her grandparents’ home. The therapist feels Catherine progresses in understanding instead of condemning herself. The relationship is really collaborative now. Self-insight expands to the aftermath of early experiences. The discourse is about former external relationships living on in the present as a result of internalisation. It is about internal interpersonal relationships, ‘the other in me’, not of course the historical but the narrative other. Catherine makes contact with the forlorn child she once was and deep down still is. It didn’t take genetic interpretations for her to do so. The therapist says Catherine is on track of an important echo of her past into the present. It may be a painful trail but in the end it will be rewarding. 4.7. At the seventh level the focus shifts to the relationship the patient maintains with herself as the consequence of identification with internal-interpersonal relationships. ‘It may be true’, Catherine says, ‘that my parents from the past still live on in me but how is it possible that they still have so much power over me? I’m not a little girl anymore!’ She has not had contact with her father in years. She considers her mother a bore, still she phones her daily. She has a boyfriend who repeatedly and unambiguously demonstrates he finds her attrac- tive, smart and loving. Annoyingly enough, she doesn’t seem to believe him and secretly fantasizes about flirting with other men, if not about more than that. She feels guilty and shameful about her thoughts. The therapist says she has yet another secret: ‘I am a stupid, fat girl who is not allowed to cry or to be unhappy and who has to do nothing but work and care for others.’ This, the therapist explains, is not establishing a fact. It is the passing of a judgement by a judge (‘You are a stupid, fat girl etc.’) on a convict who resigns herself to the verdict. Does it get through to her that once the judges were her father and mother, but now she is the judge herself? That she has become the judge her parents once were? It dawns on Catherine that she is treating herself as she was treated as a child in her remembered, subjective past. ‘Yes, I am the one who thinks I am unattractive and dumb, yes, I condemn myself to work and care for others, yes, I forbid myself to cry or to be unhappy.’ In other words, she realizes that she has identified with internal interpersonal relationships. At present, it is not her father who belittles her, nor is it her mother who dismisses her sorrow. She is treating herself the way her parents treated her, or at least the way she feels they treated her. The therapist muses: the more Catherine realizes what she is doing to herself, how she does it, why she does it and what the normal consequences of it are, the better/higher the chances are she will be able to alter it. 4.8-9. At level eight and nine the focus shifts to how the problems discussed at the levels four to seven manifest themselves in the relationship with the therapist. It regards working in the transference (level 8) and transference neurosis (level 9). It is not the SPSP therapist’s intention to elaborately or deeply discuss the transference. With some patients, however, some discussion of transference is possible and indicated or even unavoidable, e.g., when the patient’s transference interferes with the working alliance.

70 spsp starts at level one and in many cases rises to level five. The levels six and seven are reached considerably less often. Level eight may be reached for some time with some patients but can hardly be worked through fully. Level nine is unattainable. The therapeutic course is normally characterized by considerable fluctuations in the level of discourse. The therapist is expected to adapt to this variability.

4.5 Position of SPSP among other psychodynamic orientated therapies

SPSP obviously differs from long-term therapies, but it also varies from the existing short-term psychoanalytic therapies. Markowitz (Markowitz et al., 1998) characterize Short-Term Psychodynamic Psychotherapy (STPP) as ‘a treatment of less than 40 sessions that focuses on the patient’s re-enactment in current life and the transference of largely unconscious conflicts deriving form early childhood’; ‘key techniques are psychoanalytic, such as confrontation, interpretation, and work in the transference.’ ‘STPP, even when emphasizing events, focuses on transference in the office and the linking of extra-session interpersonal events to the transference.’ These characteristics refer to an inter- pretative and not a supportive technique. There are many non-psychoanalytic short-term therapies. We will not attend to this subject in this article. However, the relationship between SPSP and InterPersonal Therapy (IPT) deserves some discussion. Both modalities share common features. Klerman (Klerman et al., 1984) reports: ‘In training IPT psychotherapists and in discussions with clinical and research colleagues, we find a close relationship between IPT and dynamic psychotherapies. Many experienced, dynamically trained and psychoanalytically oriented psychotherapists report that the concepts and techniques of IPT are already part of their standard approach’. Elkin (Elkin et al., 1985) character- izes cognitive behaviour therapy and IPT, as ‘(...) to some extent, representa- tives of two different and important basic orientations to psychotherapy, the behavioural and the psychodynamic.’ In their research project, only therapists who had been trained in psychoanalytic psychotherapy were eligible as IPT therapists. On the other hand, there are significant differences between IPT and SPSP. Elkin (Elkin et al., 1985) states: ‘this approach uses techniques derived from psychodynamically oriented therapies, but treatment is focused on the patient’s current life and interpersonal relationships’. Klerman (Klerman, 1984) writes: ‘IPT is concerned with interpersonal, not intrapsychic phenomena’. The IPT focus focuses on the current, depression-related relational context. Using the aforementioned nine levels of discourse one might say that IPT is characterized on level 3 (the interpersonal level). The aim of SPSP is to achieve the discourse that is necessary and possible. Obviously in some cases it is not possible to go beyond level 3 (or even to exceed level 1). In these instances there may be not much difference between SPSP and IPT. Clearly at this moment we do not know which approach is preferable for an individual patient. A direct head to head comparison is required to shed more light on this.

71 4.6 Empirical validation of SPSP

Until now, the efficacy of SPSP in ambulatory patients presenting a DSM-IV defined, mild to moderate major depressive disorder has been tested in three Randomized Clinical Trials by De Jonghe (Jonghe et al., 2001; Jonghe et al, 2004) and Dekker (Dekker et al., 2007). A single research group conducted the three trials in different but similar study populations, using an identical research design, with each trial lasting six months. Patients were treated with SPSP, pharmacotherapy or combined therapy, i.e., the combination of SPSP and pharmacotherapy. De Maat (Maat et al., in press) performed a mega-analysis pooling the original data of the three RCTs. The mega-analysis compared SPSP (n=97), pharmacotherapy (n=45) and combined therapy (n=171). Independ- ent observers (17-item Hamilton Depression Rating Scale, HDRS), therapists (Clinical Global Impression of Severity and of Improvement, CGI-S and CGI-I) and patients (depression sub-scale of the ninety Symptom Checklist, SCL-90, and the Quality of Life Depression Scale, QLDS) assessed efficacy. Pearson chi-square calculations (level of significance .05) were used to compare base, dropout, response and remission rates. ANCOVA analyses (including baseline scores for outcome measures as covariates) were used to test inter-group differences. The results showed no significant differences in dropout rates, although dropout from both pharmacotherapy and SPSP was lower in combined therapy than in the corresponding single treatments. Table 4.1 summarizes the efficacy results, expressed in success percentages at treatment week 24 (per protocol design). as far as symptom reduction was concerned, independent observers (p=0.214) did not find differences between SPSP and pharmacotherapy. Patients (p=0.036) and therapists (p=0.026) did, in favour of SPSP. All three consistently found differences favouring combined therapy over pharmacotherapy (patients, p=0.000; therapists, p=0.024; independent observers, p=0.024). Independent observers (p=0.062) and therapists (p=0.430) found no differences between combined therapy and SPSP, but patients (p=0.016) found combined therapy superior. As far as improvements in quality of life were concerned, patients found no differences between SPSP and pharmacotherapy (p= 0.073) or between SPSP and combined therapy (p=0.217). However, they did find combined therapy superior to pharmacotherapy (p=0.015). The results of this mega-analysis suggest that, in the treatment of outpatients with

Table 4.1: Efficacy of pharmacotherapy, SPSP and combined therapy. Scale Rater Success percentage Pharmacotherapy SPSP Combined therapy HDRS Independent 24% 31% 40% CGI-I Therapist 49% 66% 65% SCL Patient 44% 64% 73% QLDS Patient 40% 53% 58%

72 mild to moderate major depressive disorder, independent observers, patients and therapists alike found SPSP plus pharmacotherapy to be more efficacious than pharmacotherapy alone, for both symptom reduction and quality of life improvement. They also suggest that independent observers found SPSP and pharmacotherapy equally efficacious. There is some indication that patients and therapists favour SPSP over pharmacotherapy for symptom reduction, but not for quality of life improvement. No efficacy difference was found between SPSP and combined therapy, except that patients thought that combined therapy was more efficacious in terms of symptom reduction. Kool (Kool et al., 2003) demonstrated that the superior efficacy of combined therapy over pharmacotherapy shows particularly in cases presenting depression combined with personality pathology.

4.7 Conclusion

SPSP is a valuable extension of the existing options for the treatment of depressed patients. It emphasizes the therapeutic potentials of psychoanalytically defined support. Further research regarding its efficacy in depression and other disorders is warranted, including its relative efficacy compared to that of other approaches.

References

Balint, M. (1952). Primary love and psychoanalytic technique. London: Hogarth Press. Bion, W.R. (1962). Learning from experience. London: Heinemann Medical Books. Bowlby, J. (1969). Attachment and Loss. New York: Basic Books. Dekker, J., Molenaar, P., Kool, S., et al. (2005). Dose-effect relations in time-limited combined psycho- pharmacological treatment for depression. Psychological Medi- cine 35: 47-58. Elkin,I., Parloff, M.B., Hadley, S.W., et al. (1985). NIMH Treatment of depression collaborative research program: Background and research program. Archives of General Psychiatry 42: 305-316. Fava, G. (2006). The intellectual crisis of psychiatric research. Psychotherapy and Psy- chosomatics 75: 202-208. Fonagy, P., Target, M. (1996). Playing with reality I: Theory of mind and the normal development of psychic reality. International Journal of Psychoanalysis, 77: 217- 234. Freud, S. (1957). Instincts and their vicissitudes, 1915. Standard Edition, 14: 117-140. London: The Hogarth Press. Freud, S. (1961). The Ego and the Id, 1923. Standard Edition, 19: 12-66 London: The Hogarth Press. Jonghe, F. de, Kool, S., Aalst, G. van, Dekker, J., Peen, J. (2001). Combining psychotherapy and antidepressants in the treatment of depression. Journal of Affective Disorders 64: 217-229. Jonghe, F., de, Hendriksen, M., Aalst, G., van, et al. (2004). Psychotherapy alone and combined with pharmacotherapy in the treatment of depression. British Journal of Psychiatry 185: 37-45.

73 Klein, M. (1975). The Writings of Melanie Klein, volume 1-4. London: The Hogarth Press. Klerman, G.L., Weissman, M.M., Rounsaville, B.J., Chevron, E.S. (1984). Interpersonal therapy of depression. New York: Basic Books. Kohut, H. (1971). The analysis of the self. New York: International Universities Press. Kohut, H. (1977). The restoration of the self. New York: International Universities Press. Kool, S., Dekker, J., Duijsens, IJ., de, Jonghe, F., Puite, B. (2003). Efficacy of combined therapy and pharmacotherapy with depressed patients with or without personality disorders. Harvard Review of Psychiatry 3: 133-141. Maat, S, de, Dekker, J., Schoevers, R., van, Aalst, G., Gijsbers-van Wijk, C., Hendriksen, M., Kool, S., Peen, J., Van, R., de, Jonghe, F.(2007). Short Psychodynamic Sup- portive Psychotherapy, Antidepressants and their Combination in the Treatment of Major Depression: A mega-analysis based on three randomized clinical trials. Depression and Anxiety (in press). Markowitz, J.C., Svartberg, M., Swartz, H.A. (1998). Is IPT time-limited psychodynamic psychotherapy? Journal of Psychotherapy Practice Research 7: 185-195. Winnicott, D.W. (1965). The maturational processes and the facilitating environment. London: The Hogarth Press.

74 Chapter 5 Support and personality change: a psychoanalytic view

Frans de Jonghe, Saskia de Maat, Rien Van, Marielle Hendriksen, Simone Kool, Gerda van Aalst, Robert Schoevers and Jack Dekker Submitted

Abstract

In the classic, psychoanalytical views, interpretation leads to insight, which in turn results in personality change, if all goes well. We do no challenge these views but contend that a second, largely neglected process also plays an important part: support may lead to personality change, which in turn may result in insight. Essential to our proposition is the notion that internal relationships represent vital personality aspects. Their alteration is crucial to personality change. We reformulate the longstanding interpretation versus relationship controversy as an interpretation versus support debate. We propose a psychoanalytic definition of support: proper gratification of unmet developmental needs, as they appear in the archaic aspects of the therapeutic relationship. Then, the controversy comes down to divergent views on the relative roles of interpretation and gratification. We stress the importance of differentiating between adequate and inadequate support. We present some ideas about the putative mode of action of support, suggesting that it may reside in its power to evoke in the patient the experience of a “dissonance” between two incongruous aspects of the therapeutic relationship. Both are simultaneously felt in the present, one being determined by it, the other by the past.

5.1 Introduction

From the very beginning, Freud made clear that the psychoanalytic therapeutic endeavour primarily regards personality change: “Psychoanalytic therapy was created through and for the treatment of patients permanently unfit for existence, and its triumph has been that it has made a satisfactorily large number of them permanently fit for existence” (1905, p. 263). “Being unfit for existence” we nowadays call “being vulnerable”. Being able to cope with the vicissitudes of life depends, to a large extent, on the maturity of the individual’s personality. Therefore, pursuing personality change (“structural change”) is fostering development. As Loewald states: “If structural changes in the patient’s personality mean anything, it must be that we assume that ego-development is resumed

75 in the therapeutic process in psychoanalysis” (1960, p. 16) Ego-development is probably best understood as increasing the individual’s differentiation and integration. There is a long-standing controversy among psychoanalysts about the relative value of interpretation and support in forwarding growth (Schech- ter, 2007). first of all, it must be clear that when we use the word support we mean a feeling state of the patient. A “supportive” therapy, relationship or technique refers to the intention of the therapist. However, the ultimate criterion is the patient. “Support” is actually supportive only as far as the patient feels supported. In this article, first we outline a psychoanalytic, relational view on personality. Secondly, we reassess the interpretation-relationship controversy. Next, we define support psychoanalytically. Fourthly, we differentiate between adequate and inadequate support. Then, we present some ideas about the therapeutic mode of action of support. Finally, we try to clarify the relationship between support, personality change and insight. Psychoanalytic interpretation, important as it is, is not our subject. We illustrate our views with vignettes from a case that has been modified for discretion’s sake. Still we trust the vignettes convey the essence of the message.

5.2 A psychoanalytic, relational view on personality

Our views regarding support and personality change are based on a psychoanalytic, relational perspective on personality. Hereafter we summarize the concepts we consider most important for our present discussion. from our point of view, present-day psychoanalysis mainly consists of six part-theories: instinct-theory (Freud, 1915a), ego-psychology (Freud, 1923), object-relations theory (Klein, 1975), self-psychology (Kohut, 1971, 1977), attachment theory (Bowlby, 1969) and primary love theory (Balint, 1952). Alto- gether they regard six innate, basic, social needs: sexuality, aggression, the need to engage in relationships, and the needs to be loved, protected and esteemed. The last four are called “developmental” because they must be met adequately in early infancy in order to allow the first growth stages to unfold. The social need concept implies relationships. There are no social needs without relationships, for they are relationally defined. Otherwise there are no relationships without social needs, for relationships are need-driven. We consider relationships cognitively, affectively, conatively and behaviourally loaded interactions with others and with oneself. We distinguish between “external” and “internal” relationships, and contend that internal relationships represent vital personality aspects. External relationships are interactions with others present “out there in the outer world”. We call them “external-interpersonal”. One type of internal relationship (the other is to be discussed hereafter) regards interactions with others present only in the “inner world of the person’s mind”. We call them “internal-interpersonal”. In a process called internalisation, external-interpersonal relationships are internalised into internal-interpersonal relationships.1 It is not the object, the affect

76 or the need that is internalised; it is the relationship. It may be conscious, partly conscious or unconscious, as is exemplified by Steven’s case.

Steven, a thirty-six old bench man goes at great lengths to help people. What happened the other day is a telling example. A tourist asked him to point out the way to his hotel. It was not nearby and not easy to explain either. Steven kindly suggested bringing the man to his hotel and invited him into his car. Overcoming some misgivings, the man accepted. Unfortunately, but not surprisingly, Steven ran into several traffic jams in the charmingly narrow streets of old Amsterdam. He ended up driving more than an hour to the other side of town and back. It caused him to be late at work, where his boss showed little understanding. Steven realizes he makes a fool of himself with this and similar actions. At first, he is not aware how much he is still obeying his late mother, who used to say: “We do not own our own lives, we are there to help others”. He gradually realizes she meant he was there for her. For years he has not thought of her, he did not miss her, he did not attend her grave. Now he understands she inhabits him, wrapped in the dismissed thoughts, feelings and wants that did not pass away with her. “When your parents die”, he says, “they move in with you. My mother may be stone-dead, she still is with me”. She used to say: “Don’t cry, sing a song.”

A second type of internal relationship we call the “IntraPersonal Relationship” (IPR). The IPR consists of the interaction between I and Me. It too may be conscious, or partly conscious or unconscious.

Steven’s life has not been easy. He comes from a large and poor family. The education he received was limited. From age fourteen Steven has been working in various jobs that required hard physical labour. It hasn’t brought in much. “They still consider me nothing”, he agrees with a sigh. The therapist wonders with whom he agrees. For a long time during therapy, it is with the whole world. There are convincing examples in abundance. Gradually it dawns on Steven that he is collecting proof of what he knows for sure beforehand. Then he realizes he agrees with himself rather than the world: “I still consider myself nothing”. His parents both died when he was in his early twenties. He has lost two brothers to cancer. Five years ago sudden death struck a dear nephew, aged twenty-four. Steven did not mourn, he did what he had always done: work hard. He steadily becomes aware that he is obeying his father’s order: “Cut the crap and get down to work”, but now he his both ruler and slave at once.

There is a causal link between the internal-interpersonal relationships and the IPR. The IPR results from a step following the internalisation of external-interpersonal relationships. in this second step, the subject identifies with (some aspects) of (some) internal-interpersonal relationships. Figure 5.1 illustrates the issue. even when the two types of internal relationships are separately perfectly conscious, the connection between them may be partly or totally unconscious.

Steven becomes conscious of the fact that he forbids himself to mourn, compels himself to work and considers himself worthless. He still can hear his father saying: “Do not give the dead too much thought, or else you might as well lie down next to them in their graves.’, and “We are nothing, we have to work hard.” However, he still wonders why his own opinions and that of his father are so similar.

77 External Internal

Internalization

Interpersonal External-interpersonal Internal-interpersonal

Identification

Intrapersonal ……………………… Intrapersonal

Figure 5.1: External and internal relationships.

Internal-interpersonal The other The subject

Identification Splitting

Intrapersonal The subject The subject

Figure 5.2: From internal-interpersonal to intrapersonal relationship.

The identification process, which brings about the rising of the IPR out of the internal-interpersonal relationships, is more complex than indicated in figure 5.1. It implies a developmentally normal splitting (“horizontal splitting”) indicated by Freud already in 1933: “The ego can take itself as an object, can treat itself like other objects, can observe itself, criticize itself and do Heaven knows what with itself. In this, one part of the ego is setting itself over against the rest.” (p. 58) In a process called introjective internalisation, one part of the split subject identifies with aspects of “the other” (of the internal-interpersonal relationships), as indicated in figure 5.2. in this view, it is not only the internalised relationships of the patient with significant others, especially in childhood, that hamper his life, but even more so the relationship he maintains with himself. Everybody presents an IPR. We consider it at the core of personality. What differentiates people, to a large extent, is the content of their IPR discourse. As a rule, patients seek help, not for an unconscious malevolent IPR itself, but for its natural, logical, painful conscious consequences.

78 Steven “buys” appreciation with pleasing behaviour. On the other hand, he is suspicious of his loving partner, being never sure that she will stay. His unreasonable fits of anger may well make his darkest prophecies come true. He feels chronically rejected, lonely, inferior, depressed. Against all expectations, he desperately craves and hopes for recognition. He fears and expects rejection and consequently shuns contacts by withdrawing into secret grandiosity. All this he considers: “That’s me, that’s who and how I am”. He asks help but does not believe these personality characteristics are remediable.

Developmental needs inadequately met in early infancy persist in adults as early infantile, internal relationships, that are malignant and ongoing. Both types of internal relationships (internal-interpersonal and intrapersonal) present benevolent and malevolent aspects. They act as Procrustes’ bed, i.e., they are distorting moulds or templates, tending to recreate old relationships out of potentially new ones. Internalisations and externalisations combine into a spiralling process, as indicated in figure 5.3. the externalisation process results in a distorted external reality, creating “illusory others” or “personifications”. It is twofold. First, in a process called projection (“projection onto the other”), the patient attributes his own attitudes to the other. Concerned care easily experienced as coerciveness (Schafer, 2005) is an example. Secondly, in a process called projective identification (“project- ing into the other”), the patient invites, pushes, or even coerces the therapist into enacting a role. It goes without saying that the twofold process indicated above equally applies to the therapist, but one may assume to a lesser degree. These views apply to external relationships in general. They are, in various aspects and degrees, externalised relationships. In therapy and in life in general, the differentiation between external and externalised problems is tremendously important. External solutions may be efficacious with external problems but they are impotent with externalised problems.

Lovelorn Steven seriously considers moving to another town. He admits this would entail loosing a comfortable home, a satisfying job and a circle of friends but he wants to start again with a clean slate. The therapist understands the wish to get a new perspective on new things but wonders whether getting a new perspective on old things is not equally good and certainly better than keeping an old perspective on new things.

Internal-interpersonal Internalisation relationship

External-interpersonal relationship Externalisation Intrapersonal relationship

Figure 5.3: The spiral of internalisations and externalisations.

79 Early on in life, a vicious circle (or better: spiral) comes into being, as (partly) externalised relationships are in their turn internalised afterwards. The externalisation-internalisation process is a tremendously conservative factor throughout mental life. It presents a major obstacle to growth, is a major determinant of the compulsion to repeat and keeps persons prisoners of their pasts. We hold it highly responsible for “an enduring pattern of inner experience and behavior that deviates markedly from the expectations of the individual’s culture, is pervasive and inflexible, has an onset in adolescence or early adulthood, is stable over time and leads to distress or impairment”, the very DSM-IV-R definition of personality disorder (p. 629). It may help explain “the tripartite failure involving the self system, kinship relationships and societal relationships” characteristic for personality disorder (Livesley, 1998) and the “harmful dysfunction” implied by “the failure to solve life tasks involving the development of integrated representations of self and others, and the capacity for adaptive kinship and societal relationships.” (Livesley and Jang, 2000) personality change, be it structuring or restructuring, mainly consists of altering the templates, verily a high aiming attempt. The therapeutic situation is propitious for eliciting the expression of the internal relationships of both participants in the therapeutic relationship. More often than not, they lay bare in the (counter)transference and archaic aspects of the therapeutic relationship.

Steven assumes that his actually neutral therapist professionally conceals his criticism on Steven’s actions. He guesses the therapist internally chuckles at his failing attempts to cope with his wife and boss. Besides that, Steven correctly perceives some boredom in the therapist without realizing to what extent he induces this feeling in the latter.

5.3 The interpretation-relationship controversy revisited

In the classic views, growth is the result of self-insight, which in its turn results from interpretation, the hallmark of psychoanalytic technique. In 1954, Bibring put the idea clearly into words: “In psychoanalysis proper, all therapeutic principles are employed to a varying degree, in a technical, as well as in a curative sense, but they form a hierarchical structure in that insight through interpretation is the principal agent and all others are – theoretically and practically – subordinate to it.” (p. 762). In 1987, Cooper still had no doubts: “Despite the diversity of analytic views that abound today, analysts seem to agree on the centrality of the transference and its interpretation in analytic process and cure, differing only in whether transference is everything or almost everything.” (p. 87); and: “The transference and its interpretation are at the centre of all considerations of analytic theory and technique.” (p. 97) Even in 2004, Paniagua, although acknowledging that consensus on the conceptualisation of “interpretation” has proven elusive, still pithily stated: “Psychoanalysis is a discipline of interpretation”. (p. 1105) Over the last decades, the classic views have increasingly been debated. Growth fostering mechanisms other than interpretation, mostly called “rela-

80 tional”, “interactional” or “intersubjective”2, have more and more become recognised as important. In 1971, Greenson voiced the doubts of many when he communicated: “My clinical experience leads me to believe that the final solution of the transference neurosis depends to a great extent on the transference neurosis being replaced by a real relationship. I do not share the tradi- tional psychoanalytic point of view that interpretation alone can resolve the transference neurosis. Interpretation has to be supplemented by a realistic and genuine relationship to the person of the analyst, limited though it may be, for the transference neurosis to be replaced.” (p. 231) In 1988, Stern et al. stated: “It is by now generally accepted that something more than interpretation is necessary to bring about therapeutic change”. (p. 903) With “something more”, the quality of the therapeutic relationship is meant. Today, the “intersubjective school” in psychoanalysis (R. Stolorow, G. Atwood, D.M. Orange, among others) even deprives interpretation of its prerogative as the principal agent of psychic change, transferring this to the intersubjective relationship itself (Andrade, 2005). Renik (2004), one of the most ardent champions of the new approach, contends countertransference enactment and self-revelation are not only unavoidable but even recommendable: “… in order to facilitate intersubjective exchange in the clinical situation, the analyst must be willing to make his/her own relevant experience as fully available to the patient as possible.” (p. 1056) In this approach the “something more”, if there is any, no longer is the relationship, but interpretation. in our view, the interpretation-relationship controversy starts from false premises. The opposite of “interpretive” is not “relational” but “non-interpretive”. Whether the approach is interpretive rather than non-interpretive does not depend on the relationship only. The therapeutic setting, frame, contract, attitude, interventions and technique contribute too. “Non-interpretive” being a definition by negation, we suggest replacing it by the affirmative term “supportive”. So, the controversy is not between interpretation and relationship, but between interpretation and support. It regards the relative importance attributed to interpretation and support, the latter being derived from the relationship or otherwise. Unfortunately, from the very beginning of psychoanalysis up to now, support presents three striking features: it bears negative connotations among most psychoanalysts, its definition is unsatisfactory and its mode of action is poorly understood. Indisputably, its bad reputation has historical grounds. Freud (1919) calls it “suggestion”, an improper element that must be prevented from corroding the pure gold of psychoanalysis. (p. 168) Bibring (1954) calls it “manipulation”, assuring that he uses the word in a completely neutral sense. A more important reason, however, may be that adequate support is seldom differentiated from inadequate support. Well intended but inadequate support is omnipresent. It blurs the existence of adequate support. “Interpretive” is self-evidently seen as “adequately interpretive”. Unfortunately, “supportive” seems to mean “inadequately supportive” to many. The controversy, if there is one, should regard adequate interpretation and adequate support. there is not just one psychoanalytic therapy, there are many. Luborsky (1984) locates them on a continuum extending from a “purely expressive” to a

81 “purely supportive” pole. We agree with the continuum idea but would call the “purely expressive” pole “purely interpretive” and would add that the “purer” the therapy, the poorer its effectiveness probably is. In short, the main difference between the various psychoanalytic therapies regards the relative importance attributed to interpretation and support. Simply put, instinct-theory, ego-psychology and object-relations theory underlie an interpretive approach, while self-psychology, attachment theory and primary love theory underlie a supportive approach. Considering the disrepute of the term “support”, we suspect Kohut, Bowlby and Balint would turn in their graves if they learnt that we call their approaches supportive, unless we could explain to them what we mean with “psychoanalytically defined support”. We also consider the mentalization- based treatment of Bateman and Fonagy (2004) supportive. In this, we find Siegel (1999) on our side: ‘Earned’ secure/autonomous status is most often achieved through supportive personal or therapeutic relationships (for example, marriage or psychotherapy)”. (p. 313) (italics added)

5.4 Psychoanalytically defined support

In our view, to support, psychoanalytically speaking, means to gratify (which does not imply that to gratify necessarily means to support). To gratify means to recognize and fulfil a need (Loewald, 1960). Support furthers growth by properly gratifying unmet developmental needs as they appear in the archaic aspects of the therapeutic relationship. “To gratify” might be a less equivocal expression than “to support”. Alas, to many this term bears even more negative connotations. Ferenczi’s unconven- tional technique surely did not do things any good. Early on, however, Freud (1915) was convinced that the analyst could supply nothing but “ersatz” gratification: “And what we could offer would never be anything else than a surrogate, for the patient’s condition is such that, until her repressions are removed, she is incapable of getting real satisfaction.” (p. 165) This would entail that real support is impossible. Understandably, “interpretation instead of gratification” is the leading idea of the classic psychoanalytic technique. Non-interpretive interventions are considered ancillary at best and “suggestion”, “influence” or “gratification of an illusion” at worst. Transference gratification is especially cautioned against, as it would lead to a “transference cure by strengthening defenses” at best. Countertransference gratification is considered wrong alto- gether. No doubt transference-countertransference gratification exists and often takes the form of inadequate support, but we are interested in adequate support. today, most analysts agree that patient and analyst continually influence each other in a dyadic, intersubjective system. Meissner (1998) argues: “Effec- tive neutrality and abstinence require a capacity to shift sensitively along the continuum between the poles of gratification and frustration, depending on the intensity of the needs and desires emanating from the patient.” (p. 1104) For Greenberg (2001), it goes without saying that the analyst gratifies a wide

82 range of needs and desires, and that this motivates the patient to enter and stay in treatment. An “affirmative stance” (Kohut, 1971) and a “position of functional neutrality” (Kris, 1990) are far less controversial now than they were first. Schechter (2007) breaks a lance for “validation”, understood as the experience of having been perceived, understood and accepted as legitimate by an important other. (p. 106)

5.5 Adequate and inadequate support

A young tree may need support in order to grow, an old tree in danger of top- pling down needs buttressing and supporting a sturdy tree probably does more harm than good. Supporting a hungry man may consist in giving him a fish or in teaching him how to fish. However, to teach a starving man, a desert-dweller or a fisherman how to fish is pointless. The examples may illustrate the fact that support may be fostering progression, maintaining the status quo or furthering regression. These three support types may be adequate or inadequate, resulting in six possibilities, as figure 5.4 tries to elucidate. the first support type fosters progression. It counters regression and for- wards development, i.e., it stimulates the resumption of arrested and the correction of crooked growth. It is (1) adequate if progression is needed and possible. It is (2) inadequate if the support overestimates the patient’s needs or capacities. The second support type maintains the status quo and counters decay. Perhaps it deserves the name “buttressing” better. “Buttressing therapy” accepts regression and developmental problems without trying to alter them. Its aim is to prevent deterioration. It is (3) adequate if the patient cannot cope with a more challenging approach, temporarily or permanently. Other- wise, it (4) underrates the patient’s needs and capabilities and is consequently inadequate. The third support type fosters regression and developmental arrest. The setting of psychoanalysis proper fosters a mild to moderate, temporary regression “in the service of the ego” (Kris, 1936, p. 290). In that context it may be (5) adequate, in all other circumstances it is (6) inadequate. Unfortunately, this inadequate variant is a very frequent consequence of well-intended help by therapists deluded by the patients’ regressive position and projective identifications.

Steven is ambivalent about therapy. Can the therapist guarantee that it will help? He is distraught with fear that his wife will never come back. Can the therapist talk

Support fostering Adequate Inadequate Progression 1 2 Status quo 3 4 Regression 5 6

Figure 5.4: Three types of adequate and inadequate support.

83 to her? His position at work deteriorates and rightly so for he is absent-minded nowadays. Can the therapist give him sound advice? Is taking sick leave a good idea or shall he bite the bullet? The therapist realizes Steven is asking for inadequate support. He tries to avoid the pitfall. Yes, he understands perfectly well the wish for success guarantees, practical interventions and advice but, no, he doesn’t think this would be the right way to help him. However, he may offer Steven something valuable. Why not take the chance that the therapy would help? Why not discuss what went wrong between him and his wife and how to proceed now in his best interests? Why not discuss how to cope with his understandable absent-mindedness problem at work? Why not consider all the arguments in favour or against taking sick leave?

Hereafter, when we use the term “support” we mean adequate, progression fostering support, i.e., support that counters regression and furthers growth in accordance to the person’s needs and capacities. in humans, the young always need support in order to thrive. What about adults? Adults by definition are full-grown, at least physically. Can they still grow psychologically? As stated before, growth or development consists of an increase in integrated differentiation. As this is the ultimate goal of psychoanalytic therapy at any age or life stage, therapists apparently assume - we trust on good clinical grounds - that in adulthood, later adulthood and even in old age, some, perhaps many, present un-actualised growth potentials that await support. It is a well-known fact that Freud had his doubts on that matter, although he made his major contributions to psychoanalysis well past his fifties.

5.6 The therapeutic action of support: the support-experience track

The specific experience evoked by support is probably the agent of change understood least well. Its intangible nature is most likely due to the fact that at its deepest level it is not reflexive, explicit, verbal, symbolic, declarative or repressed. At that “procedural knowledge level” it is unconscious but not dynamically so. Supposing Freud had heard of it, he probably would not have been interested, at least not in 1917: “Where no repression (or analogous psy- chical processes) can be undone, our therapy has nothing to expect.” (p 435) Present day psychoanalysts, at least some of them, are the more interested. We suggest the therapeutic action of support resides in its power to evoke a specific experience: dissonance, friction, incongruity. If we focus, for simplici- ty’s sake, on the therapeutic relationship (omitting other also important factors such as setting and technique), support resides in the experience of a “relational dissonance” in the therapeutic relationship. Figure 5.5 tries to elucidate the process between a patient with basically rather malignant and a therapist with basically rather benign internal relationships.

The figure may be clarified using Steven’s case (1) steven experiences the therapist as interested, listening and understanding. Somehow, the therapist even seems to enjoy the contact with him. This is quite unfamiliar to him. Preconditions to this new experience are that the

84 Patient’s relatively Patient’s relatively malignant internal 6 intact reality testing relationships Internalisation of the 1 2 prevailing ‘new’ relationship Patient’s experience of a Patient’s experience of an ‘new’, relatively benign ‘old’, relatively malignant therapeutic relationship therapeutic relationship 5

Patient’s experience of a 4 The ‘new’ relationship relational dissonance prevails over the between ‘old’ and ‘new’ ‘old’ one

Figure 5.5: The therapeutic action of support.

therapist offers support and that Steven presents sufficient reality testing (and sufficiently benevolent, internal relationships) to experience the therapist as he really is. (2) Steven is punctual for the therapy appointments. Flu does not keep him from attending. His major concern is to prevent that the therapist, who works so hard for him, looses his time with him. Steven too works hard but he is not accustomed to talk about himself. How long will it take before the therapist will get sick of him? He probably soon will send him away. This would not surprise him, on the contrary, it would be exactly what he expected. Preconditions to this old experience are Steven’s rather malevolent, internal relationships, especially a malevolent IPR, expressing themselves in projections and projective identifications. (3) The punctuality and predictability of his therapist amaze Steven. Who has ever reckoned with him? Besides, his therapist seems interested in the twad- dle he produces. He listens, or is he feigning? What’s more, he seems to find the weirdest ideas understandable, not laughable. Could he be worth the effort ? The result of (1) and (2) is that the Steven experiences a “relational dissonance” between two incongruous external-interpersonal relationships, both simultaneously felt in the present, one being determined by the past and the other by the present. They both regard the same developmental needs as they manifest themselves in the archaic aspects of the therapeutic relationship. (4) Dissonance is not a persisting phenomenon. Sooner or later either the new or the old relationship will prevail.

85 (5) sooner or later, Steven will internalise the prevailing relationship and, once internalised, he will identify with it. If the “new”, benevolent relationship prevails and if its internalisation-identification occurs, the action is therapeutic. However, if the “old”, malevolent relationship prevails, and subsequently internalisation-identification occurs, the action is anti-therapeutic. (6) The internalised therapeutic relationship will influence the existing internal relationships. The nature of this effect is still a matter of debate on two possibilities, and their combination. It may be that the existing malevolent internal relationships are disconfirmed and therefore weakened (that they would be eradicated seems illusory). Another likelihood is that the existing internal relationships stay the way they were, but get company of a, possibly rather dominant, new one. Sandler et al. (1969) argue: “… in fact we probably do not ever get a real replacement of one type of relationship by another. What we see is rather the addition of various new types of object relationship developing collaterally, being integrated with and dominating the old, but not necessarily replacing them. Thus even in the person who has attained the most mature type of object love, the infantile aspects, for example the purely need-satisfying aspects, remain, although they may be subordinated to the higher-level developments.” (p. 643)

Insofar the therapist presents basically rather benign, internal relationships, most probably the situation for him will be opposite: in the therapeutic situation, a malevolent new relationship arises next to a benevolent old one. Func- tioning as a therapist requires that his “old”, mainly benign internal relationships are resilient enough to prevail in the end. The therapist’s capacity for reflective thought is a decisive factor here. the dual dissonance phenomenon arises when the felt qualities of a new, external relationship come close enough, but not too close, to those of an old one. If the new experience bears too much novelty, i.e., if it differs too much from the old one, the two feeling states pass each other with no wringing, jarring contact, hence without influencing each other. On the other hand, if the new experience is “not new enough”, i.e., if it resembles the old one too much, the old feeling prevails with retraumatization as a result. providing psychoanalytically understood support is not just giving the patient what he asks for, e.g., reassurance. Feldman (1993) pertinently remarks that the patient, fearing an encounter with a bad object, will try to reassure himself by identifying the therapist with either his good or bad-objects. If the therapist lends himself for it, consciously or not, the patient will feel reassured - a telling example of inadequate support. The therapist being neither a bad nor a good object is what frightens and destabilises the patient. We agree with Feldman that disappointment by and disillusion in the therapist are unavoidable and useful aspects of support. The patient will experience the adequately gratifying therapist as more gratifying than the bad object feared for, but less than the good object hoped for. Therefore, an adequately gratifying therapist as a rule is less gratifying than the patient wishes.

86 there are several concepts that approximate to “relational discordance”: “corrective emotional experience” (Alexander et al., 1946), “integrative experience” (Loewald, 1960), “dyadic states of resonance” (Siegel, 1999), “affect attunement” (Stern, 1985), and “present moments” (Stern, 2004). The terms mostly used are “new experience” and “new object”, the latter being defined as empathizer, container, holder or need-satisfier. However, the new experience is only one part of dissonant relationships. the dissonance idea is not entirely new. As early as 1934, Strachey hints at it: “If all goes well, the patient’s ego will become aware of the contrast between the aggressive character of his feelings and the real nature of the therapist, who does not behave like the patient’s ‘good’ or ‘bad’ archaic objects. The patient, that is to say, will become aware of a distinction between his archaic phantasy object and the real external object.” (p. 143) In the same vein Loewald (1960) asserts: “The patient can dare to take the plunge into the regressive crisis of the transference neurosis, which brings him face to face again with his childhood anxieties and conflicts, if he can hold on to the potentiality of a new object-relationship, represented by the analyst.” (p. 17-8) Stern et al. (1998) identify the “something more than interpretation that therapeutic change requires” as the impact of the “shared implicit relationship” between patient and therapist on the patient’s “implicit relational knowledge”. They call that impact “mutative information”. Chused (1996) describes the “informative experience creating emotional dissonance”: “Informative experiences arise out of those analyst- patient interactions in which the anticipated, the consciously or unconsciously desired or provoked reaction, does not occur, and what does occur is so jarring and discordant with expectations that action and thoughts are derailed, and expectations become suspect.” (p. 1051) He calls it “the dissonance between what is expected and what occurs.” (p. 170) however, there are differences between our views and those of the authors mentioned above. First, the basic needs most of them have in mind are sexuality and aggression, not developmental needs. Secondly, in their views the therapist evokes dissonance by abstinence, not by gratification. Chused (1996), e.g., states that “informative experience” is the desired result of an analyst maintaining a relatively neutral and abstinent stance. Ryle (2003) concurs that the therapist’s explicit non-reciprocation of problematic relationship procedures is the essential component of the relationship. Thirdly, most of them consider dissonance only possible after interpretation. According to Loewald (1960), e.g., it is the objective interpretation by the therapist of the patient’s transference distortions, resting on the collaboration of both therapist’s and patient’s observing ego’s, which makes the potentially new object actually available as such. We concede that longstanding psychoanalytic practice bears out these views. Nonetheless, we question whether interpretation is a necessary condition in all cases. Fourthly, and most importantly, the aforementioned authors consider dissonance an important phenomenon preceding successful interpretation, not a therapeutic factor in itself. For Chused (1996), e.g., “an informative experience precedes most successful analytic interpretation.” (p. 1070)

87 5.7 Support, personality change and insight

If all goes well, insight results from interpretation and results in personality change, in the classic views. In what could be called the post-classic views, personality change results from support and results in insight, if all goes well. Figure 5.6 contrasts the views. the dissidence seems insurmountable. It is indeed, if the adherents of one, or worse, both viewpoints, claim their theory to be necessary and sufficient to explain all psychoanalytic phenomena. The dissidence seems to evaporate if a radical partition is enforced: the classic views regard conflict pathology, the post-classic developmental pathology. Bad luck, this solution is too simple. it is generally agreed that, in conflict pathology, support plays at least an ancillary role. It favourably influences the therapeutic relationship and it helps interpretation to be effective. We take it a step further and contend that a patient who feels adequately supported may loosen up his defenses. They are not circumvented, broken through, or interpreted. The patient abandons them and proceeds to self-exploration. We do not exclude that support alone may suffice in some cases. in developmental pathology, many acknowledge support as an important therapeutic factor, be it under different denominations. Support is “establishing meaning by affirmative interventions”, not “revealing meaning by interpretive interventions” (Killingmo, 1989). It is stimulating progress according to the “developmental mode”, not the “integrative mode” or the mode of “strengthening of preferred defensive strategies” (Abrams, 1990). According to Kris (1993), support is “a sense of endorsement or affirmation and consists of the reversal of punitive, unconscious self-criticism”. It may be produced by interpretation from a position of “functional neutrality” but also by non-verbal means. Already in 1946, Alexander and French state that insight is frequently the result of emotional adjustment, not the cause. This is not to say that interpretation has no role in this area. It is necessary in some cases in order to make support possible. Besides, the supportive aspects of an interpretation cannot be ignored.

Interpretation Support

Insight Personality change

Personality change Insight

Figure 5.6: Interpretation, support and personality change.

88 interpretation furthers growth by diminishing what hampers development (defenses). Support furthers growth by providing what is necessary for development (adequate gratification). Jointly they may stand a better chance on achieving that the patient’s present life no longer is haunted by his past.

Conclusions

1. internal relationships represent vital aspects of personality. Their alteration is an important aspect of personality change. 2. the interpretation versus relationship controversy may be better formulated as an interpretation versus support debate. 3. psychoanalytically defined support consists of properly gratifying unmet developmental needs as they appear in the archaic aspects of the therapeutic relationship. 4. support may foster progression, maintain the status quo or further regression. Each type may be adequate or inadequate. 5. We suggest the therapeutic action of support resides in its power to evoke the experience of a “dissonance” between two incongruous aspects of the therapeutic relationship, both simultaneously felt in the present, one being determined by the past and the other by the present. 6. support may lead to growth, i.e. personality change, which in its turn may lead to insight.

Notes

1. We do not favour the terminology that calls external-interpersonal phenomena “intersubjective”, internal-interpersonal ones “intrapsychic”, and internalisation “introjection” or “incorporation”. 2. Many authors distinguish “intersubjective” from “intrapsychic” factors. The first term refers to what we call “external-interpersonal”, the second as well to “internal- interpersonal” as to “intrapersonal”.

References

Abrams, S. (1990). The psychoanalytic process: the developmental and the integrative. Psychoanalysis Quarterly, 59: 650-75. Alexander, F., French, T.M., et al. (1946). Psychoanalytic therapy. New York: Ronald Press. Andrade VM (2005). Affect and the therapeutic action of psychoanalysis. International Journal of Psychoanalysis, 86: 677-97. Balint, M. (1952). Primary love and psychoanalytic technique. London, Hogarth Press. Bateman, A.W., Fonagy, P. (2004). Mentalization-based treatment of BPD. Journal of Personality Disorders, 18: 36-51. Bibring, E. (1954). Psychoanalysis and the Dynamic Psychotherapies. Journal of the American Psychoanytical Association, 2: 745-70.

89 Bowlby, J. (1969). Attachment and Loss. New York, Basic Books. Chused, J.F. (1996). The therapeutic action of psychoanalysis: abstinence and informative experiences. Journal of the American Psychoanytical Association, 44: 1047- 71. Cooper, A.M. (1987). Changes in psychoanalytic ideas: transference interpretation. Journal of the American Psychoanytical Association, 35: 77-98. Feldman, M. (1993). The dynamics of reassurance. International Journal of Psychoa- nalysis, 74: 275-285. Freud, S. (1905). On psychotherapy. Standard Edition, 7: 257-68. London: Hogarth Press, 1964. Freud, S. (1915a). Instincts and their vicissitudes. Standard Edition, 14: 118-9. London: Hogarth Press, 1964. Freud, S. (1915b). Transference-love. Standard edition, 12: 159-71. London: Hogarth Press, 1964. Freud, S. (1917). Introductory lectures on psycho-analysis. Lecture 27. Transference. Standard edition, 16: 431-47. London: Hogarth Press, 1964. Freud, S. (1919). Lines of advance in psycho-analytic therapy. Standard Edition, 17: 157-73. London: Hogarth Press, 1964. Freud, S. (1923). The Ego and the Id. Standard Edition, ..-… London: Hogarth Press, 1964. Greenberg, J. (2001). The analyst’s participation. Journal of the American Psychoanyti- cal Association, 49: 359-81. Greenson, R.R. (1971). The “real” relationship between the patient and the psychoanalyst. In Kanzer, M., ed. The Unconscious Today. New York: Int Univ Press. pp. 213- 232. Killingmo, B. (1989). Conflict and deficit: implications for technique. International Journal of Psychoanalysis, 0: 65-79. Klein, M. (1975). The Writings of Melanie Klein, volume 1-4. London, Hogarth Press. Kohut, H. (1971). The analysis of the self. New York, International Universities Press. Kohut, H. (1977). The restoration of the self. New York, International Universities Press. Kris, E. (1936). The psychology of caricature. International Journal of Psychoanalysis, 17: 285-33. Kris, A.O. (1990). Helping patients by analyzing self-criticism. Journal of the American Psychoanalytic Association, 38: 605-36. Kris, A.O. (1993). Support and psychic structural change. In Horowitz et al. (eds) 1993. Psychic structure and psychic change. Madison: Int Univ Press, pp. 95-115. Livesley, W.J. (1998). Suggestions for a framework for an empirically based classification of personality disorder. Canadian Journal of Psychiatry, 43: 137-47. Livesley, W.J., Jang, K.L. (2000). Toward an empirically based classification of personality disorder. Journal of Personality Disordesr, 14: 137-51. Loewald, H.W. (1960). On the therapeutic action of psycho-analysis. International Jour- nal of Psychoanalysis, 41: 16-33. Luborsky, L. (1984). Principles of psychoanalytic psychotherapy. New York: Basic Books. Meissner, W.W. (1998). Neutrality, abstinence, and the therapeutic alliance. Journal of the American Psychoanalytic Association, 46: 1089-128. Paniagua, C. (2003). Problems with the concept ‘interpretation’. International Journal of Psychoanalysis, 84: 1105-23. Renik, O. (2004). Intersubjectivity in psychoanalysis. International Journal of Psy- choanalysis, 85: 1053-64. Ryle, A. (2003). Something more than the ‘Something more than interpretation’ is needed: a comment on the paper by the Process of Change Study Group. Interna- tional Journal of Psychoanalysis, 84: 109-18. Sandler, J., Holder, A., Kawenoka, M., Kennedy, H.E., Neurath, L. (1969). Notes on some

90 theoretical and clinical aspects of transference. International Journal of Psychoa- nalysis, 50: 633-645. Schafer, R. (2005). Caring and coercive aspects of the psychoanalytic situation. Journal of the American Psychoanalytic Association, 53: 771-87. Siegel, D.J. (1999). The developing mind. Toward a neurobiology of interpersonal experience. New York: The Guilford Press. Stern, D.N. (1985). The interpersonal world of the infant: a view from psychoanalysis and developmental psychology. New York: Basic Books. Stern, D.N. (2004). The present moment in psychotherapy and everyday life. New York: Norton & Company. Stern, D.N., Sander, L.W., Nahum, J.P., Harrison, A.M., et al. (1998). Non-interpretive mechanisms in psychoanalytic therapy. The “something more” than interpretation (The Boston Change Process Study Group, Report No. 1) International Journal of Psychoanalysis, 79: 903- 21. Strachey, J. (1934). The nature of the therapeutic action of psycho-analysis. Interna- tional Journal of Psychoanalysis, 15: 127- 59.

Chapter 6 Short Psychodynamic Supportive Psychotherapy, Antidepressants and their Combination in the Treatment of Major Depression: a mega-analysis based on three randomized clinical trials

Saskia de Maat, Jack Dekker, Robert Schoevers, Gerda van Aalst, Cecile Gijsbers-van Wijk, Marielle Hendriksen, Simone Kool, Jaap Peen, Rien Van, Frans de Jonghe Depression and Anxiety 2007 (in press)

Abstract

The efficacy of Short Psychodynamic Supportive Psychotherapy (SPSP) has not yet been compared to pharmacotherapy. A mega-analysis based on three original RCT’s was performed. Patients with (mild to moderate) major depressive disorder were randomized in (24 weeks) SPSP (n=97), pharmacotherapy (n=45) or their combination (n=171). Efficacy was assessed by the Hamilton Depres- sion Rating Scale, Clinical Global Impression of Severity and of Improvement, the Ninety Symptom Checklist (depression sub-scale) and the Quality of Life Depression Scale. Pearson chi-square calculations were used to compare success rates. ANCOVA analyses were used to test inter-group differences. success rates indicated that independent observers (HDRS) found no differences in symptom reduction between SPSP and pharmacotherapy (p=0.214), but therapists (CGI-S, p=0.026), and patients (SCL, p=0.036) favoured SPSP. Combined therapy was found superior to pharmacotherapy by all three (patients (p=0.000), therapists (p=0.024), independent observers (p=0.024). Inde- pendent observers (p=0.062) and therapists (p=0.430) found no differences between combined therapy and SPSP, but patients (p=0.016) found combined therapy superior. As far as quality of life is concerned, success rates indicated that patients (QLDS) found no differences between SPSP and pharmacotherapy (p= 0.073) or between SPSP and combined therapy (p=0.217). However, they found combined therapy superior to pharmacotherapy (p=0.015). The results of the mega-analysis suggest that combined therapy is more efficacious than pharmacotherapy. SPSP and pharmacotherapy seem equally efficacious, except for some indications that patients and therapists favour SPSP for symptom reduction. Combined therapy and SPSP also seem equally efficacious, except that patients think that the first is better in symptom reduction.

93 6.1 Introduction

Short-term Psychoanalytic Supportive Psychotherapy (SPSP, De Jonghe, 2005) is a treatment modality developed, from 1992 on, in the context of the Depres- sion Research Project of JellinekMentrum in Amsterdam. SPSP is a face-to- face, individual form of psychotherapy. It is a six-month cure, consisting of sixteen sessions (first eight weekly sessions, then eight biweekly sessions). It is rooted in psychoanalytic theory of normality, pathology, and of technique. In SPSP, the primary goal is to cure depression. The secondary goal is to reduce patient’s vulnerability to depression. The SPSP therapist is expected to apply a primarily supportive technique. The SPSP therapist does not shun interpretation, even transference interpretation if it is possible and useful, yet, its technique is largely situated on the supportive side of the interpretive-supportive continuum. Until now, three Randomized Clinical Trials (RCT’s) were conducted to test the relative efficacy of Short Psychodynamic Supportive Psychother- apy (SPSP) in the treatment of mild to moderate depression in ambulatory psychiatric patients. These trials provided the data for this mega-analysis. Combined therapy (SPSP plus antidepressants) was compared with pharmacotherapy alone (De Jonghe et al, 2001) and with SPSP alone (De Jonghe et al., 2004), and two intensities of combined therapy (medication plus 8 or 16 sessions of SPSP) were compared (Dekker et al. 2005). Combined therapy was clearly found more efficacious than pharmacotherapy alone on all outcome measures. The results of the comparison of combined therapy with SPSP were equivocal: neither therapists nor observers found differences, but patients found combined therapy superior in symptom reduction. Never did SPSP perform better than combined therapy. Finally, there were no significant differences found between the two intensities of combined therapy. spsp has not yet been compared with pharmacotherapy alone in a head-to- head comparison. In addition, the effect of SPSP on quality of life has not been investigated or compared with that of the other two treatment modalities. the aim of this paper is to compare, by pooling the original data of the three trials mentioned above, SPSP with pharmacotherapy alone and with combined therapy. The comparisons cover the effects on both symptom reduction and improvements in quality of life. the technique used is the mega-analysis of individual patient data. This is distinct from meta-analysis, which is performed on the outcome of trials (N= the numbers of studies reviewed). Mega-analysis allows for more powerful statistical analyses and the comparison of more interventions. Homogeneity in treatment protocols, patient samples and research methods is particularly important in mega-analysis (Thase et al, 1997). Since the same research group performed the three trials in question, and used an identical research design for similar study populations, high levels of clinical and methodological homogeneity can be expected. A limitation of mega-analysis is that some of the comparisons do not result from direct randomization of the subjects.

94 Based on the results of the original trials we formulated three hypotheses for this mega-analysis regarding efficacy in mild to moderate depressive disorder. As combined therapy clearly outperformed pharmacotherapy in the original trial, the first hypothesis is: Combined therapy is more efficacious than pharmacotherapy. As SPSP never performed better than combined therapy and the latter was superior to SPSP on some measures, the second hypothesis is: Combined therapy is more efficacious than SPSP. As the difference between combined therapy and pharmacotherapy was clearly larger than that between combined therapy and SPSP, the third hypothesis is: SPSP is more efficacious than pharmacotherapy.

6.2 Method of the three original trials

The following is a shortened version of the method section of the original trials. Details about patient selection, measurements and treatment protocols can be found in the publications relating to the original trials (De Jonghe et al., 2001, De Jonghe et al., 2004, and Dekker et al., 2005).

Study samples The samples of the three RCTs consisted of consecutively newly registered depressed patients at two outpatient clinics in Amsterdam. General practitioners referred the patients. In addition to written informed consent, inclusion criteria were: age between 18 and 65 years and a DSM-III-R (APA, 1980) or DSM-IV (APA, 1994) defined Major Depression with or without dysthymia. The DSM diagnosis was assessed by means of a semi-structured interview (Huyser et al., 1996). A further inclusion criterion was a 17-item Hamilton Depression Rat- ing Scale (HDRS, Hamilton, 1967) baseline score of 14 points or more in two trials, and of 12-24 points in the trial comparing psychotherapy to combined therapy. In this last trial, patients with severe depression (HDRS score 25 or more points) were excluded because it was not considered ethical to withhold severe depressed patients medication. The aim was to select for this trial patients presenting mild to moderate major depression. We defined mild and moderate depression as specified by a HDRS score between 12 and 18 points, and between 18 and 24 points respectively. The exclusion criteria were: presenting a psycho-organic disorder, drug abuse, a psychotic disorder and/or a dissociative disorder, not reliable enough to participate in a clinical trial (e.g. doctor shopping), serious communication problem (e.g., language barrier), physical restrictions (e.g., the patient will soon leave the country), being “too ill” and/ or “too suicidal” (e.g. hospitalisation is unavoidable), pregnancy or a wish to become pregnant. Non-response to an established psychotherapy was not an exclusion, nor was treatment resistance in prior depressive episodes. Exclusion criteria associated with medication were: a contra-indication for one of the antidepressants prescribed by the pharmacotherapy protocol, a history of adequate treatment with antidepressants during the present depressive episode, and use of psychotropic medication not prescribed by the pharmacotherapy protocol.

95 All patients were allocated (through a randomized and concealed allocation procedure) to one of the treatment conditions.

Treatment All treatments were intended to last for 6 months. Short Psychodynamic Supportive Psychotherapy (SPSP) is based on the principles described by, among others, de Jonghe et al. (1994). Approximately half of the therapists were experienced psychotherapists with at least 4 years of experience, half of them were residents with at least 2 years of experience in giving psychotherapy and working with (depressed) psychiatric patients. Residents required weekly group supervision of 90 minutes by an experienced psychoanalytic therapist (two of the authors, F. de Jonghe and H. Van), where tape recorded sessions were discussed. F. de Jonghe formulated the guidelines of SPSP. The ratio of experienced therapists and residents was comparable in the three trials. We did not explicitly investigate the possible effects of experience on outcome. the protocol required from the psychotherapists that they would, at each interview, consider alternative treatments if the patients deteriorated or did not benefit from SPSP. In these cases the patient was considered a drop-out and alternative treatment according to best medical practice was to be applied. Pharmacotherapy was provided in accordance with an antidepressant medication protocol allowing for changes in medication in response to inefficacy or intolerance. In two trials first a SSRI (fluoxetine) was given before switching to a TCA (nortriptyline), and finally an RIMA (moclobemide). In the trial comparing psychotherapy to combined therapy the order was first an SNRI (venlafaxine XR), then an SSRI (fluoxetine or, as a second option, fluvoxamine), then a TCA (nortriptyline), and finally lithium alongside nortriptyline. All patients were treated by a psychiatrist or by an advanced, supervised resident in psychiatry. The fifteen-minute medication consultations took place once every two weeks during the first two months of treatment, and once a month thereafter. The intended medication period was 6 months. Over the course of the 24-weeks trial, 64% of the completers in the pharmacotherapy condition stayed on the first choice of antidepressant, 31% of them changed from first choice to the second choice of medication and in 5% of the cases there were more than two steps necessary. In the pharmacotherapy conditions of the combined therapies taken together, these percentages were 76%, 20% and 4% respectively. The exact protocol detailing the types of medication and their dosages are mentioned in the publications of the earlier trials. Combined therapy consisted of treatment with SPSP and pharmacotherapy simultaneously for a period of six months. Different therapists provided the psychotherapy and pharmacotherapy.

6.3 Flow of participants and design

Table 6.1 shows the design of the three original trials.

96 Table 6.1: Design and flow of participants. Combined therapy Combined therapy Short versus very short versus pharmacotherapy versus psychotherapy combined therapy (de Jonghe et al., 2001) (de Jonghe et al., 2004) (Dekker et al., 2004) Assessed for eligibility 525 824 463 Excluded 358 616 360 Not meeting inclusion criteria 113 158 80 Meeting exclusion criteria 213 433 251 Refused 32 25 29 Randomized 167 208 103 Treatment Pharmaco- Combined Psycho- Combined Combined Combined conditions therapy therapy therapy of therapy therapy therapy including consisting including including including 16 SPSP 16 SPSP 16 SPSP 8 SPSP 16 SPSP sessions sessions sessions sessions sessions Randomized (ITT sample) 84 83 107 101 49 54 Started treatment (Per Protocol Sample) 57 72 106 85 45 45 Per protocol sample mega- analysis * 45 62 97 73 Not included 36 * excluded HDRS scores <14 and >24

6.4 Assessment of efficacy

The assessment of efficacy in each trial was based on data drawn from three sources: the patients, the treating therapists and independent observers. The independent observers (main criterion) used the 17-item HDRS (Hamilton, 1967). This is a scale of 17 items that assesses the severity of depression. It regards questions about symptoms such as sleep, feelings of anxiety, feelings of guilt, eating and weight, hypochondrias and suicidal ideations. In a recent review of Bagby et al. (2004) the inter-rater reliability (Pearson´s r) ranged between 0.82 to 0.98. The independent observers were university students who almost completed their masters degree in Psychology. They were selected in an application procedure by J. Dekker. Before administering the HDRS to patients the observers were trained by F. de Jonghe, who has developed a manual for a semi-structured interview for the HDRS (de Jonghe, 1994). During the studies

97 the independent observers discussed their audiotaped assessments monthly with F. de Jonghe. Inter-rater reliability for the independent observers was not statistically assessed. the therapists used the Clinical Global Impression of Severity (CGI-S) and Improvement (CGI-I) (Guy, 1976). This assessment instrument uses a 7-step Likert Scale to assess how depressed a patient is (from normal (1) to extremely ill (7) on the CGI-S) and how much a patient did improve (from very much improved (1) to very much worse (7) on the CGI-I). The CGI is widely used in clinical research (i.e. Zaider et al., 2003, Schneider & Olin, 1996). In the combined therapy condition the CGI was provided by the treating psychiatrist. the patients used the depression subscale of the Ninety Symptom Check- list (SCL-90, Arrindell & Ettema, 1986.). The SCL-90 is a self-report questionnaire that consists of 90 items measuring a broad range of psychopathological symptoms, clustered in 8 sub-scales. The sub-scales are Agoraphobia (7 items), Anxiety (10 items), Depression (16 items), Somatic complaints (12 items), Insufficiency in thinking and acting (9 items), Hostility (6 items), Sus- picion and interpersonal sensitivity (18 items) and Sleep problems (3 items). The remaining 9 items are miscellaneous items. Illustrative SCL-90 items are ‘feeling anxious’ and ‘having a feeling of emptiness’. Items are rated on a 5- point Likert Scale (1=not at all and 5=very much). The total score is seen as a general index of self-reported psychopathology. Patients also used the Quality of Life Depression Scale (QLDS, Tuynman-Qua and de Jonghe, 1992). This questionnaire contains 34 items that patients have to answer with ‘true’ or ‘not true’. Examples of items are: ‘I just want time to pass’, ‘I feel hopeful about the future’, ‘I take good care of myself’, ‘I can’t be bothered with my friends’, ‘I enjoy my food’ and ‘My life has no meaning’. The test-retest correlations of this scale are 0.94 in the United Kingdom and 0.87 the Netherlands. Internal consistency alpha-coefficients are 0.95 and 0.92, respectively. The QLDS was shown to correlate relatively highly with established measures of well-being, and scores obtained with this scale appeared related to severity of depression as assessed by the HDRS. The scale has wide applicability and has been shown to be user-friendly, both for respondents and administrators (Tuynman-Qua and de Jonghe, 1992).

6.5 Mega-analysis method

Study sample Some patients were excluded from the mega-analysis. Firstly, the condition consisting of combined therapy with eight sessions of SPSP, from the study examining a dose-response relationship (Dekker et al., 2005), was excluded because we considered it too different from the other combined therapy conditions (16 SPSP sessions). Even though in the original trial no efficacy differences were found between the eight and sixteen session condition, we excluded the first condition in order to enhance the clinical homogeneity of the mega- analysis sample. Secondly, the trials differed with respect to the required 17-

98 item HDRS baseline score: 14 points or more in two trials, and 12-24 points in the trial comparing psychotherapy and combined therapy. In order to enhance homogeneity, only patients with HDRS baseline scores of 14-24 were included in the mega-analysis; patients with HDRS scores < 14 and > 24 were excluded. This led to the exclusion of 12 cases in the pharmacotherapy condition, 9 in the SPSP condition and 36 in the combined therapy conditions. We realized that this is a considerable loss of patients, especially in the pharmacotherapy condition. In order to evaluate the influence of this loss on our results, we repeated all analyses including also these patients from the three original trials.The mega- analysis sample therefore consisted of 313 patients: 45 in pharmacotherapy, 97 in SPSP and 171 in combined therapy (see table 6.1).

Outcome measures Relative efficacy is expressed in this study in two different ways: inter-group differences in remission rates or response rates, and in mean scores. Assess- ments reflect the points of view of the independent observers (HDRS, main criterion), therapists (CGI-S and CGI-I), and of the patients (SCL-D and QLDS). Remission is defined (as in the original trials) as a final HDRS score of 7 points or less, as a CGI-S or CGI-I score of 2 points or less, and as an improvement of at least 1 standard deviation from base rate on the SCL-D depression subscale and the QLDS. Response is defined as a 50% reduction in the HDRS baseline score. Differences in dropout rates were calculated.

Statistical analyses Pearson chi-square calculations (two sided, level of significance .05) were used to compare base rates and dropout rates. Pearson chi-square calculations (one sided, level of significance .05) were used to compare response rates and remission rates (Altman, 1991, Bland & Bland, 1994). ANCOVA analyses (including baseline scores for outcome measures as covariates) were used to test inter- group differences. Our results are calculated in all patients who started with the treatment they were allotted to (a Per-Protocol or modified Intention-to-treat sample). The Last Observation Carried Forward approach was used.

6.6 Results of the mega-analysis

Study sample A total of 313 patients were included in our mega-analysis (45 in pharmacotherapy, 97 in SPSP and 171 in combined therapy). Table 6.2 presents the characteristics of the study sample. there were four significant differences between the treatment conditions. The mean CGI-severity baseline score in the pharmacotherapy condition was somewhat higher than in the SPSP and combined condition (p=0.01 and p=0.04 respectively). Furthermore, the SCL-D-score in pharmacotherapy group was somewhat lower than in the SPSP-group (p= 0.03) and the QLDS score on the combined group was higher than in the SPSP group (p=0.019). The three

99 Table 6.2: Characteristics of the Per protocol sample. Pharmaco- Psycho- Combined therapy therapy therapy 1 vs 2 1 vs 3 2 vs 3 (n=45) (n=97) (n=171) Sex Male 35,6 33,0 32,2 Female 64,4 67,0 67,8 Age 19-29 year 40,0 33,0 35,7 30-39 year 33,3 34,0 35,7 >40 year 26,7 33,0 28,7 Educational level Low 13,3 15,1 15,4 Middle 35,6 34,4 37,3 High 51,1 50,5 47,3 Living situation Alone 46,7 45,7 41,8 Living with at least one person 53,3 54,3 58,2 Marital status Married 20,0 20,2 20,6 Divorced/widowed 13,3 7,4 11,2 Not married 66,7 72,3 68,2 Job status Sick/social benefit/other 53,3 53,8 55,3 Job 46,7 46,2 44,7 Duration present episode < 1 year 65,9 74,2 70,8 1-2 year 13,6 12,4 16,1 > 2 year 20,5 13,5 13,0 Medication 3 months Yes 13,3 27,8 22,8 before study No 86,7 72,2 77,2 Depressed episodes 0 47,7 69,0 61,7 within previous 5 yrs 1 25,0 17,2 22,8 >= 2 27,3 13,8 15,4 Psychiatric treatment Yes 43,2 30,3 38,3 during present episode No 56,8 69,7 61,7 HDRS Mean 19,0 18,7 18,7 SD 2,8 3,0 3,0 CGI-severity Mean 4,8 4,4 4,5 .011 .041 SD 0,6 0,7 0,7 SCL-90 depression Mean 46,4 50,3 48,5 .030 subscale SD 11,2 8,8 9,4 SD 8,8 7,7 7,5 QLDS Mean 16,6 15,8 17,8 .019 SD 7,4 6,3 6,8 Personality disorder No 33,3 30,8 34,6 Yes 66,7 69,2 65,4

100 samples did not differ statistically significant with regard to the presence of dysthymia. Although the absolute differences were probably clinically insignificant, the baseline scores for the outcome measure were included as covariates in the ANCOVA analysis of that outcome. Mean HDRS score in the first trial was 20.43, in the second trial 19.90 and in the third trial 18.07. The mean HDRS for the total sample of the mega-analysis was 18,7.

Dropout Table 6.3 presents numbers and rates for dropout. as can be seen in table 6.3, the dropout in pharmacotherapy was 38%, in combined therapy it was 28% and in SPSP it was 25%. The dropout rates for the three conditions did not differ significantly. Dropout of both SPSP and pharmacotherapy was lower in the combined therapy condition than in monotherapy conditions (10% versus 25% for SPSP and 25% versus 38% for pharmacotherapy). The difference regarding SPSP was significant (p=0.001). for reasons explained above (method section), we repeated the analyses including patients with HDRS scores < 14 and > 24 from the three trials. With regard to drop-out rates we found the results to be comparable, except for one finding: A formerly not found significant difference between SPSP and pharmacotherapy was detected (in the new analysis respectively 25,5% and 40,4%, p=0.049).

Remission rates and response rates The remission rates and response rates are shown in table 6.4. no significant differences in HDRS remission and response, CGI-I and QLDS, were found between SPSP and pharmacotherapy. CGI-S and SCL-D indicated a significant difference in favour of SPSP. All outcomes (except for HDRS-response) indicated significantly higher remission rates in combined therapy than in pharmacotherapy. No differences were found between combined

Table 6.3: Dropout rates. Pharmaco- Psycho- Combined therapy therapy therapy 1 vs 2 1 vs 3 2 vs 3 N % N % N % Pearson p Pearson p Pearson p Chi2 Chi2 Chi2 Pharmaco- therapy dropout 17 37,8% - - 43 25,1% 2,83 0,092 Psycho- therapy dropout - - 24 24,7% 17 9,9% 10,46 0,001 Total dropout 17 37,8% 24 24,7% 48 28,1% 2,54 0,111 1,60 0,206 0,35 0,555

101 Table 6.4: Remission and response rates at treatment termination (week 24). Pharmaco- Psycho- Combined therapy therapy therapy 1 vs 2 1 vs 3 2 vs 3 % N % N % N Pearson One Pearson One Pearson One Chi2 sided P Chi2 sided P Chi2 sided P HDRS remission 24,4 45 30,9 97 40,4 171 0,63 0,214 3,87 0,024 2,36 0,062 HDRS response 35,6 45 40,2 97 48,0 171 0,28 0,298 2,21 0,068 1,50 0,110 CGI-Severity success 48,9 45 66,0 97 64,9 171 3,76 0,026 3,87 0,024 0,03 0,430 CGI- Improvement success 60,0 45 71,0 93 73,2 168 1,66 0,099 2,98 0,042 0,15 0,348 SCL- depression success 44,4 45 60,6 94 73,4 169 3,23 0,036 13,51 0,000 4,57 0,016 QLDS 40,0 45 53,1 96 58,1 167 2,11 0,073 4,67 0,015 0,61 0,217

Table 6.5: Mean scores for the outcome measures at treatment termination (week 24). Pharmacotherapy Psychotherapy Combined therapy 1 vs 2 1 vs 3 2 vs 3 Mean SD N Mean SD N Mean SD N F One F One F One sided P sided P sided P HDRS 12,82 7,79 45 11,45 7,13 97 10,28 6,86 171 0,84 0,180 4,22 0,020 1,86 0,087 CGI- Severity 2,73 1,29 45 2,18 1,30 97 2,19 1,31 171 6,02 0,007 5,56 0,009 0,23 0,317 CGI- Improve- ment 2,51 1,24 45 2,09 1,27 93 1,99 1,07 168 3,41 0,033 7,48 0,003 0,39 0,265 SCL- depression 35,76 13,16 45 36,95 13,91 97 31,38 11,81 170 0,15 0,349 7,58 0,003 9,87 0,001 QLDS 22,8 8,79 45 23,16 8,63 97 25,95 7,60 170 0,84 0,180 6,33 0,006 3,48 0,031

102 therapy and SPSP, except for the SCL-D favouring combined therapy over SPSP. For reasons explained above (method section), we repeated the analyses including patients with HDRS scores < 14 and > 24 from the three trials. With regard to remission and response rates we found the results to be comparable, except for two findings: 1.A formerly found trend in the QLDS success rates comparing SPSP and pharmacotherapy became a significant difference favouring SPSP (in the new analysis respectively 53,3% versus 37,5%, one sided p=0.027). 2.A formerly found trend in the HDRS response rates became a statistically significant difference favouring combined therapy above pharmacotherapy (in the new analysis respectively 33,3% versus 47%, one sided p=0.033).

Differences in mean scores Table 6.5 presents the mean scores for depression on the HDRS, CGI-Severity, CGI-Improvement, SCL-D and QLDS. all outcome measures indicated that combined therapy was superior to pharmacotherapy. No significant differences were found between SPSP and pharmacotherapy, except for two (CGI-S and CGI-I) favouring SPSP. All outcomes, except two (SCL-D and QLDS, favouring combined therapy), showed no significant differences between combined therapy and SPSP. for reasons explained above (method section), we repeated the analyses including patients with HDRS scores < 14 and > 24 from the three trials. With regard to differences in mean scores we found the results to be comparable, except for one finding: A formerly not found significance on the QLDS was now found to be a significant difference, favouring SPSP over pharmacotherapy (in the new analysis respectively 23,43 versus 20,89 points, one sided p=0.047).

6.7 Discussion

In this mega-analysis we pooled the original data of three RCT’s published earlier. This allowed for three head-to-head comparisons: SPSP and pharmacotherapy (a comparison that was not included in the earlier publications), SPSP and combined therapy and pharmacotherapy and combined therapy. In addition, we assessed the effect of the treatment modalities on improvements in quality of life. The same research group conducted the RCT’s in similar populations using an identical research design, thereby ensuring clinical and methodological homogeneity.

SPSP and pharmacotherapy Turning to symptom reduction, the main outcome of the mega-analysis (HDRS) suggests that SPSP and pharmacotherapy are equally efficacious in the treatment of mild to moderate depression. These findings are in line with the literature about comparisons between cognitive psychotherapy and antidepressants (e.g. Keller et al., 2000, Jarrett et al., 1999, Blackburn & Moore, 1997), and between IPT and antidepressants (DiMascio et al., 1979, Elkin et al., 1989). Similar conclusions were reached in a recent meta-analysis comparing psy-

103 chotherapy with pharmacotherapy (De Maat et al.,2006). However, secondary outcome measures show that there are indications that the clinicians (CGI-S) found SPSP superior to pharmacotherapy, as did the patients with regard to symptom reduction (SCL-D). It seems that more research is necessary to reach final conclusions in this area. We found no differences between the two treatment modalities in terms of improvements in quality of life. We did not find any data in the literature for comparison with our findings.

Combined therapy and pharmacotherapy Patients, therapists and independent observers alike find combined therapy more efficacious than pharmacotherapy alone, both for symptom reduction and improving quality of life. These results are in line with the results of the original trial. As far as symptom reduction is concerned, our findings concur with the results of Pampallona et al. (2004) who report an Odds Ratio of 1.86 in favour of combined therapy compared with drug treatment alone in the treatment of depression.

Combined therapy and SPSP With respect to symptom reduction, independent observers and therapists did not discern a difference between combined therapy and SPSP. Patients, on the contrary, favoured combined therapy over SPSP. These results are again in line with the results of the original trial. Apparently, the patients, or the scale they use, register subtle differences that go unnoticed by therapists and independent observers. In the literature, the relative merits of combined therapy and psychotherapy are still a matter of debate. Some reviewers find no differences (Conte et al., 1986, Wexler & Cicchetti, 1992, Blom et al., 2000) others find combined therapy to be superior (Friedman et al., 2004, Segal, 2002). Our results are mostly in line with the former. that there were generally no differences found in outcome between this conditions may be due to the fact that our population was mildly to moderately depressed. In the mega-analysis of Thase et al. (1997), the authors did find that combined therapy was superior to psychotherapy in more severe depression (HDRS>20), but not in less severe cases. Differences in symptom reduction (according to the patients) had no clear correlation with differences in improvements of quality of life. The difference measured in QLDS remission rates was not, but in QLDS mean end score was significant (favouring combined therapy). Of course, it is the question whether a difference of 3.15 points on the QLDS scale is clinically significant according to the patients.

Dropout rates Dropout rates in SPSP (25%) were similar to psychotherapy dropout rates in other studies, varying from 16% (DeRubeis et al., 2005) to 44% (Hollon et al., 1992). In pharmacotherapy, at week 24, 38% of the patients had prematurely stopped taking medication. In the original trial (De Jonghe et al., 2001), the pharmacotherapy dropout rate at week 8 was only 5%. The increase in dropout

104 of pharmacotherapy during the treatment period had various reasons, one of them being adverse effects, even up to week 24. Some of the reasons in the later stages of the treatment were also ‘not feeling better’ and, perhaps somewhat surprisingly, ‘feeling better’. A publication regarding the detailed analyses of dropout in the pharmacotherapy conditions and the corresponding reasons is being prepared. (Aalst, van, G. et al., in preparation). Pharmacotherapy compliance in the first 8 weeks of our study can be considered rather fair. Recent reviews (Linden et al., 2000, Anderson, 2000) mention dropout rates of 33% to 48% in pharmacotherapy conditions within the first 6-8 weeks. the dropout rates in the combined therapy condition in our study were 28%. Combining SPSP with medication reduced the dropout rates for both treatment components (in SPSP from 25% to 10% and in pharmacotherapy from 38% to 25%). in our study the overall dropout rates did not differ between the three treatment modalities. This is in accordance with findings in literature. Casacalenda et al. (2002) report a dropout rate for psychotherapy of 22% and of 37% for medication, the difference not being statistically significant. A review by Fried- man et al. (2004) shows no significant differences between dropout rates of combined therapy versus psychotherapy. Finally, Pampallona et al. (2004) conclude that the dropout rates for combined therapy and pharmacotherapy do not differ.

Limitations Our study has several limitations. Firstly, the study sample suffers from the selection bias inherent to all RCT’s. Most trials exclude patients with, for example, serious comorbidity, suicidal intentions or substance abuse. Patients meeting all inclusion and no exclusion criteria applied in RCT’s represent only a small proportion of the depressed target population. (In our studies, 26% of the assessed patients were finally included.) A cautious approach is therefore required to the generalization of conclusions from RCT’s. Second, there are some differences at baseline between the study samples of the three original trials. Since the baseline scores for outcome measures were included in analyses as covariates, these differences have been accounted for. Thirdly, the pharmacotherapy protocols in the three original trials differed slightly (the SPSP manual, on the other hand, was identical). We did not, however, find significant differences between the mean scores or remission rates of combined therapy in the three samples of the original RCT’s. Furthermore, meta-analytic studies have shown that there are no clinically significant differences between antidepressants in terms of treating depressed outpatients (Anderson, 2000). Fourthly, we are well aware that the comparison of SPSP and pharmacotherapy is not based on the randomization between these two treatments. This may have affected the results. However, we believe both groups are comparable as the studies were performed in the same outpatient facilities with the same study design and within the same population. Furthermore, we tested for pre-treatment differences and performed covariance analyses. Fifthly, in a secondary analysis, we repeated all analyses including patients with HDRS scores of < 14

105 and > 24 from the three original trials in order to evaluate the influence on our results of the loss patients due to restricting our main calculations to patients with HDRS scores 14-24. The findings were in line with those of our original analyses. However, dropout rates and quality of life favoured SPSP above pharmacotherapy. This we consider an indication that in, an broader range of HDRS scores, SPSP, on some points, may outperform pharmacotherapy. Sixthly, as there are many statistical comparisons in this mega-analysis, it could be argued that a Bonferroni correction is needed to prevent type I statistical errors. To this end a significance level of for instance 0.016 (0.5/3 as we had three sources of information) might be taken as threshold. If this correction is applied, 8 of the 17 significant differences disappear, four of them concerning the differences in success rates between combined therapy and pharmacotherapy, three concerning the comparison of SPSP and pharmacotherapy. This means all the more that our results must be interpreted with caution. A final limitation is that, from sheer necessity (we did not have follow-up data at our disposal) we restrict our mega-analysis to the short-term effects of time-limited treatments.

Strengths Firstly, in our opinion the strong clinical and methodological homogeneity of this mega-analysis adds to its strength. Secondly, our data come from three sources (independent observers, patients and therapists). There are two reasons why we consider this to be a strength. Evidence-based medicine values the opinions of patients and clinicians highly, alongside those of independent observers. In addition, a recent systematic review by Bagby et al. (2004) shows that the Hamilton depression scale, which is widely used in depression research, has considerable psychometrical and conceptual flaws. A final strength of our mega-analysis is the inclusion of the effect of treatment on quality of life, which is, after all, the ultimate goal of the treatment.

Conclusions In the treatment of outpatients with mild to moderate major depressive disorder the results of the mega-analysis suggest that independent observers, patients and therapists alike found SPSP plus pharmacotherapy to be more efficacious than pharmacotherapy alone, for both symptom reduction and improving quality of life. The results also suggest that independent observers found SPSP and pharmacotherapy equally efficacious. There is some indication that patients and therapists favour SPSP over pharmacotherapy in regard to symptom reduction, but, then again, not in improving quality of life. No efficacy difference was found between SPSP and combined therapy, except that patients thought that combined therapy was more efficacious in terms of symptom reduction.

References

Aalst, van, G., Schoevers, R., Jonghe, de, F., Jonghe, de, A., Peen, J., Dekker, J. Pre- mature termination of antidepressant medication in monotherapy and combined therapy for Major Depression. In preparation.

106 Altman, D.G. (1991). Practical Statistics for medical research. Chapman & Hall, Lon- don. American Psychiatric Association (1980) Diagnostic and Statistical Manual of Mental Disorders (3rd edn) (DSM-III). Washington DC: APA. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders. Fourth edition (DSM-IV). Washington: APA. Anderson, I.M. (2000) Selective serotonin reuptake inhibitors versus tricyclic antidepressants: a meta-analysis of efficacy and tolerability. Journal of Affective Disorders, 58, 19-36. Arrindell WA, Ettema JMM. SCL-D-90 (1986) Handleiding bij een multidimensionele psychopathologie-indicator. Lisse: Swets & Zeitlinger. Bagby RM, Ryder AG, Schuller DR, Marshall MB. (2004) The Hamilton Depression Rating Scale: Has the Gold Standard Become a Lead Weight? American Journal of Psychiatry; 161, 2163-2177. Blackburn, I.M., & Moore, R.G. (1997). Controlled acute and follow-up trial of cognitive therapy and pharmacotherapy in out-patients with recurrent depression. British Journal of Psychiatry, 171, 328-334. Bland, J.M. & Bland, D.G. (1994) Statistics Notes: One and two sided tests of sifnifi- cance. British Medical Journal, 309: 248. Blom, M.B.J., Jonker, K, Haffmans, P.M.J., et al (2000) Combinatie van medicatie en psychotherapie bij depressie: een overzicht. Tijdschrift voor Psychiatrie, 42, 99-407. Casacalenda, N., Perry, J.C., Looper, K. (2002) Remission in Major Depressive Disorder: A comparison of pharmacotherapy, psychotherapy, and control conditions. Ameri- can Journal of Psychiatry, 159: 1354-1360. Conte, H.R., Plutchik, R., Wild, K., et al (1986) Combined psychotherapy and pharmacotherapy for depression. Archives of General Psychiatry, 43, 471-479. Dekker J, Molenaar PJ, Kool MA, van Aalst G, Peen J, de Jonghe F. Dose-effect relations in time-limited combined psycho-pharmacological treatment for depression. Psychological Medicine, 35(1): 47-58. DeRubeis, R.J., Hollon, S.D., Amsterdam, J.D., Shelton, R.C., Young P.R., Salomon, R.M., O’Reardon, J.P., Lovett, M.L., Gladis, M.M., Brown, L.L., Gallop, R. (2005). Cognitive therapy vs medication in the treatment of moderate to severe depression. Archives of General Psychiatry, 62: 409-416. Dimascio A, Weissman MM, Prusoff BA, Neu C, Zwilling M, Klerman GK. (1979) Dif- ferential Symptom Reduction by Drugs and Psychotherapy in Acute Depression. Archives of General Psychiatry. 36, 1450-1456. Dwight Johnson, M. (2000) Treatment preferences among depressed primary care patients. Journal of General International Medicine; 15.8, 527-534. Elkin, I., Shea, M.T., Watkins, J.T., Imber, S.D., Sotsky, S.M., Collins, J.F., Glass, D.R., Pilkonis, P.A., Leber, W.R., Docherty, J.P., Fiester, S.J., Parloff, M.B. (1989). National Institue of Mental Health Treatment of Depression Collaborative Research Program. Archives of General Psychiatry, 46, 971-982. Friedman, M.A., Detweiler-Bedell, J.B., Leventhal, H.E., et al (2004) Combined psychotherapy and pharmacotherapy for the treatment of major depressive disorder. Clinical Psychology; Science and Practice, 11, 47-68. Guy W. (1976). ECDEU Assessment Manual for Psychopharmacology. Revised DHEW Pub. (ADM). Rockville, MD: National Institute for Mental Health. Hamilton, M. (1967). Development of a rating scale for primary depressive illness. British Journal of Social and Clinical Psychology, 6, 278-296. Hollon, S.D., DeRubeis, R.J., Evans, M.D., Wiemer, M.J., Garvey, M.J., Grove, W.M., Tuason, V.B. (1992). Cognitive therapy and Pharmacotherapy for Depression. Archives of General Psychiatry, vol. 49, 774-781. Huyser, J., Jonghe, F. de, Jonkers, F., et al. (1996) The manual for the Diagnostics of Major Depression (MDMD): description and reliability. International Journal of Methods in Psychiatric Research, 6, 1-4.

107 Jarret, R.B., Schaffer, M., McIntire, D., Witt-Browder, A., Kraft, D., & Risser, R.C. (1999). Treatment of atypical depression with cognitive therapy of phenelzine. Archives of General Psychiatry, vol. 56, 431-437. Jonghe F, Rijnierse P, Janssen R. (1994). Psychoanalytic supportive psychotherapy. Journal of the American Psychoanalytic Association; 42, 421-446. Jonghe, F., de. (1994) Leidraad voor het scoren van de Hamilton Depression Rating Scale. Amsterdam: Benecke Consultants. Jonghe, F., de, Kool, S., Aalst, G., Dekker, J., Peen, J. (2001). Combining psychotherapy and antidepressants in the treatment of depression. Journal of Affective Disorders, 64, 217-229. Jonghe, F., de, Hendriksen, M., Aalst, G., et al. (2004) Psychotherapy and Combined Therapy (pharmacotherapy plus psychotherapy) in the treatment of depression. British Journal of Psychiatry, 185 (1), 37-45. Jonghe, F., de (2005) Kort en krachtig. Kortdurende Psychoanalytische Steungevende Psychotherapie. Benecke N.I. Amsterdam. Keller,M.B., McCullough, J.P., Klein D.N., Arnow, B., Dunner, D.L., Geleberg, A.J., Markowitz, J.C., Nemeroff, C.B., Russel, J.M., Thase, M.E., Treivedi, M.H., Zajecka, J. (2000). A comparison of Nefazodone, the Cognitive Behavioral-Analysis System of Psychotherapy and their combination for the treatment of chronic depression. The New England Journal of Medicine, 342, 1462-1470. Linden M, Gothe H, Dittmann RW, Schaaf B. (2000) Early termination of antidepressant drug treatment. Clinical Psychopharmacology, 20.5: 523-30. Maat, S., de, Dekker, J., Schoevers, R., Jonghe, F., de. (2006). Relative efficacy of psychotherapy and pharmacotherapy in the treatment of depression: a meta-analysis. Psychotherapy Research, 16(5): 562-572. Pampallona, S., Bollini, P., Tibaldi, G., Kupelnick, B., Munizza, C. (2004) Combined pharmacotherapy and psychological treatment for depression. A systematic review. Archives of General psychiatry; 61: 714-719. Segal, Z, Vincent, P., Levitt, A. (2002) Efficacy of combined, sequential and crossover psychotherapy and pharmacotherapy in improving outcomes in depression. Journal Psychiatry Neurosci, 27 (4), 281-290. Thase ME, Kupfer DJ. (1996) Recent developments in the pharmacotherapy of mood disorders. Journal of Consulting and Clinical Psychology; 64.4, 646-659. Thase ME, Greenhouse JB, Frank E, Reynolds III CF, Pilkonis PA, Hurley K, Grocho cinsky V, Kupfer DJ. (1997). Treatment of major depression with psychotherapy or psychotherapy-pharmacotherapy combinations. Archives of General Psychiatry; 54, 1009-1015. Tuynman-Qua, H., de Jonghe, F. (1992). Quality of Life Depressie Schaal. Ibero, Houten. Wexler, B.E., Cicchetti, D.V. (1992) The outpatient treatment of depression. Implica- tions of outcome research for clinical practice. The Journal of nervous en mental disease, 180, 277-286.

108 Chapter 7 Comparison of Short Psychodynamic Supportive Psychotherapy, Pharmacotherapy and their Combination on sub dimensions of the HDRS and sub scales of the SCL-90: a mega-analysis of three randomized controlled trials regarding depressed patients

Saskia de Maat, Robert Schoevers, Frans de Jonghe, Gerda van Aalst, Cecile Gijsbers-van Wijk, Marielle Hendriksen, Simone Kool, Jaap Peen, Rien Van, Jack Dekker Submitted

Abstract

Background: In an earlier mega-analysis (based on three RCT’s) we compared three treatments for depression: Short Psychodynamic Supportive Psychother- apy (SPSP), pharmacotherapy (Ph) and their combination (CT). We used the sum score of the HDRS and the SCL-90 Depression subscale. However, the HDRS is a multidimensional scale and the SCL-90 consists of eight subscales. We therefore may have lost important information. Aim: To compare the efficacy of three depression treatment modalities on possibly relevant factor levels of the HDRS and on three subscales of the SCL-90: Depression, Anxiety and Somatic complaints. Method: Factor analysis, based on the patient sample of the RCT’s, yielded two HDRS factors. The first factor was labelled ‘Mental’, loading on items like depressive mood, feelings of guilt, hypochondrias and anxiety. The second factor was labelled ‘Somatic’, loading on items like insomnia, gastrointestinal, general somatic complaints. Mega-analysis was applied in order to compare the efficacy of the three treatment modalities on the level of the two HDRS factors and of the three SCL subscales. Results: In regard to the ‘Mental’ HDRS factor SPSP and Ph did not differ (p=0,077) nor was there a difference found between CT and SPSP (p=0,280), but CT outperformed Ph (p=0,029). In regard tot the ‘Somatic’ HDRS factor SPSP and Ph did not differ (p=0,422) nor was there a difference between CT and Ph (p=0,053) but CT outperformed SPSP (p=0,014). In regard to the SCL Depression subscale SPSP and Ph yielded similar results (p=0,350), and CT was superior to both of them (resp. p=0,003 and p=0,001). In regard to the SCL-Anxiety subscale SPSP and CT were better than Ph (resp. p=0,003 and

109 p=0,000), while CT and SPSP yielded similar results (p=0,067). In regard to the SCL-Somatic complaints subscale no differences between treatments were found. Summary and conclusions: Our results suggest that, in regard to the ‘Mental’ HDRS factor, CT outperformed Ph while SPSP holds an intermediate position. In regard to the ‘Somatic’ HDRS factor CT outperformed SPSP while Ph holds an intermediate position. According to the SCL Depression subscale, CT was the more efficacious treatment. According to the SCL Anxiety subscale CT and SPSP were the more efficacious treatments. According tot the SCL Somatic complaints subscale the three treatment modalities were equally efficacious. it seems that CT is the first choice treatment according to the SCL Depres- sion subscale. As monotherapy is, other things being equal, to be preferred over combined therapy, it may tentatively be concluded that SPSP is the first choice treatment in regard to the HDRS ‘Mental’ factor and SCL Anxiety subscale. In regard to the HDRS ‘Somatic’ factor, it is Ph. As far as the SCL Somatic complaints subscale is concerned, it is SPSP or Ph. Looking at efficacy results at dimensional and subscale levels may yield valuable information about the working components of treatments.

7.1 Introduction

Most scales that are being used for measuring depression (such as de HDRS, but also the Beck Depression Inventory (BDI) and the SCL-90) present different symptom clusters discovered by factor analyses (Bagby, 2004; Arrindell & Ettema, 1986; Shafer, 2006). However, depression studies in general, including our own, use the overall scores of these scales. With a multidimensional scale this can present problems for both assessment and treatment. Shafer (2006) presents a clear overview of these problems. The author mentions that patients with different symptom profiles are likely to have different prognoses and may require different treatment. He presents the following arguments for this. Firstly, some studies found that specific depression symptom factors within a test have different patterns of correlations with external variables such as gender or race (Barefoot et al., 2000; Holm et al., 1994; Tashakkori et al., 1989, Williams&Richardson, 1993). Secondly, a number of studies have shown that different groups have different mean levels of symptomatology in terms of specific depression symptom factors (Devins et al., 1988; Endler et al., 1999; Kilgore, 1999; Ross&Mirosky, 1984; Yin et al., 2000; Zemore&Earmes, 1979). (We add that for the SCL-90 this is demonstrated in studies of Dolan et al., 1990; Dyer et al., 2005; Cornman, 1997; Feldman et al., 1983.) Thirdly, some studies have demonstrated that treatment can affect specific depression factors and not others (Muller&Benkert, 2001; Rickels et al., 1973). Fourthly, changes in one specific dimension could be missed by looking only at the overall score. Two meta-analyses of pharmacotherapy studies concluded that different one-dimensional subscales were more sensitive to change than

110 the overall HDRS score (Faries et al., 2000). In short, it can be informative to look at outcomes of treatment effects on a more specific dimensional level. in our own mega-analysis (De Maat et al., 2007), we also encountered this issue. In the mega-analysis we compared the relative effectiveness of three treatment modalities; Short Psychodynamic Supportive Psychotherapy (SPSP, De Jonghe, 2005), pharmacotherapy and their combination in the treatment of mild to moderate depression in ambulatory psychiatric patients (De Jonghe et al, 2001, De Jonghe et al., 2004, Dekker et al. 2005). Symptom reduction was evaluated by patients (using the depression sub scale of the SCL-90 (SCL- 90-D, Arrindell & Ettema, 1986), independent observers (using the Hamilton Depression Rating Scale (HDRS, Hamilton, 1967), and therapists (using the Clinical Global Impression scale (CGI, Guy, 1979). The results of the mega- analysis suggest that, overall, combined therapy is more efficacious than pharmacotherapy. SPSP and pharmacotherapy are equally efficacious according to the HDRS scores, but both patients (SCL-90-D) and therapists (CGI) appear to slightly favour SPSP for symptom reduction. Combined therapy and SPSP are also equally efficacious (HDRS), except that patients (SCL-90-D) estimate the first to be better in symptom reduction. in this mega-analysis we used the overall scores of the HDRS and one sub scale of the SCL-90. However, patients filled in two more sub scales of the SCL- 90 (Anxiety and Somatic Complaints) and we did not look at subdimensions of the HDRS. The aim of this article, applying the method of mega-analysis, is twofold. First we compare the effectiveness of SPSP, pharmacotherapy and combined therapy on the possible relevant factor levels of the HDRS. Secondly, we compare the three treatment modalities on the Depression, Anxiety and Somatic complaints sub scales of the SCL-90. as Shafer (2006) detected two common factors in four meta-analyses of four depression questionnaires (a Depression Severity factor and a small Somatic Symptoms factor), we expected to find at least two factors in the HDRS: a core depression factor (a more ‘psychological’ symptom cluster) and a somatic factor (a more ‘vegetative’ symptom cluster). Based on the findings of the original trials and following the often assumed but unsubstantiated hypothesis that antidepressants work predominantly on more vegetative aspects (such as sleep, appetite, tiredness, somatic complaints), and that psychotherapy has more influence on the core depression aspects (like feelings of guilt, depressed mood, hypochondriasis), we formulated the following hypotheses. Firstly, we predicted that SPSP would outperform pharmacotherapy on a core depression factor of the HDRS, SCL-Depression and SCL-Anxiety. Secondly, our hypothesis was that pharmacotherapy would yield better results than SPSP on a HDRS vegetative factor and SCL-Somatic complaints. Finally, our third hypothesis was that Combined therapy would perform better than both monotherapies.

111 7.2 Method of the three original RCT’s

7.2.1 Study samples

The samples of the three RCTs consisted of consecutively newly registered depressed patients at two outpatient clinics in Amsterdam. General practitioners referred the patients. In addition to written informed consent, inclusion criteria were: age between 18 and 65 years and a DSM-III-R (American Psychiatric Association, 1980) or DSM-IV (American Psychiatric Association, 1994) defined Major Depression with or without dysthymia. The DSM diagnosis was assessed by means of a semi-structured interview (Huyser et al., 1996). A further inclusion criterium was a 17-item Hamilton Depression Rating Scale (HDRS, Hamilton, 1967) baseline score of 14 points or more in two trials, and of 12-24 points in the trial comparing psychotherapy to combined therapy. In this last trial, patients with severe depression (HDRS score 25 or more points) were excluded because it was not considered ethical to withhold severe depressed patients medication. The aim was to select for this trial patients presenting mild to moderate major depression. We defined mild and moderate depression as specified by a HDRS score between 12 and 18 points, and between 18 and 24 points respectively. The exclusion criteria were: presenting a psycho-organic disorder, drug abuse, a psychotic disorder and/or a dissociative disorder, not reliable enough to participate in a clinical trial (e.g. doctor shopping), serious communication problem (e.g., language barrier), physical restrictions (e.g., the patient will soon leave the country), being “too ill” and/or “too suicidal” (e.g. hospitalisation is unavoidable), pregnancy or a wish to become pregnant. Non-response to an established psychotherapy was not an exclusion, nor was treatment resistance in prior depressive episodes. Exclusion criteria associated with medication were: a contra-indication for one of the antidepressants prescribed by the pharmacotherapy protocol, a history of adequate treatment with antidepressants during the present depressive episode, and use of psychotropic medication not prescribed by the pharmacotherapy protocol. All patients were allocated (through a randomized and concealed allocation procedure) to one of the treatment conditions.

7.2.2 Treatment

All treatments were intended to last for 6 months. short Psychodynamic Supportive Psychotherapy (SPSP) is based on the principles described by, among others, de Jonghe et al. (1994). Approximately half of the therapists were experienced psychotherapists with at least 4 years of experience, half of them were residents with at least 2 years of experience

112 in giving psychotherapy and working with (depressed) psychiatric patients. Residents required weekly group supervision of 90 minutes by an experienced psychoanalytic therapist (the last two authors), where tape recorded sessions were discussed. The last author formulated the guidelines of SPSP. The ratio of experienced therapists and residents was comparable in the three trials. We did not explicitly investigate the possible effects of experience on outcome. the protocol required from the psychotherapists that they would, at each interview, consider alternative treatments if the patients deteriorated or did not benefit from SPSP. In these cases the patient was considered a dropout and alternative treatment according to best medical practice was to be applied. pharmacotherapy was provided in accordance with an antidepressant medication protocol allowing for changes in medication in response to inefficacy or intolerance. In two trials first a SSRI (fluoxetine) was given before switching to a TCA (nortriptyline), and finally an RIMA (moclobemide). In the trial comparing psychotherapy to combined therapy the order was first an SNRI (venlafaxine XR), then an SSRI (fluoxetine or, as a second option, fluvoxamine), then a TCA (nortriptyline), and finally lithium alongside nortriptyline. All patients were treated by a psychiatrist or by an advanced, supervised resident in psychiatry. The fifteen-minute medication consultations took place once every two weeks during the first two months of treatment, and once a month thereafter. The intended medication period was 6 months. Over the course of the 24-weeks trial, 64% of the completers in the pharmacotherapy condition stayed on the first choice of antidepressant, 31% of them changed from first choice to the second choice of medication and in 5% of the cases there were more than two steps necessary. In the pharmacotherapy conditions of the combined therapies taken together, these percentages were 76%, 20% and 4% respectively. The exact protocol detailing the types of medication and their dosages are mentioned in the publications of the earlier trials. combined therapy consisted of treatment with SPSP and pharmacotherapy simultaneously for a period of six months. Different therapists provided the psychotherapy and pharmacotherapy.

7.2.3 Flow of participants and design

Table 7.1 shows the design of the three original trials.

7.2.4 Assessment of efficacy

The assessment of efficacy in each trial was based on data drawn from three sources: the patients, the treating therapists and independent observers. The independent observers (main criterion) used the 17-item HDRS (Hamilton, 1967). In the review of Bagby et al. (2004) the inter-rater reliability (Pearson´s r) ranged between 0.82 to 0.98. The independent observers were university students who almost completed their masters degree in Psychology. They were selected in an application procedure by the second author. Before administering the HDRS to patients the observers were trained by the last author, who has

113 Table 7.1: Design and flow of participants. Combined therapy Combined therapy Short versus very short versus pharmacotherapy versus psychotherapy combined therapy (de Jonghe et al., 2001) (de Jonghe et al., 2004) (Dekker et al., 2004) Assessed for eligibility 525 824 463 Excluded 358 616 360 Not meeting inclusion criteria 113 158 80 Meeting exclusion criteria 213 433 251 Refused 32 25 29 Randomized 167 208 103 Treatment Pharmaco- Combined Psycho- Combined Combined Combined conditions therapy therapy therapy of therapy therapy therapy including consisting including including including 16 SPSP 16 SPSP 16 SPSP 8 SPSP 16 SPSP sessions sessions sessions sessions sessions Randomized (ITT sample) 84 83 107 101 49 54 Started treatment (Per Protocol Sample) 57 72 106 85 45 45 Per protocol sample mega-analysis * 45 62 97 73 Not included 36 * excluded HDRS scores <14 and >24

developed a manual for a semi-structured interview for the HDRS (de Jonghe, 1994). During the studies the independent observers discussed their audio taped assessments monthly with the last author. Inter rater reliability for the independent observers was not statistically assessed. the therapists used the Clinical Global Impression of Severity (CGI-S) and Improvement (CGI-I) (Guy, 1976). The patients used the Ninety Symptom Checklist (SCL-90, Arrindell & Ettema, 1986.).

7.3 Method of the mega-analysis

7.3.1 Factors of the HDRS

Principal component analysis with varimax rotation was performed on the HDRS scores of the diagnostic sample (n=915). This sample concerns all the patients of table 7.1 that were assessed minus the patients that met exclusion criteria. hDRS factor solutions with three or four factors yielded unsatisfactory relia- bilities. Therefore we used the two factor solution which is presented in table 7.2.

114 Table 7.2: Principal Component Analysis (with Varimax-rotation) on the HDRS (diagnostic sample=963) (the final factor assignment printed bold). Factor 1 Factor 2 8 Depressed mood 0,64 0,21 14 Psychic anxiety 0,62 15 Guilt 0,58 -0,22 9 Work/interests 0,48 0,36 6 Somatic anxiety 0,45 0,19 16 Suicide 0,44 0,15 7 Hypochondrias 0,39 -0,14 10 General somatic (tiredness) 0,38 0,31 13 Agitation 0,28 0,11 2 Middle insomnia 0,59 4 Gastrointestinal 0,25 0,57 1 Early insomnia 0,18 0,52 5 Weight loss 0,50 3 Late insomnia 0,47 11 Retardation 0,31 0,36 12 Loss of libido 0,22 0,36 17 Loss of insight 0,18 Eigenvalue 2,40 2,11 Variance explained (%) 14,10 12,39 Reliability of the constructed factors Cronbach’s alfa (stand.) all items .70 factor 1: ‘Mental’ (items 6,7,8,9,13,14,15,16) .60 factor 2: ‘Somatic’ (items 1,2,3,4,5,10,11,12) .60

The model that best fitted our data the best, consisted of two factors in the HDRS. Both factors have ‘eigen values’ above 2. These factors explain 14,1% and 12,4% of the variance respectively. Factor one had the highest factor loadings on the items 6,7,8,9,10,13,14,15 and16. Factor two had the highest factor loadings on the items 1,2,3,4,5,11,12 and 17. The first was labelled the ‘Mental’ factor, loading the highest on the items addressing depressive mood, anxiety, suicide, hypochondriasis, psychic anxiety, guilt, loss of interest, agitation. To enhance the homogeneity of the factors we decided to put the item concerning ‘general somatic complaints’ under the second ‘Somatic’ factor, where the loading was still reasonable. This second factor also loaded on the items: insomnia (early, middle, late), weight loss, retardation, loss of libido gastrointestinal. We decided to leave out the item of ‘loss of insight’ since the loading was too low. It seems that in our sample, the HDRS covers two dimensions. One dimension regards the more ‘psychological’ core aspects of depression: a sombre mood, feelings

115 and thoughts of guilt and suicide, a loss of interest in the world around, worries about health and other anxieties. The second factor seems to cover the more vegetative aspects: insomnia, loss of appetite, tiredness, retardation and loss of libido. These factors had sufficient reliability (both Cronbach’s alfa=.60).

7.3.2 Study sample

Some patients were excluded from the mega-analysis. Firstly, the condition consisting of combined therapy with eight sessions of SPSP, from the study examining a dose-response relationship (Dekker et al., 2005), was excluded to enhance homogeneity at it was considered too different from the other combined therapy study conditions (16 SPSP sessions). Secondly, the trials differed with respect to the required 17-item HDRS baseline score: 14 points or more in two trials, and 12-24 points in the trial comparing psychotherapy and combined therapy. In order to enhance homogeneity, only patients with HDRS baseline scores of 14-24 were included in the mega-analysis; patients with HDRS scores < 14 and > 24 were excluded. This led to the exclusion of 12 cases in the pharmacotherapy condition, 9 in the SPSP condition and 36 in the combined therapy conditions. The mega-analysis sample therefore consisted of 313 patients: 45 in pharmacotherapy, 97 in SPSP and 171 in combined therapy (see also table 7.1). Table 7.3 presents the characteristics of this sample. there was only one significant difference between the treatment conditions. The SCL-D-score in pharmacotherapy group was somewhat lower than in the SPSP-group (p= 0.03). Although the absolute difference was probably clinically insignificant, the baseline scores for the outcome measures were included as covariates in the ANCOVA analysis of that outcome. The three samples did not differ statistically significant with regard to the presence of dysthymia.

7.3.3 Outcomes and statistical analyses

Relative efficacy is expressed in this study in inter-group differences in mean scores at week 24 (treatment termination). Assessments reflect the points of view of the independent observers, differentiated by the two factors found in our factor analyses of the HDRS, and of the patients, differentiated by three subscales of the SCL-90: Depression, Anxiety and Somatic complaints; the SCL- D, SCL-A and SCL-S. ancOVA analyses (including baseline scores for outcome measures as covariates) were used to test inter-group differences (one sided, level of significance .05; Altman, 1991, Bland & Bland, 1994). Our results are calculated in all patients who started with the treatment they were allotted to (a Per-Protocol or modified Intention-to-treat sample). The Last Observation Carried Forward approach was used. All statistical analyses were performed in SPSS (version 15.0).

116 Table 7.3: Characteristics of the Per protocol sample of the mega-analysis. Pharmaco- Psycho- Combined therapy therapy therapy 1 vs 2 1 vs 3 2 vs 3 (n=45) (n=97) (n=171) Sex Male 35,6 33,0 32,2 Female 64,4 67,0 67,8 Age 19-29 year 40,0 33,0 35,7 30-39 year 33,3 34,0 35,7 >40 year 26,7 33,0 28,7 Educational level Low 13,3 15,1 15,4 Middle 35,6 34,4 37,3 High 51,1 50,5 47,3 Living situation Alone 46,7 45,7 41,8 Living with at least one person 53,3 54,3 58,2 Marital status Married 20,0 20,2 20,6 Divorced/ widowed 13,3 7,4 11,2 Not married 66,7 72,3 68,2 Job status Sick/social benefit/other 53,3 53,8 55,3 Job 46,7 46,2 44,7 Duration present < 1 year 65,9 74,2 70,8 episode 1-2 year 13,6 12,4 16,1 > 2 year 20,5 13,5 13,0 Medication 3 months Yes 13,3 27,8 22,8 before study No 86,7 72,2 77,2 Depressed episodes 0 47,7 69,0 61,7 within previous 5 yrs 1 25,0 17,2 22,8 >= 2 27,3 13,8 15,4 Psychiatric treatment Yes 43,2 30,3 38,3 during present episode No 56,8 69,7 61,7 HDRS Sumscores Mean 19,0 18,7 18,7 SD 2,8 3,0 3,0 HDRS F1: Mental Mean 11,8 12,1 12,0 SD 2,9 2,4 2,4 HDRS F2: Somatic Mean 7,2 6,6 6,6 SD 2,5 2,6 2,4 SCL-90 Depression Mean 46,4 50,3 48,5 .030 SD 11,2 8,8 9,4 SCL-90 Anxiety Mean 23,3 24,9 24,8 SD 7,6 7,1 7,1 SCL-90 Somatic compl. Mean 25,0 27,1 26,5 SD 8,8 7,8 7,5 Personality disorder No 33,3 30,8 34,6 Yes 66,7 69,2 65,4

117 7.4 Results of the mega-analysis on the HDRS factors and SCL-90 subscales

Table 7.4 presents the mean scores at week 24 (treatment termination) of the HDRS ‘Mental’ factor and the ‘Somatic’ factor, and the sub scales SCL-D, SCL- A, SCL-S and the total of the three SCL sub scales. intra group pre-post comparisons for all factors and sub scales were significant. SPSP outperformed pharmacotherapy on the SCL-Anxiety subscale. Com- bined therapy was superior to pharmacotherapy on three outcome measures: the HDRS ‘Mental’ factor of the HDRS, SCL-Depression and the SCL-Anxiety. Combined therapy yielded better results than SPSP on the ‘Somatic’ factor of the HDRS, the SCL-Depression and de sum of the three SCL sub scales.

7.5 Discussion

We compared the efficacy of SPSP, pharmacotherapy and their combination in the treatment of mild to moderate depression on the level of two HDRS factors and three subscales of the SCL-90 (Depression, Anxiety, Somatic complaints).

Mega-analysis on HDRS factors No differences were found between SPSP and pharmacotherapy on both factors of the HDRS. The findings are not in concordance with our hypothesis that SPSP would be superior on the HDRS ‘Mental’ factor and pharmacotherapy on the HDRS ‘Somatic’ factor. Apparently the treatments perform equally well on both dimensions. These results are in line with the original general mega-analysis, where no difference was found on the HDRS sum scores (p=0,18). still, a number of factors could be taken to indicate that SPSP does indeed address more mental or psychological depression aspects. Combining pharmacotherapy with SPSP enhances the efficacy on the HDRS ‘Mental’ factor significantly, compared to pharmacotherapy alone, but there was no statistically significant difference when comparing combined therapy to SPSP alone. Apparently, SPSP holds an intermediate position between combined therapy and pharmacotherapy with regard to the ‘Mental’ factor. in our earlier mega-analysis combined therapy was superior to pharmacotherapy (HDRS, p=0,02). The current study further differentiates this and shows that the difference is attributed to changes in one specific symptom cluster, the mental factor. Combined therapy is better than pharmacotherapy on the HDRS ‘Mental’ factor (possibly due to the addition of SPSP), but not on the HDRS ‘Somatic’ factor. there are also indications that pharmacotherapy is more efficacious than SPSP in more vegetative aspects of depression. Adding pharmacotherapy to SPSP enhances the efficacy on the HDRS ‘Somatic’ factor significantly, compared to SPSP alone, but not to pharmacotherapy alone. Apparently, with regard to this factor, pharmacotherapy holds the intermediate position between combined therapy and SPSP. No difference was found between combined therapy and

118 p

ther.

vs. 0,014 0,001 0,003

SPSP 0,34 0,280 9,87 2,26 0,067 Combined

F sided

vs. ther.

0,029 0,003 0,000 2,59 0,055 7,59 3,63 1,14 0,144 1,39 0,120 2,65 0,053 4,98 7,58 Pharmaco

17,47 Combined

p F sided

0,003 SPSP Pharmaco

, pharmacotherapy and their combination (week group

4,69 313 2,05 0,077 7,67 312 0,00 0,488 3,23 313 0,04 0,422 7,49 312 7,91

SPSP 6,80 4,15

therapy Total

4,74 171 7,17 170 21,21 2,86 171 6,83 170 18,43

-90 subscales for 6,48 3,75

Combined SCL

4,33 97 8,17 97 20,78 3,65 97 7,44 97 17,50

factors and 6,82 4,59

RS D H 5,15 45 3,48 45 8,40 45 22,10 9,21 45 18,79

sided 7,96 4,73 Pharmacotherapy SPSP Mean SD N Mean SD N Mean SD N Mean SD N F One- One- One-

Mean scores on d,a,s 77,78 28,30 45 77,85 26,68 97 69,66 23,13 170 73,38 25,31 312 0,08 0,392

Mental Somatic

SCL somatics 20,87 depression 35,76 13,16 45 anxiety 36,95 13,91 97 31,38 11,81 21,16 170 33,74 12,92 312 0,15 0,350

Sum HDRS: HDRS: SCL

Table 7.4: 24). Treatment

SCL SCL

119 SPSP in the original mega-analysis (HDRS, p=0,087), and this clearly obscured the fact that adding pharmacotherapy to SPSP improves the more vegetative symptom cluster of depression. This finding is conform our hypothesis that pharmacotherapy would address this dimension more than psychotherapy.

Mega-analysis on the subscales of SCL-90 According to patients combined therapy is superior to both monotherapies in SCL measured Depression. This sub scale is comparable to the ‘Mental’ factor of the HDRS (see Appendix 1). So, although independent observers (HDRS) are not so clear in the superiority of combined therapy, patients (SCL) find it clearly preferable over monotherapy with regard to psychological depression aspects. spsp is superior to pharmacotherapy in SCL-Anxiety. It may surprise that SPSP is more potent than pharmacotherapy in reducing anxiety, as SSRI’s and SNRI’s are first choice anxiolytic drugs. A possible explanation for this effect of SPSP might be that it advocates a strong supportive stance of the therapist (de Jonghe, 2005). This may reduce anxiety of depressed patients to a high extent. Not surprisingly, adding SPSP to pharmacotherapy yields better results in this subscale than pharmacotherapy alone (but not than SPSP alone). no differences were found with regard to the SCL Somatic complaints subscale, and in this regard either mono therapy is equal to combined therapy. The Somatic complaints subscale of the SCL-90 does not seem to be comparable with the vegetative aspects of depression (see Appendix 1). the sum scores of the three SCL subscales reveal that there are no differences found between SPSP and pharmacotherapy and between combined therapy and pharmacotherapy. Again it shows that sum scores sometimes may prevent the discovery of changes in specific dimensions of the illness. In addition, a difference found in sum scores between combined therapy and SPSP is explained in more detail by the finding that this difference is mainly accounted for by the better performance of the first treatment on the SCL Depression subscale.

Limitations Our study presents several limitations. Firstly, the method of factor analysis is based on subjective decisions of researchers that choose the model that in their opinion best fits the data. In addition, finding the most fitting clinical solution, is often obtained at the prices of a reduction in variance explained (Marcos & Salamero, 1990). In a systematic review Bagby et al. (2003) examined 15 studies that performed factor analyses on the HDRS. The number of factors found ranged from 2 to 8. The poor comparability of these findings could be partially explained by the heterogeneity of the studies’ patient samples. Included were geriatric patients, general ‘psychiatric patients’, schizophrenic patients, depressed patients, general ‘medical’ patients and inpatients as well as outpatients. Furthermore, the studies cover a variety of countries across the world, adding to the heterogeneity, and used different methods of factor analyses. One study in the Bagby review (Steinmeyer&Möller, 1992), regarding depressed inpatients, found 2 factors or facets as they have called them. One

120 facet was labelled an indicator for depressive severity and contained the items: depressive mood, guilt, delayed insomnia, work and interest, psychic anxiety, general somatic. The other was labelled a stabile indicator for somatisation aspects of depression and contained the items: somatic anxiety, gastrointestinal and genital. Although our solution did differ with the findings of Steinmeyer and Möller (1992) on some items , it could be concluded that the dimensions are quite similar, with a more ‘mental depression’ facet (that they labelled indicator of depression severity or core depression items) and a more ‘vegetative’ or ‘somatic’ one. Our findings are also in line with the results of the extensive meta-analysis of Shafer (2006). The author analysed four depression scales: the Beck Depression Inventory (BDI), the HDRS, the Center for Epidemiological Studies Depression Scale (CES-D) and the Zung Self-Rating Depression Scale (SDS). He found two factors in all four tests: a General Depression Severity factor and a small Somatic Symptoms factor. secondly, the study sample of the mega-analysis suffers from the selection bias inherent to all RCT’s. Most trials exclude patients with, for example, serious comorbidity, suicidal intentions or substance abuse. Patients meeting all inclusion and no exclusion criteria applied in RCT’s represent only a small proportion of the depressed target population. A cautious approach is therefore required to the generalization of conclusions from RCT’s. thirdly, there is one difference at baseline between the study samples of the three original trials. Since the baseline scores for outcome measures were included in analyses as covariates, this difference has been accounted for. fourthly, the pharmacotherapy protocols in the three original trials differed slightly (the SPSP manual, on the other hand, was identical). We did not, however, find significant differences between the mean scores or remission rates of combined therapy in the three samples of the original RCT’s. Furthermore, meta-analytic studies have shown that there are no clinically significant differences between antidepressants in terms of treating depressed outpatients (Anderson, 2000). fifthly, we are well aware that the results comparing SPSP and pharmacotherapy are not based on a head to head comparison between these treatments in one RCT. This may have affected the results. However, we believe both groups are highly comparable as the studies were performed in the same outpatient facilities with the same study design and within the same population. Furthermore, we tested for pre-treatment differences. sixthly, we did not investigate whether patients with higher initial scores on one of the factors or subscales may benefit more from one of the treatment modalities. Therefore we can not conclude that the clinical implication is that higher scores predict therapy success of one of the treatment modalities. How- ever, our findings may serve as the basis for hypotheses in research that address these questions. This is highly relevant as it furthers the insight in the tailoring of treatments towards the individual needs and possibilities. finally, as the sample size of the pharmacotherapy group was relatively small, it is conceivable that differences between this treatment and the comparison categories are not easily detected. Some trend findings in our results do

121 suggest that differences may exist in larger samples that did not reach statistical significance in the current analyses.

Strengths Firstly, our factor analysis is based on a large sample (n=915), that is by its characteristics representative for ‘real world’ psychiatric practice. Secondly, in our opinion the strong clinical and methodological homogeneity of the mega- analysis adds to its strength and justifies the pooling of the data. Furthermore, we have included the opinions of both independent observers and patients on the efficacy of three treatment modalities. We have focused on subscales of the HDRS and SCL-90 thereby adding information on the specific components that are probably effective in the different treatments.

Summary and conclusions Our results suggest that, in regard to the ‘Mental’ HDRS factor, combined therapy outperforms pharmacotherapy, while SPSP holds an intermediate position. In regard to the ‘Somatic’ HDRS factor combined therapy outperforms SPSP while pharmacotherapy holds an intermediate position. According to the SCL Depression subscale, combined therapy is the more efficacious treatment. According to the SCL Anxiety subscale combined therapy and SPSP are the more efficacious treatments. According tot the SCL Somatic complaints subscale the three treatment modalities are equally efficacious. it seems that combined therapy is the first choice treatment according to the SCL Depression subscale. As monotherapy is, other things being equal, to be preferred over combined therapy, it may tentatively be concluded that SPSP is the first choice treatment in regard to the HDRS ‘Mental’ factor and the SCL Anxiety subscale. In regard to the HDRS ‘Somatic’ factor, it is pharmacotherapy. As far as the SCL Somatic complaints subscale is concerned, it is SPSP or pharmacotherapy. looking at efficacy results at subscale level may yield valuable information about the working components of treatments. The exact clinical implications of our findings need to be closer investigated, especially the question whether initial scores on sub dimensions have predictive value for treatment success. However, the results indicate that treatment indications for (mild to moderate) depression probably can be more focused.

References

Altman, D.G. (1991). Practical Statistics for medical research. Chapman & Hall, Lon- don. American Psychiatric Association (1980). Diagnostic and Statistical Manual of Mental Disorders (3rd edn) (DSM-III). Washington DC: APA. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders. Fourth edition (DSM-IV). Washington: APA. Anderson, I.M. (2000). Selective serotonin reuptake inhibitors versus tricyclic antidepressants: a meta-analysis of efficacy and tolerability. Journal of Affective Disorders, 58, 19-36.

122 Arrindell WA, Ettema JMM. SCL-D-90 (1986). Handleiding bij een multidimensionele psychopathologie-indicator. Lisse: Swets & Zeitlinger. Bagby RM, Ryder AG, Schuller DR, Marshall MB. (2004). The Hamilton Depression Rating Scale: Has the Gold Standard Become a Lead Weight? American Journal of Psychiatry; 161, 2163-2177. Barefoot, J., Brummet, B. Helms, M., Mark, D., Siegler, L., Williams, R. (2000). Depres- sive symptoms and survival of patients with coronary artery disease. Psychosomatic Medicine, 62: 790-795. Bland, J.M. & Altman, D.G. (1994). Statistics Notes: One and two sided tests of significance. British Medical Journal, 309(6949): 248. Cornman, B.J. (1997). Female adolescent response to childhood sexual abuse. Journal of Children and Adolescent Psychiatry and Nursing, 10(2), 17-25. Dekker J, Molenaar PJ, Kool MA, van Aalst G, Peen J, de Jonghe F. (2005). Dose-effect relations in time-limited combined psycho-pharmacological treatment for depression. Psychological Medicine, 35(1): 47-58. Devins, G.M., Orme, C.M., Costello, C.G., Binik, Y.M., Frizell, B., Stam, H.J., Pulin, W.M. (1988). Measuring depressive symptoms in illness populations :Psychometric properties of the Center forEpidemiological Studies Depression (CES-D) Scale. Psy- chology & Health, 2, 139-156. Dolan, B., Morton, A., & Wilson, J. (1990). Selection of admissions to a therapeutic community using a group setting: association with degree and type of psychological distress. International Journal of Social Psychiatry, 36(4), 265-271. Dyer, S.J., Abrahams, N., Mokoena, N.E., Lombard, C.J., van der Spuy, Z.M. (2005). Psy- chological distress among women suffering from couple infertility in South Africa: a quantitative assessment. Human Reproduction, 20(7), 1938-1943. Endler, N., Rutherford, A., & Denishoff, E. (1999). Beck Depression Inventory: Explor- ing its dimensionality in a non-clinical population. Journal of Clinical Psychology, 55, 1307-1312. Faries, D., Herrera, J., Rayamajhi, J., Debrota, D., Demitrack, M., Potter, W. (2000). The responsiveness of the Hamilton Depression Rating Scale. Journal of Psychiatric Research, 34, 3-10. Feldman, L.B., Grodman, J., & Behrman, D. (1983). Psychiatric symptoms of maritally distressed wives and husbands. International Journal of Social Psychiatry, 29(2), 140-145. Guy W. (1976). ECDEU Assessment Manual for Psychopharmacology. Revised DHEW pub.(adm). Rockville, MD: National Institute for Mental Health. Hamilton, M. (1967). Development of a rating scale for primary depressive illness. British Journal of Social and Clinical Psychology, 6, 278-296. Holm, J., Penzien,, D., Holroyd, K., Brown, T. (1994). Headache and depression: Con- founding effects of transdiagnostic symptoms. Headache, 34, 418-423. Huyser, J., Jonghe, F. de, Jonkers, F., Schalken, H. (1996) The manual for the Diagnos- tics of Major Depression (MDMD): description and reliability. International Journal of Methods in Psychiatric Research, 6, 1-4. Jonghe F, Rijnierse P, Janssen R. (1994). Psychoanalytic supportive psychotherapy. Journal of the American Psychoanalytic Association, 42, 421-446. Jonghe, F., de. (1994). Leidraad voor het scoren van de Hamilton Depression Rating Scale. Amsterdam: Benecke Consultants. Jonghe, F., de, Kool, S., Aalst, G., Dekker, J., Peen, J. (2001). Combining psychotherapy and antidepressants in the treatment of depression, Journal of Affective Disorders, 64, 217-229. Jonghe, F., de, Hendriksen, M., Aalst, G., Kool, S., Peen, J., Van, H., Eijnden, van den, E., Dekker, J. (2004). Psychotherapy and Combined Therapy (pharmacotherapy plus psychotherapy) in the treatment of depression. British Journal of Psychiatry, 185 (1), 37-45.

123 Jonghe, F., de (2005). Kort en krachtig. Kortdurende Psychoanalytische Steungevende Psychotherapie. Benecke N.I. Amsterdam. Kilgore, W. (1999). Emprirically derived factor indices for the Beck Depression Inven- tory. Psychological Reports, 84, 1005-1013. Maat, S.C.M., Dekker, J., Schoeves, R., van Aalst, G., Gijsbers-van Wijk, C., Hendriksen, M., Kool, S., Peen, J., Van, H., d Jonghe, F. (2007). Short Psychodynamic Supportive Psychotherapy, Antidepressants and their Combination in the Treatment of Major Depression: A mega-analysis based on three randomized controlled trials. Depres- sion and Anxiety (in press). Marcos, T., & Salamero, M. (1990). Factor study of the hamilton Rating Scale for Depres- sion and the Bech Melancholia Scale. Acta Psychiatrica Scandinavica, 82, 178-181. Muller, M.J., & benkert, O. (2001). Lower self-reported depression in patients with erectile dysfunction after treatment with sildenafil. Journal of Affective Disorders, 66, 255-261. Rickels, K., Downing, R.W., Lipman, R.S., Fisher, E., & Randall, A.M. (1973). The Self- Rating Depression Scale (SDS) as a measure of psychotropic drug response. Diseases of the Nervous System, 34, 98-104. Ross, C.E., & Mirowsky, J. (1984). Components of depressed mood in married men and women: The Center of Epidemiological Studies Depression Scale. American Journal of Epidemiology, 119, 997-1004. Shafer, A. (2006). Meta-analysis of the factor structures of four depression questionnaires: Beck, CES-D, Hamilton, and Zung. Journal of Clinical Psychology, 62(1): 123-146. Steinmeyer, E.M., Möller, H.J. (1992). Facet theoretic analysis of the Hamilton-D scale. Journal of Affective Disorders, 25, 53-61. Tashakkori, A., Barefoot, J., & Mehryar, A. (1989). What does the Beck Depression Inventory measure in college students? Evidence from a non-Western culture. Jour- nal of Clinical Psychology, 45, 595-602. Williams, A.C., & Richardson, P.H. (1993). What does the BDI measure in chronic pain? Pain, 55, 259-266. Yin, S., Robins, C.J., & Lin, N. (2000). A cross-cultural comparison of depressive symptom manifestation: China and the United States. Journal of Consulting and Clinical Psychology, 68, 993-999. Zemore, R., & Eames, N. (1979). Psychic and somatic symptoms of depression among young adults, institutionalized aged and non- institutionalized aged. Journal of Gerontology, 34, 716-722.

124 Chapter 8 The effectiveness of long-term psychotherapy: methodological research issues

Saskia de Maat, Jack Dekker, Robert Schoevers, Frans de Jonghe Psychotherapy Research (2007) 17(1): 59-65

Abstract

Introduction. In Evidence Based Medicine (EBM) hierarchy randomized controlled trials (RCT’s) are ranked higher than cohort studies. Long-term Psycho- Therapy (LPT) has often been investigated with cohort intervention studies but hardly with RCT’s. As a consequence it is generally concluded that its effectiveness has not been demonstrated. This conclusion may be unwarranted. Method: A critical discussion of the acceptability, feasibility and decisive power of RCT’s and cohort studies regarding LPT. Results: The only essential and unchangeable difference between RCT’s and cohort studies is that the first always include randomized control groups and the second never. As far as internal validity is concerned, this gives RCT’s a head start that cohort studies cannot ever level. However, randomization nearly always has dramatic consequences for LPT research: the control conditions that are most informative (no treatment, waiting list, placebo) are so unacceptable for the patients that decisive RCT’s are in most cases unfeasible. Better feasible RCT’s are less decisive. This is a serious limitation of RCT’s, not of LPT. Cohort studies are more feasible than RCT’s. Their decisive power is determined by their methodological quality and by the knowledge of the natural course of the investigated disorders. Cohort studies are equally well as RCT’s capable of meeting all quality criteria for intervention research, except for randomization. The knowledge of the natural course of the disorders suitable for LPT treatment is limited but not inexistent. Conclusion: In most cases of LPT research, decisive RCT’s present insurmountable, method inherent feasibility problems. In these cases they represent, not the highest but an irrelevant level of evidence, meaning that cohort studies provide the best available evidence.

8.1 Introduction

Psychotherapy is not an applied science but a scientifically based art and skill. The general criteria of science, among others, apply to the field. The Ameri- can Psychological Association (Task Force on Promotion and Dissemination

125 of Psychological Procedures, 1995; Chambless and Hollon, 1998) has defined quality criteria to assess whether or not the effectiveness of a psychotherapeutic method has been sufficiently substantiated empirically to deserve the qualifi- cation Empirically Supported Psychological Therapy (ESPT) (Kendall, 1998). ESPTs meet the standards of Evidence Based Medicine (EBM): ‘The conscien- tious, explicit and judicious use of current best evidence in making decisions about the care of individual patients. The practice of evidence-based medicine requires the integration of individual, clinical expertise with the best available, external, clinical evidence from systematic research and our patients’unique values and circumstances’(Center of Evidence Based Medicine Toronto, Glos- sary, 2005 (CEBMT)). researchers in the field of treatment effectiveness (‘outcome research’, not ‘process research’) agree that empirical evidence can and must be ordered in a hierarchical system. A widely accepted hierarchy (Center of Evidence Based Medicine Oxford, Levels of evidence, 2001 (CEBMO)) differentiates between five levels.

Hierarchy of empirical evidence (CEBMO)* 1.a. a systematic review of RCTs with consistent results 1.b. one high quality RCT 1.c. one all-or-none study 2.a. a systematic review of cohort studies or of patiënt-control studies with consistent results 2.b. one cohort- or patiënt-control study or one lower quality RCT 2.c. outcome research 3.a. a systematic review of case-control studies 3.b a case-controle study 4. case series or a lower quality cohort study 5. expert’s opinions or generally accepted therapeutic methods * Center of Evidence Based Medicine Oxford, Levels of evidence, 2001

Randomized Controlled Trials (RCT’s), or their systematic reviews and meta- analyses, occupy the first place in this hierarchy. Cohort studies hold the second place. A cohort is a well-defined group of people, who are followed over a certain period of time in order to observe whether a certain ‘outcome’ occurs. The relation between ‘a determinant’ (in our case LPT) and outcome (the effect) is scrutinized. If the determinant is an intervention, this design is also called a quasi –experimental study: ‘experiments that lack random assignment (…) but that otherwise have similar purposes and structural attributes to randomized experiments.’ (Shadish et al., 2002). Quasi-experimental studies may or may not include a comparison group. In this paper we will apply the term cohort study, because we feel it is most frequently used in EBM-literature. psychotherapy has been extensively investigated with RCT’s, as is testified, e.g., by the systematic reviews of Crits-Christoph (1992), Perry, Banon and Ianni (1999), Leichsenring and Leibing (2003) and Leichsenring, Rabung and

126 Leibing (2004). Together, these four reviews mention twenty-eight different RCT’s. Twenty-seven of them regard Short-term PsychoTherapy (SPT), only one refers to Long-term PsychoTherapy (LPT). RCT’s have frequently studied short-term and rarely long-term therapies in general and especially in psychotherapy. This state of affairs has led to the common opinion that LPT is not an ESPT. Its effectiveness would not have been researched scientifically, let alone convincingly demonstrated. This conclusion is based on two premises: (a) that conclusive RCT’s are feasible in LPT research and (b) that RCT’s are superior to cohort studies in all important methodological aspects. In this article we discuss these views and assess their validity. We apply an arbitrary distinction between SPT and LPT: the latter consists of at least fifty sessions and lasts at least one year.

8.2 Randomized Controlled Trials and long-term psychotherapy

8.2.1 Introduction

RCT’s derive their name from their most important characteristic, the randomization of patients. This study design is also called the ‘confirmatory-deductive methodology’ and is applied in ‘efficacy research’. RCT’s strive for a maximum internal validity, so that it can be assumed beyond reasonable doubt that differences found between treatment and control groups (the dependent variable) are explained by the therapeutic intervention (the independent variable). Within reasonable limits this method ensures like no other that chance is the only alternative explanation for the differences found between the studied groups. Factors possibly (co)determining results (‘confounders’) are neutralised as much as possible. To this end, the groups have to be, apart from the intervention concerned, comparable to each other (e.g., regarding demographical and clinical characteristics). The tool to achieve this is random assignment. This means patients are allocated by chance, either to the experimental treatment or to the control condition. As Kunz and Oxman (1998) state: ‘It is a paradox that unpredictability is introduced into the design of clinical trials by using random allocation to protect against the unpredictability of the extent of bias in the results of non-randomized clinical trials.’ The advantages of randomization are obvious and justify the high ranking of RCT’s within the hierarchy of empirical evidence. However, this does not take away the fact that, especially over the past decade, RCT’s have been severely criticised as well, not only in the field of psychotherapy and regard- less of treatment length. (Seligman, 1995; Hollon, 1996; Howard, Moras, Brill, Martinovich and Lutz et al., 1996; Marks, 1997; Crits-Christoph, 1997; Healy, 2001; Kaptchuck, 2001; Leichsenring, 2004). The criticism especially focuses on the external validity of RCT’s, i.e., on the generalizability of RCT results to daily practice. Many patients refuse participation, are excluded or drop out. The manuals and protocols applied, the many assessments, etcetera, poorly connect to daily patient care. Therefore, the question arises whether conclusions drawn

127 from such research also stand ‘outside the laboratory’. In short, RCT’s opt for internal validity at the expense of external validity. In order to offer a synthesis in this discussion Leichsenring (2004) has argued that the RCT and the quasi- experimental study do not differ in principle concerning their external and internal validity, given their own specific context (respectively the laboratory and the field). hereafter, we mention but do not expand on these issues, which have already received much attention in literature. We focus on the consequences of randomization for the feasibility and conclusive power of RCT’s regarding LPT, as we are convinced that this is even more of a problem for long-term treatments than for short-term treatments.

8.2.2 Acceptability, feasibility and conclusive power of RCT’s in LPT

Based on chance, RCT’s compare a treatment to be examined (the ‘experimental’ one, in this case LPT) to a control condition. The following seven alternatives qualify as control conditions:

1. no treatment 2. a waiting list 3. a ‘placebo’ treatment 4. ‘Treatment As Usual’ (TAU) 5. the experimental LPT in a ‘low dose’ 6. another LPT with an assumed but unproven efficacy 7. another LPT with an established efficacy

RCT’s comprising one of the first three control conditions mentioned above (no treatment, waiting list or placebo) provide the strongest proof of efficacy. As far as LPT is concerned, however, they merely are theoretical options for hardly any well informed patient will participate in a study lasting at least one year and offering a fifty percent chance on no treatment, a waiting list or a treatment only resembling psychotherapy. Should anyone accept the chance on such control conditions and indeed be allocated to one, it can be safely assumed that he will seek treatment elsewhere in the course of the year. And if, by any chance, that patient would be compliant with the control condition, he is most certainly not representative for the study population. In short, regarding LPT these RCT’s are almost unfeasible when they are conclusive, and they are inconclusive when they are feasible. Small wonder they are not found in literature. rct’s can compare LPT to Treatment As Usual (TAU). This research type has been published by Linehan, Armstrong, Suarez, Allmon and Heard ( 1991), Linehan, Schmidt, Dimeff, Craft, Kanter and Comtois (1999), Linehan, Dim- eff, Reynolds, Comtois, Welch and Kivlahan (2002), Verheul, van den Bosch, Koeter, de Ridder, Stijnen and van der Brink (2003), and Bateman and Fonagy (1999). These publications all regard patients with Borderline Personality Dis- order (BPD). This design is feasible due to the fact that there is an acceptable TAU for BPD patients, namely treatment at a psychiatric outpatient clinic. Bate-

128 man and Fonagy (2000) clearly favour this study type: ‘Waiting-list controls cannot be used to control for change over long periods of time, and so are of limited use. The most stringent control group without ethical problems is TAU which should be used in future, even though heterogeneity of interventions and differential responsiveness within groups may obscure results.’ As far as we know, there are no RCT’s comparing LPT to TAU in non-BPD-patients. The reason is not that there is no LPT for these patients, but there is no acceptable TAU for well-informed patients. Without exaggeration it may be said that TAU mostly consists of minimal therapy, hardly rising above the first three options mentioned before. It has been said that TAU comes down to cheap, low-frequency treatment, provided by minimally trained paraprofessionals with an overwhelming caseload, struggling to get by (Scheel, 2000). What patient would be prepared to chance such control condition in a long-term study? Borderline patients do, what else could they do? In short, regarding LPT the feasibility of RCT’s comprising TAU as control condition is clearly limited. a fifth study design entails that LPT in a dose considered ‘normal’ is compared to itself in a ‘lower dose’. In pharmacological studies this mostly means a dose considered ineffective, in fact a placebo in disguise, unless several doses are tested in search of a ‘dose-response’ curve. In the context of psychotherapy the term ‘dose’ refers to session frequency. This control condition is possibly more acceptable to patients, but here another problem arises. It has to do with the possible outcome of this research type. Only an apparent difference in effectiveness is conclusive, because finding no difference may as well signify that the two treatments are efficacious as that they are both inactive. To correctly interpret a ‘negative’ result (finding no difference between the two dosages) it is necessary to use the ‘natural’ course of the disorder (the course of the disorder untreated) as a reference. If the natural course is known, however, cohort studies would suffice. there is yet another drawback to comparing two psychotherapy doses. A large dose difference (e.g. two sessions weekly versus one fortnightly) increases the probability of finding a difference in effect, thus enhancing the conclusive power of the study. However, it decreases the acceptability of the control condition to patients, thus reducing the feasibility of the study. A small dose difference on the other hand, e.g. four sessions weekly versus three, increases the acceptability and thus the feasibility of the study, but it decreases the probability of finding a difference in effect and thus the conclusive power of the study. This RCT type is either conclusive but hardly feasible or highly feasible but poorly conclusive. It seems that the conclusive power and the feasibility of RCT’s are two enemies convicted to one another. the next study type compares two LPT’s the efficacy of which is assumed though it has not been proven in conclusive RCT’s. An RCT comparing Schema- Therapy (Young, Klosko and Weishaar, 2003) to Transference-Focused Psycho- therapy (Kernberg, 2002) is an example. Which of the two treatments represents the control condition depends on the point of view of the researchers. The acceptability and thus the feasibility of this design are fair. However, here too only finding a difference in effect is conclusive (unless the natural course of the

129 disorder is known, what would render the RCT superfluous). When researchers sincerely believe that two treatment methods are roughly equivalent, the probability of finding a difference is low. In short, in most cases this study design is fairly feasible but poorly conclusive. the last option is a comparison of LPT with another LPT with an undisputed efficacy. Dialectic Behaviour Therapy (Linehan et al., 1991, 1999, 2002; Verheul et al., 2003) and Mentalisation Based Therapy (Bateman and Fonagy, 1999) in the treatment of BPD-patients qualify as such control conditions. Acceptability and thus feasibility of this RCT type are passable. However, two new problems arise here. The first regards the conclusive power of this research type. Finding a difference in favour of the experimental treatment would naturally be instruc- tive, but chances on that are slim. This study design is mostly applied when the aim is to find no difference in order to conclude that the new method is equal in efficacy to the established one (the ‘me too’ method). No difference may be found, but that does not necessarily mean that the new method differs as clearly as the established one from a control condition that implies no treatment, a waiting list or a placebo. The second and most prominent problem is that there are no LPT’s with RCT-proven efficacy for non-BPD patients.

8.3 Cohort studies and long-term psychotherapy

8.3.1 Introduction

Where RCT’s are not feasible, what is the highest level of empirical evidence? According to the CEBMO hierarchy of evidence the answer must be: cohort studies. Compared to RCT’s, cohort studies have a lower internal validity. In contrast with this, however, the external validity is higher because cohort studies present less selection bias. In addition, in most designs cohort studies do while RCT’s do not investigate treatment methods conforming to norms that hold in daily practice.

8.3.2 Acceptability, feasibility and conclusive power of cohort studies in LPT

In cohort studies only the acceptability of the treatment condition counts, not that of a control condition as well. Therefore, the feasibility of cohort studies is, especially in LPT, better than that of RCT’s. the conclusive power of cohort studies is still a matter of debate. Many feel that cohort studies tend to overestimate the effects of treatments, as there is no correction for potential confounders, meaning that the comparability of the treated and the untreated group may be doubted. (Sacks, Chalmers and Schmidt, 1982; Kunz and Oxman, 1998). Theoretically this is certainly possible, but whether overestimation does actually occur in practice remains to be seen. An exhaustive and impressive study, published in the New England Journal of Medicine (Concato, Shah and Horwitz, 2000), compared the results of RCT’s to those of cohort studies regarding the same interventions in five

130 different somatic disorders. Analyses of seventy-two meta-analyses of RCT’s, twenty-four meta-analyses of cohort studies or case-control studies and six meta-analyses involving both designs brought to light that the results of the observational studies remarkably resembled those of the RCT’s. Concato et al. conclude: The popular belief that only randomized, controlled trials produce trust-worthy results and that all observational studies are misleading does a disservice to patient care, clinical investigation, and the education of health care professionals. It may be clear that these findings are important in light of the discussion on research into the effectiveness of LPT. the conclusive power of cohort studies depends on the methodological quality of studies themselves and on the knowledge of the natural course of the disorders studied. We will discuss these two aspects separately.

8.3.3 The methodological quality of cohort studies

The actual practice of cohort studies is often ‘looser’ than that of RCT’s. This has resulted in the general opinion that cohort studies are per se of lower quality than RCT’s. This is a misconception. Except for randomization, the assessment criteria for the methodological quality of intervention-cohort studies do not differ from those for RCT’s. Both research types can be of superior or inferior quality. It is in this light that Leischenring (2004) has proposed a hierarchy of evidence for both RCT’s and quasi-experimental studies, based on criteria (see below) that do not represent a sine qua non condition for high-level research. As a rule they are not categorical (yes/no) but dimensional (studies meet them from a slight to considerable extent). The following list is not exhaustive:

• prospective study design • a control group, preferably via extensive matching • specific, homogeneous and representative study groups • clearly defined, specific and representative treatments of adequate duration • follow-ups of adequate duration • pre-, post and follow-up assessments • reliable and valid instruments applied by independent assessors • adequate statistical processing • acceptable dropout percentage during treatment and follow-up • ‘Intention-to-treat’ and ‘per protocol’ approaches and dropout accountabi- lity.

Cohort studies are prospective or retrospective. They can but do not necessarily have to include a (of course not randomized) control group (Kazin, 1998, Shadish et al., 2002). ‘Single-group cohort designs’ follow one study group. There is no control group. Pre- and post-measuring makes each patient functioning ‘as his own control’. ‘Multi-group cohort designs’ follow at least two groups. One group is exposed to the intervention (e.g. psychoanalysis), the other is often an otherwise treated control group (e.g. psychoanalytic psychotherapy). If the groups are matched on relevant variables the results of

131 both groups can be cautiously compared. Matching regards variables, e.g., demographical and diagnostic characteristics, which might contribute to the differences in effect. Matching is not an all-or-none phenomenon. The smaller the number of variables included in the matching, the less the ‘control group’ deserves its name. Epidemiological data regarding the general population may also serve as a control for the results of cohort studies. Untreated groups in epidemiological research (which reflect the ‘natural course’ of a condition) can be compared to treated cohorts. Again, the more the persons from epidemiological studies are matched to the cohort, the stronger the evidence.

8.3.4 Natural courses of the disorders treated with LPT

Knowledge of the ‘natural course’ of the disorder enhances the conclusive power of cohort studies considerably. The natural course regards the ‘spontaneous remission’ rate in untreated patients presenting the disorder concerned. An example may clarify this issue. Although the efficacy of the surgical removal of an inflamed appendix has not been investigated in any RCT, nobody considers this a problem. The reason is that people are convinced they know sufficiently well the natural course of appendicitis to consider RCT’s in this case superfluous and on ethical grounds unacceptable. Knowledge of the natural course of disorders can be based on clinical experience (as is the case with appendicitis) but it is preferable that results from epidemiological research. In practice it proves difficult to determine the real (because untreated) natural course of disorders treated with LPT. A number of studies may serve as an example. Spijker, de Graaf, Bijl, Beek- man, Ormel and Nolen (2000) studied the course of major depression (MDD) in the Netherlands. It appeared that fifty percent of all new cases recover within the first three months. After that period the probability of remission diminishes and becomes nil after twelve months. After two years twenty-six percent are still not remitted. Two studies suggest they provide relevant information on the natural course of BPD. Perry, Banon and Ianni (1999) come to the conclusion, based on a survey of five studies, that two- and ten years after treatment termination respectively nineteen and forty-eight percent of BPD patients have recovered. This would mean a remission rate of 3.6 percent (round up 4%) per year between the second and tenth year. Zanarini, Frankenburg, Hennen and Silk (2003) come to the conclusion, based on one study (n=290), that two and six years after treatment termination respectively thirty-five and sixty-six percent have recovered. This would mean a remission rate of 7.7 percent (round up 8%) per year between the second and sixth year. Taking these two publications into account, the remission percentage of BPD could roughly be estimated at six percent a year. In this context, however, ‘recovered’ only means that the patient no longer meets the DSM-criteria for BPD. Unfortunately, the three studies mentioned above relate to the course, not to the natural course of the disorder. The studies of Perry and of Zanarini relate to the course of BPS after termination of

132 intensive treatment. In addition, most patients of Spijker, of Perry and of Zanarini had sought and received help during the follow-up period. It may be assumed that the natural course of disorder is even less favourable than presented (unless one assumes that treatment is to no avail or even worsens the outcome). in short, knowledge of the natural course of disorders is scarce, but epidemiological research of untreated disorders is necessary to enhance the conclusive power of any long-term treatment study.

8.4 Discussion

This article discusses two premises that are more or less explicitly taken for granted in research regarding the efficacy of LPT: (a) that conclusive RCT’s are feasible in LPT research and (b) that RCT’s are superior to cohort studies in all important methodological aspects.

8.4.1 The feasibility of conclusive RCT’s

In RCT’s randomization warrants unsurpassed internal validity, but not only the treatment condition but also the control condition must be acceptable, otherwise the study is impracticable. Acceptability of the control condition is often difficult enough in SPT studies. In LPT studies the problem, as a rule is almost insurmountable. The consequences are considerable. First, the more limited the acceptability of the control condition, the smaller the percentage of the study population that will agree to the study conditions, the larger the selection bias. But our main argument regards another aspect. Conclusive control conditions (no treatment, waiting list, placebo) are, as far as LPT is concerned, unacceptable. More acceptable control conditions (other treatment modalities), if they exist at all, are less conclusive. All too often, researchers have to choose between either hardly feasible designs with high conclusive powers or passably feasible designs with low conclusive power. so far, literature has paid insufficient attention to the limited acceptability of control conditions to patients and the subsequent problems for RCT’s. There is an extensive literature on patients dropping out after treatment has started. There is also a good deal of literature on patients refusing to actually start with the treatment they have been allotted to after randomization (the difference between Intention-to-treat design and Per Protocol design). There are, however, almost no publications regarding patients who refuse to participate in a study because they do not accept the possibility of being allotted to the control condition. the problems outlined before are not limitations of the treatment method but of the research method. The factors determining the limitations relate to the duration, not to the type of treatment. They are by no means specific to LPT research but apply in general to research into long-term treatments, whether they are of a somatic or a psychological nature.

133 8.4.2 The methodological qualities of RCT’s and cohort studies

There are general quality criteria applying to intervention research. Some of them have been mentioned before (sub 8.3.3). Neither RCT’s nor cohort studies can meet all criteria, the first because of randomization (at the expense of external validity), the last because of lack of randomization (at the expense of internal validity). Apart from these inherent limitations, RCT’s and cohort studies in actual practice vary in quality and are both able to meet all other criteria for intervention studies. We discussed two specific problems of cohort studies. The inclusion of a control group enhances the credibility of cohort studies. The greater the number of variables included in the matching, the more decisive power the study has. However, the relevance of matching must not be overesti- mated. After all, it is never certain whether all relevant factors have been taken into account in the matching. furthermore, the conclusive power of cohort research is codetermined by the ‘control’ knowledge of the ‘natural’, i.e., ‘untreated’ course of the disorders studied. Progress in epidemiological research is badly needed but there are unresolved, possibly insoluble problems. People that suffer, understandably seek help and in our society they often get it. Consequently, the natural, untreated course of many disorders suitable to be treated with LPT is hardly known. There is still another problem if data regarding the general population are gathered. This group differs from the group treated in a cohort study in one essential aspect (possibly an important confounder): help-seeking behaviour. Where scientific arguments make default, clinical experience retains its rights, or stated differently: for lack of epidemiological knowledge regarding the natural course of the disorders suitable for LPT treatment, one has to accept the guidance of clinical experience. LPT seems mostly indicated for long-existing problems. Leichsenring & Leibing (2003) conclude in their review that ‘several studies reported more improvement in personality disorders after longer treatment durations’. Kopta et al (1994) also found that improvements in character problems take longer than symptom changes. In the context of stepped care, LPT is also indicated in case SPT appears insufficiently effective. The latter is by no means exceptional, on the contrary; there are strong indications that the effects of short-term therapies are short-lived. In an extensive review Gloaguen et al. (1998) showed that relapse rates after short-term treatment for depression with psychotherapy and pharmacotherapy on average varied from 30% to 60%. Two time honoured adages apply here: (a) The longer the duration of disorders and (b) the more frequently they have been treated to no avail, the less probably they will remit ‘spontaneously’. Time rarely heals old and ineffectively treated wounds.

8.5 Conclusion

Up to now, with the exception of BPD patients, RCT’s constitute an inadequate research methodology for studying LPT effectiveness. In this area RCT’s do not

134 represent the highest but an irrelevant level of empirical evidence. Where RCT’s are an impracticable and therefore inadequate research method, cohort studies provide the best available evidence.

References

Bateman A., Fonagy P. (1999). The effectiveness of partial hospitalisation in the treatment of of borderline personality disorder – a randomized controlled trial. American Journal of Psychiatry, 156: 1563-1569. Bateman, A.W., Fonagy, P. (2000). Effectiveness of psychotherapeutic treatment of personality disorder. British Journal of Psychiatry, 177: 138-143. Center of Evidence Based Medicine Toronto, Glossary. Available at: http://www.cebm. utoronto.ca. Accessed March, 2005. Center of Evidence Based Medicine Oxford, Levels of evidence (2001). Available at: http://www.cebm.net. Accessed March, 2005. Chambless, D.L., Hollon, S.D. (1998). Defining empirically supported therapies. Journal of Consulting and Clinical Psychology, 66 (1), 7-18. Concato, J., Shah, N., Horwitz, R.I. (2000) Randomized controlled trials, observational studies and the hierarchy of research designs. New England Journal of Medicine, 342 (25), 1887-1892. Crits-Christoph, P. (1992). The efficacy of brief dynamic psychotherapy: a meta-analysis. American Journal of Psychiatry, 149: 151-158. Crits-Christoph, P. (1997). Limitations of the “dodo bird” verdict and the role of clinical trials in psychotherapy research: comments on Wampold et al. (1997) Psychological Bulletin, 122: 216-20. Gloaguen, V., Cottraux, J., Cucherat, M., Blackburn, I. (1998). A meta-analysis of the effects of cognitive therapy in depressed patients. Journal of Affective Disorders, 49: 59-72. Healy, D. (2001). The dilemmas posed by new and fashionable treatments. Advances in Psychiatric Treatment, 7: 322-327. Hollon, S. (1996). The efficacy and effectiveness of psychotherapy relative to medications. American Psychologist, 51: 1025-30. Howard, K.I., Moras, K., Brill, P.L., Martinovich, Z., Lutz, W. (1996). Evaluation of psychotherapy: efficacy, effectiveness, and patient progress. A, Psychol., 51: 1059-64. Kaptchuck, T.J. (2001). The double-blind randomized, placebo-controlled trial: gold standard or golden calf? Journal of Clinical Epidemiology, 54: 541-549. Kazin, A.E. (1998) Research design in clinical psychology. Boston: Allyn and Bacon. Kendall, P.C. (1998). Empirically supported psychological therapies. Journal of Consult- ing and Clinical Psychology, 66: 3-6. Kernberg, O.F., Clarkin, J.F., Yeomans, F.E. (2002). A Primer of Transference Focused Psychotherapy for the Borderline Patient. New York: Jason Aronson. Kopta, S.M., Howard, K.I., Lowry, J.L., Beutler, L.E. (1994). Patterns of symptomatic recovery in psychotherapy. Journal of Consulting an Clinical Psychology, 62: 1009- 1016. Kunz, R., Oxman, A.D. (1998). The unpredictability paradox: review of empirical comparisons of randomized and non-randomized clinical trials. British Medical Journal, 317: 1185-1190. Leichsenring, F., Leibing, E. (2003). The effectiveness of psychodynamic therapy and cognitive behavior therapy in the treatment of personality disorders: a meta-analysis. American Journal of Psychiatry, 160: 1223-1232. Leichsenring, F., Rabung, S., Leibing, E. (2004). The efficacy of short-term psychodynamic psychotherapy for specific psychiatric disorders: a meta-analysis. Archives of General Psychiatry, 61 (12), 1208-1216.

135 Leichsenring, F. (2004). Randomized controlled versus naturalistic studies: A new research agenda. Bulletin of the Menninger Clinic, 68: 137-151. Linehan, M.M., Armstrong, H.E., Suarez, A., Allmon, D, Heard, H.L. (1991). Cognitive- behavioral treatment of chronically parasuicidal borderline patients. Archives of General Psychiatry, 48: 1060-1064. Linehan, M.M., Schmidt, H., Dimeff, L.A., Craft, J.C., Kanter, J, Comtois, K.A.(1999). Dialectic behavior therapy for patients with borderline personality disorder and drug-dependence. American Journal of Addiction, 8: 279-292 Linehan, M.M., Dimeff, L.A., Reynolds. S.K., Comtois, K.A., Welch, S.S., Kivlahan, D.R. (2002). Dialectic behavior therapy versus comprehensive validation therapy plus 12-step for the treatment of opioid dependent women meeting criteria for borderline personality disorder. Drug Alcohol Dependence, 67: 13-26. Luborsky, L., Diguer, L., Seligman, D. (1999). The researcher’s own therapy allegiances: a ´wild card´ in comparisons of treatment efficacy. Clinical Psychology: science and practice, 6: 95-106. Marks, H.M. (1997). The progress of experiment. Science and therapeutic reform in the United States 1900-1990. New York; Cambridge University Press. Perry, J.C., Banon, E., Iannni, F. (1999). Effectiveness of pychotherapy for personality disorders. American Journal of Psychiatry, 156: 1312-21. Sachs H, Chalmers TC, Smith H Jr. (1982). Randomized versus historical controls for clinical trials. American Journal of Medicine, 72, 233-40. Shadish, W.R., Cook, T.D., Campbell, D.T. (2002). Experimental and quasi-experimental designs for generalized causal inference. Boston: Houghton Mifflin Company. Scheel, K.R. (2000). The empirical basis of dialectic behavior therapy; summary, cri- tique, and implications. Clinical Psychology: Science and Practice, 7: 68-86. Seligman, M.E.P. (1995). The effectiveness of psychotherapy: the consumer report study. American Psychologist, 50: 965-74. Spijker, J., de Graaf, R., Bijl, R.V., Beekman, A.T., Ormel, J., Nolen, W.A. (2000). Dura- tion of major depressive episodes in the general population: results from The Neth- erlands Mental Health Survey and Incidence study. British Journal of Psychiatry, 181, 208-213. Stirman, S.W., DeRubeis, R.J., Crits-Christoph, P., Brody, P.E. (2003). Are samples in randomized controlled trials of psychotherapy reptresentative of communituy outpatients? A new methodology and initial findings. Journal of Consulting and Clinical Psychology, 71: 963-972. Task Force on Promotion and Dissemination of Psychological Procedures. (1995). Train- ing and dissemination of empirically-validated psychological treatments. Behavioral Research & Therapy, 39, 299-308. Verheul, R., van den Bosch, L.M.C., Koeter, M.W.J., De Ridder, M.A.J., Stijnen, T., van der Brink, W. (2003). Dialectical behaviour therapy for women with borderline personality disorder. British Journal of Psychiatry, 182: 135-140. Westen, D., Novotny, C.M., Thompson-Brenner, H. (2004). The empirical status of empirically supported psychotherapies: assumptions, findings, and reporting in controlled clinical trials. Psychological Bulletin, 130: 631-663. Young, J.E., Klosko, J.S., Weishaar, M.E. (2003) Schema Therapy. A Practitioner’s Guide. New York, London: The Guilford Press Zanarini, M.C., Frankenburg, F.R., Hennen, J., Silk, K.R. (2003). The longitudial course of borderline psychopathology. American Journal of Psychiatry, 160: 274-283.

136 Chapter 9 The effectiveness of long-term psychoanalytic therapy: a systematic review of empirical studies

Saskia de Maat, Frans de Jonghe, Robert Schoevers, Jack Dekker Submitted

Abstract

Background: Long-term PsychoAnalytic Therapies (LPaT) include both psychoanalysis and long-term psychoanalytic psychotherapy (henceforward called psychotherapy). The effectiveness of LPaT is still a matter of debate. Most reviews of LPaT effectiveness research do not include a systematic quality check of the individual studies and do not pool their data. Aim: To present a systematic review of studies dealing with LPaT effectiveness and published from 1970 onwards. To carry out a systematic quality check of the studies and to pool their data. Methods: A systematic literature search was performed for studies dealing with the effectiveness of individual LPaT in ambulatory, adult patients. A quality check was carried out for each study. Data about the overall effectiveness of LPaT, its impact on symptom reduction and its effect on personality changes were pooled separately, both at treatment termination and at follow- up. Results are presented as weighted mean effect sizes (ESs) and weighted mean success rates. Results: We found twenty-seven studies with a total population of 5,063 patients. All but one were cohort studies. Nineteen studies (70%) met our quality criterion. psychotherapy yielded large mean ESs (0.78 at termination, 0.94 at follow- up) and high mean overall success rates (67% at termination, 55% at follow- up) in patients with moderate pathology. The mean ES was larger for symptom reduction (1.05) than for personality change (0.57). In severe pathology, the results were similar, but it was only possible to calculate the mean ES for personality change (1.09). Psychoanalysis achieved large mean ESs (0.96 at termination, 1.18 at follow-up) and high mean overall success rates (70% at termination, 54% at follow-up) for moderate pathology. The mean ES for symptom reduction was larger (1.23) than for personality change (0.83). The available data about severe pathology was insufficient. Assessments of success from therapists and patients were in broad agreement. They thought that there was more symptom reduction than personality change.

137 Conclusions: LPaT effectiveness has been extensively investigated, mainly in cohort studies of varying quality. Our data suggest that it is an effective treatment for a large range of pathologies, with moderate to large effects that last for years after treatment termination. Subsequent research in cohort studies could, and should, be improved in several ways.

9.1 Introduction

The effectiveness of Long-term PsychoAnalytic Therapies (LPaT, consisting of psychoanalysis ‘proper’ and long-term psychoanalytic psychotherapy) has been studied from the very beginning, for instance by Coriat (1917), Fenichel (1930), Kessel and Hyman (1933), Jones (1936), Alexander (1937), Knight (1941), Schjelderup (1955) Orgel (1958), Knapp et al. (1960), Graham (1960), Klein (1960), Cremerius (1962), Bieber et al. (1962), Hamburg et al. (1967) and Feldman (1968). These studies can be seen as the first attempts to provide empirical evidence of LPaT effectiveness. However, they do not meet contemporary criteria of scientific research, and so their actual relevance is exclusively historical. In recent decades, efforts have been made in order to bring LPaT effectiveness research more into line with current standards of evidence-based medicine. There are several relatively recent reviews and over- views in this field. We mention only a few. Bachrach et al. (1991) performed an extensive review of literature dealing with the effectiveness of psychoanalysis. They discussed at length the methodology, design and results of six systematic studies, including a total of 550 patients. The authors found success rates in the 60 to 90 percent range, and significant effect sizes. Doidge (1997) conducted an overview of the empirical evidence for psychoanalytic psychotherapies and psychoanalysis. He concluded that: ‘There is considerable experimental and clinical data for efficacy of a range of individual psychoanalytic psychotherapies, from short-term psychoanalytic psychotherapy to long-term psychoanalytic psychotherapy and psychoanalysis’. The Open Door Review (Fonagy, 1999) described the study designs and outcomes of scores of LPaT studies. Its importance is not easily overrated. It covered both process and outcome studies in a wide variety of patient groups. However, the authors, probably because of the broad range of these studies, did not pool the data or aim to reach general conclusions. The clinical heterogeneity of the studies would certainly have precluded the latter. Fonagy evaluated each study separately and listed the strengths and weaknesses of the designs, but the methodological quality of the studies was not systematically assessed or expressed in a quality score. Doidge (2001) evaluated nine quantitative studies of psychoanalysis and concluded that ‘for the properly chosen patient, psychoanalysis is effective in terms of symptom relief, character changes, and conflict resolution.’ The review of Leichsenring (2005) dealt with short- term psychodynamic psychotherapy and long-term psychoanalytic therapy. He identified 22 RCTs, all of which dealt with short-term psychotherapy. Furthermore, he reviewed four studies of long-term psychoanalytic therapy

138 that met the highest quality requirements for naturalistic effectiveness studies (Leichsenring, 2004). He concluded that LPaT was more effective than shorter forms of psychodynamic psychotherapy and that it yielded effect sizes that significantly exceeded the effects of untreated or low-dose-treated comparison groups. The strength of his study was that it included only the highest level of evidence for effectiveness studies, but the data from individual studies were not pooled. in short, the more recent reviews are of good quality, but they do not pool the data (e.g. calculate mean ESs or mean success rates) from individual studies. Although some reviews explicitly address study quality, they do not use systematic quality scores. the aim of this article is to present a systematic review of studies published from 1970 onwards. To enhance clinical homogeneity, we focus on individual, ambulatory LPaT with adult patients with ‘regular’ indications for psychoanalytic therapy. We perform a systematic assessment of study quality using an explicit quality criterion. Furthermore, we attempt to pool the data about the effectiveness of LPaT, both for symptom reduction and personality changes. Finally, we compare the results of studies meeting our quality criterion with the results of lower quality studies.

9.2 Long-Term Psychoanalytic Therapies

We define ‘long-term’ rather arbitrarily as therapy consisting of at least 50 sessions and lasting at least one year. Psychoanalysis is always long-term; psychotherapy can be short-term or long-term. LPaT, therefore, consists of psychoanalysis and long-term psychoanalytic psychotherapy. the shared feature of these treatments is that they are rooted in psychoanalytic theories. We differentiate between the two treatments according to principles generally accepted by psychoanalytic therapists. Two easily identifi- able aspects, therapy setting and session frequency, discriminate between these types of LPaT. In psychoanalysis ‘proper’ the patient lies on a couch and there are at least three sessions a week. In psychotherapy, patients sit across from the therapist and there are no more than two sessions a week. Interpretation is the hallmark of psychoanalysis; psychotherapy moves on a continuum between the poles of interpretation and support. the ultimate goals of LPaT are symptom reduction, prevention of recurrence, better social functioning, higher quality of life and higher life satisfaction, preferably for a long time after treatment termination. These goals are by no means specific to LPaT. The specificity lies in the intermediate goals, which focus on bringing about changes in some aspects of the personality of the patient. These changes are meant to be lasting, and they should enable the patient to meet the problems of living more successfully and make better use of their personal potential and the opportunities afforded by their lives. In other words, people’s vulnerabilities are reduced, and their strengths and resources are developed. The aim of LPaT is to stimulate this development. Successful

139 LPaT does not, by definition, make a patient happier; it enhances the possibility that patients will be happy when there is reason to be. It also enhances the possibility of there being such a reason, since people are (up to a certain point) the creators of their own circumstances, and certainly of their attitudes towards those circumstances. In psychoanalytic terms, the changes in personality are described as: ‘structural change’, ‘personality change’, ‘personality reconstruction or construction’, or the development of a ‘cohesive’, ‘adult’, ‘integrated’ self. Freud (1919) put it ‘simply’ when stating that: ‘So vollzieht sich bei dem analytisch Behandelten die Psychosynthese (…)’.

9.3 Methods

9.3.1 Literature search strategy

A literature search was conducted in Pubmed, Embase, the Cochrane Database of Systematic Reviews, the Cochrane Central Register of Controlled Trials, Psy- chLit and the ACP Journal Club. The following subject headings were applied: Long-term psychotherapy, Psychoanalysis, Psychoanalytic psychotherapy, Psy- chodynamic therapy. The time limit was 1970 and there were no limits set on language. Cross-references in the retrieved publications were tracked down. The Open Door Review (Fonagy, 1999) was an important source of studies regarding psychoanalytic therapies. There was no systematic effort to find additional unpublished data.

The following inclusion criteria were applied: 1. the studies were required to be ‘outcome-intervention studies’. The outcomes had to be measured in terms of symptoms or personality changes. Issues such as process variables or therapeutic variables were excluded from this review. 2. the studies were required to be RCT or cohort studies. Case studies or case series were excluded, but surveys were included. 3. the studies were required to deal with individual, ambulatory psychoanalytic therapies with adult patients (18 to 65 years of age). Studies dealing with children or the elderly and studies conducted in clinical settings (Bateman & Fonagy, 1999) or day-care settings were excluded from the review. 4. the studies were required to include patients with regular indications for psychoanalytic therapies (i.e. DSM diagnoses (axis I and II) or otherwise specified personality or symptom neuroses). Very specific disorders such as schizophrenia (Stanton et al., 1984), somatic disorders such as juve- nile-onset insuline-dependent diabetes (Fonagy en Moran, 1990), studies focusing exclusively on eating disorders (Quiroga et al., 1998) or more rare diagnoses such as Münchhausen-by-proxy, and learning difficulties with children (Heinicke, 1965) were excluded. 5. treatment was required to last a minimum of one year and to consist of at least 50 sessions.

140 9.3.2 Identification of relevant publications and quality assessment

Two reviewers checked the search results (SdM and FdJ) and all potentially suitable studies were requested. SdM and FdJ conducted independent checks of all identified studies. Any disagreements about whether a study should be included were resolved by discussion with a third member of staff (JD). Two reviewers (SdM and FdJ) evaluated the quality of the studies independently using a self developed Research Quality Score (RQS). This list and its scoring method can be found in Appendix 1. The list is a first draft and has not yet been tested. It is based on the criteria postulated by the Cochrane Collabo- ration (Cochrane Reviewers Handbook, Cochrane Collaboration, 2004) and Leichsenring (Leichsenring, 2004) and deals with aspects of the study design, patients included, interventions, outcome data, statistics, dropout and follow- up. We distinguished between three designs, each with their own criteria (which overlapped extensively) and maximum RQS: single or multiple cohort studies, cohort studies with matched control groups and RCTs. Each design can be a well-designed and well-performed study meeting all possible quality criteria. There is also a hierarchy in the quality of evidence, with RCTs coming first and single cohort studies last. A matched control group was considered to be either an untreated matched group with highly similar characteristics to the intervention group or a matched group that was treated with therapy that was already evidence-based. We did not consider cohort studies with both psychotherapy and psychoanalysis groups to be studies with matched control groups since we do not think these groups act as controls for each other. In our opinion, the indications ‘psychoanalysis’ or ‘psychotherapy’ are based on reasons that make the two patient groups inherently different (e.g. ego strength, affect tolerance, capacity for reflection and insight). We therefore classified these studies as multiple cohort studies. The quality criterion for each study design was half of the maximum possible number of points.

9.3.3 Effectiveness data and calculations of results

1. We first summarized the findings of studies meeting our quality criterion. The primary outcome data were pre-post and pre-follow-up effect sizes (ES, Cohen, 1988). We considered these to be the most ‘robust’ outcome measures, firstly because they require pre-treatment and post-treatment assessments using a specific questionnaire, and secondly because their standardisation allows for comparison. Effect sizes of less than 0.5 are considered small, the range 0.5-0.8 to be medium, and effect sizes exceeding 0.8 to be large (Cohen, 1988). When a study presented more than one effect size, a mean ES for the study was first calculated, followed by a mean ES weighted on the basis of the sample size of the study for each table. As some outcome measures may be more sensitive to change than others (Leichsenring & Leibing, 2003; Rosenthal, 1991), we also calculated separate mean ESs for scales concerning symptoms (i.e. the 90-item Symptom Check List (SCL-90, Derogatis et al., 1974)) and for scales considered to measure aspects of personality (i.e. Sense of Coherence Scale

141 (SOCS, Antonovsky, 1987); Social Adjustment Scale (SAS, Weissman & Bot- twell, 1976)). success rates were adopted as the secondary outcome measure. They consist of appraisals from patients and therapists of therapy success. We consider measures to be less ‘robust’, since they consist of more general subjective assessments (often based on five- or seven-step Likert scales like a Clinical Global Assessment (CGI, Guy, 1976)). These scales do not always differentiate between symptoms or personality changes. However, some studies differentiate between success rates based on symptom scales like the SCL-90 and success rates measured on scales relating to personality aspects, such as capacity to enjoy, interpersonal capacities, capacity to deal with conflicts and interpersonal relationships. So we first divided the results into general/symptom- related assessments and personality-related assessments. We broke down these results according to whether they represented therapist or patient assessments. Secondly, we calculated overall success rates, pooling all success rates. Scales and criteria varied per study. For this review, we defined ‘success’ as at least moderate improvement. This means that ‘success’ in this review varied from moderate to excellent improvement on a certain scale. Weighted mean success rates were calculated for each table. 2. We summarized the findings of lower quality studies in order to compare them with those of the studies meeting the quality criterion.

9.3.4 Breakdown of the results

We broke down our results using five types of differentiation: 1. psychoanalysis ‘proper’ and psychoanalytic therapy (referred to hereafter as ‘psychotherapy’). 2. ‘therapist assessment’ and ‘patient assessment’. 3. pre-post outcomes and pre-follow-up outcomes. ‘Post’ is defined as ‘at treatment termination’, follow-up as a given period after treatment termination. 4. results relating to symptom reduction (e.g. SCL-90) and results relating to personality change (e.g. SOCS, Inventory of Interpersonal Relationships (IIP, Horowitz, Strauss,& Kordy, 2000)). 5. ‘moderate’ pathology and ‘severe’ pathology. By ‘moderate pathology’ we mean the ‘regular’ indications for ambulatory LPaT as indicated by the extensive survey of Doidge et al. (2002). The authors conducted a survey of 510 analysts in the USA, Canada and Australia. The results showed that 71% of the patients had a DSM-III-R diagnosed personality disorder, 68% had a major depressive disorder and 57% an anxiety disorder. Only 3% of the patients did not present a DSM-III-R diagnosis. In addition, comorbidity was high, with the modus of patients presenting three disorders. Although these clinical diagnostic pictures are fairly complicated, we labeled pathology in patients treated in ambulatory psychoanalytic practices to be ‘moderate’, in other words to conform to the findings of Doidge et al. (2002). We classified studies as being related to ‘severe’ pathology when they included

142 only patients with more severe personality disorders such as borderline personality disorder.

9.4 Results

A total of 742 studies was initially found. Based on screening of titles and abstracts 672 studies were excluded in the first selection round. Reasons for exclusion were: patients samples based on a somatic illness and not a psychiatric illness, case studies, theoretical articles, epidemiological studies, studies dealing with training of professional therapists, studies dealing with alcohol and drugs, patient samples concerning only geriatric patients, very specific diagnoses such as Münchausen-by-proxy, articles dealing with methodological issues. The remaining 73 articles were retrieved and a further exclusion of 45 studies followed in the second selection round. Reasons for exclusion in this round were: process-outcome studies, studies focusing on children, reviews, studies on short-term therapies, studies focusing on therapists variables, studies on inpatients. Finally, 27 studies outcome studies were retrieved of LPT in regular ambulatory patients. Table 9.1 presents an overview of all relevant studies we found. Tables 9.2-4 show the results of the 19 studies that met the quality criterion. Table 9.5 presents the results of the remaining 8 studies. Finally, tables 9.6 summarizes all findings.

9.4.1 Included studies

As can be seen in table 9.1, we found 27 relevant publications meeting our inclusion and exclusion criteria. These studies included 1 RCT, 5 surveys and 21 cohort studies. Of the cohort studies, 16 were prospective and 5 retrospective. Four surveys (Hartmann & Zepf, 2004; Freedman et al., 1999; Heinzel et al., 1998, Keller et al., 1997) were replications of the Consumer Report performed by Seligman (1995). The survey of Stehle (2004) was based on inquiries among therapists. In all, the studies covered 5,063 patients. The mean number of sessions was approximately 497 (3.6 years) in psychoanalysis and 192 (3.2 years) in psychotherapy. The vast majority of the studies looked at patients with ‘moderate’ pathology; only four studies (Wallerstein, 1985; Stevenson & Meares, 1992; Giesen-Bloo et al, 2006 and Monsen et al., 1995) were considered to address ‘severe pathology’. In the study of Keller et al. (1997) we categorized the psychoanalysis condition (n=84) under the heading of ‘psychotherapy’ because the mean frequency of the sessions was 1.6 per week.

9.4.2 Study quality

A total of eight studies (30%) did not meet the quality criterion (Appendix 1): Dührssen, 1986; Erle et al., 2003, 1984; Freedman et al., 1999; Friedman et al., 1998, 2005; Hartman et al., 2004; Heinzel et al., 1995; Sashin et al., 1975 and Stehle et al., 2003. Six of these studies were retrospective studies, and five

143 Table 9.1: Studies included in review. Author Study type Treatments¹ N Diagnosis Duration² Outcome Measurements Assessment & RQS source³ of effectiveness points (1) Cohort, Psychotherapy 31 Moderate 3 years Pt: + Symptoms: Pre: + Brockman et prospective pathology: Th: + SCL-90-R, GSI During: + al., 2003 DSM-III-R Ob: + Personality: IIP-D Post: + Brockman, RQS: 38 Depression Other: - Goal attainment: Follow-up: + 2002 and anxiety Goal Attainment disorders Scale (100%) (2) Cohort, Psychotherapy 23 Severe 1 year Pt: - Symptoms: Pre: + Clarkin et al., prospective pathology: Th: + Parasuicidal history During: - 2001 Borderline Ob: + interview, treatment Post: + RQS: 41 Personality Other: - history interview. Follow-up: - Disorder plus Undifferentiated: GAF more than one Axis I disorder (3) Cohort, Psychotherapy 60 Moderate Unspecified. Pt: + Symptoms: FPI, Pre: + Dührssen, prospective pathology: Th: - Giessen test, Giessener During: - 1986 Psychoanalysis 30 unspecified Ob: - Beschwerdebogen Post: + RQS: 24 Other: + Personality: GT-scale Follow-up: - (4) Cohort Psychoanalysis 161 Moderate 5 to 6 years Pt: - Undifferentiated: CGI Pre: - Erle et al., Retrospective pathology: Th: + During: - 2003 I (’73-’77) unspecified Ob: - Post: + New York Other: - Follow-up: - study I RQS: 18 (5) Cohort Psychoanalysis 92 Moderate 5 years Pt: - Undifferentiated: CGI Pre: + Erle et al., Prospective pathology: Th: + During: + 2003 II (’84-’89) unspecified Ob: - Post: + New York Other: - Follow-up: - study II RQS: 28 (6) Survey, Psychotherapy 99 Moderate 1 year Pt: + Treatment satisfaction: Pre: - Freedman et retrospective pathology: Th: - Effectiveness During: + al., 1999 Personality Ob: - Questionnaire Post: - IPTAR RQS: 18 disorder (12%), Other: - Follow-up: + depression (58%), anxiety (17%). (7) Cohort, Psychotherapy 575 Moderate 1 year Pt: - Symptoms: structured Pre: + Friedman et retrospective pathology: Th: + interview During: + al., 1998 88% at least 1 Ob: - Personality: structured Post: - Friedman et RQS: 21 axis 1 disorder. Other: - interview Follow-up: + al., 2005 59% at least 1 Undifferentiated: GAF axis II disorder. (8) RCT Psychotherapy 42 Severe 3 years Pt: - Symptoms and Pre: + Giesen-Bloo pathology: Th: + personality: psycho- During: + et al., 2006 RQS: 77 BPDSI-IV Ob: + and personality Post: + Borderline Other: - pathology. Follow-up: + Personality Personality: BPDSI-IV disorder and others. Life satisfaction: Euroqol, WHOQOL

144 (9) Cohort, Psychotherapy 27 Moderate 2 years Pt: + Symptoms: SCL-90, Pre: + Grande et prospective pathology: Th: + GSI, PSKB-Se During: - al., 2006, 55% psycho- Ob: + Personality: IIP, Post: + Rudolf et al., neurosis, 31% Other: + Reconstruction scale Follow-up: + 2004 RQS: 48 Psychoanalysis 31 personality 2.5 years Social functioning: + PAL study disorder. Life satisfaction: + (Heidelberg- Treatmentsatisfaction: + Berlin) (10) Survey, Psychotherapy 448 Moderate 2 years Pt: + Treatment satisfaction: Pre: - Hartmann et retrospective pathology: Th: - Effectiveness During: + al., 2004 unspecified. Ob: - Questionnaire Post: - RQS: 16 Psychoanalysis 263 Other: - Follow-up: + (11) Suvey, Psychotherapy 633 Moderate 3.4 years Pt: + Symptoms and Social Pre: + Heinzel et retrospective pathology: Th: - functioning: 5-point During: - al., 1998 unspecified. Ob: - rating scale. Post: + RQS: 22 Other: - Follow-up: + (12) Cohort, Psychotherapy 37 Moderate 1.5 years Pt: + Symptoms: CGI-I Pre: + Holm-Hadulla prospective pathology: Th: + Personality: CGI-I During: - et al., 1997 Personality Ob: - Symptoms and social Post: + RQS: 38 disorder (27%) Anders: - functioning (global): Follow-up: - Dysthymic (32%) 21-item questionnaire Adjustment (35%) (13) Survey, Psychotherapy 111 Moderate 2.7 years Pt: + Symptoms: SCL-90, Pre: - Keller et al., retrospective pathology: Th: + BSS During: - 1998 Personality Ob: + Personality: Post: - Berlin Jungian RQS: 29 disorder (17%) Other: - Giessen-test Follow-up: + Study (BJS) Depression (46%) (14) Cohort, Psychoanalysis 36 Moderate 3 years Pt: + General well-being: Pre: + Leichsenring prospective pathology: Th: + VEV During: + et al., 2005 Personality Ob: - Symptoms: CGI-I, Post: + RQS: 44 disorder (69%) Other: - SCL-90 Follow-up: + Göttingen affective (69%) Personality: CGI-I, IIP study phobic (50%) Social functioning: CGI-I Life satisfaction: CGI-I, FLZ (15) Cohort, Psychotherapy 194 Moderate 4 years Pt: + Symptoms: SCL-90, Pre: - Leuzinger- retrospective pathology: Th: + BSS, SOFAS During: - Bohleber et Personality Ob: + Personality: SOC Post: - al., 2001, 2003 RQS: 31 Psychoanalysis 207 disorder (51%) 4 years Other: - Social functioning: + Follow-up: + affective (27%) Life satisfaction: IRES Treatment satisfaction: + (16) Cohort, Psychoanalysis 17 Moderate 3 years Pt: + Symptoms: HSRS, Pre: + Luborsky et prospective pathology: Th: + Undifferentiated: GAF During: + al., 2001 unspecified. Ob: + Treatment satisfaction: Post: + Penn study RQS: 30 Other: - SSI Follow-up: + (17) Cohort, Psychotherapy 25 Severe 2 years Pt: + Symptoms: HSRS, Pre: + Monsen et al., prospective pathology: Th: + SCL-90 During: - 1995, 1995 SCID Ob: + Personality: MMPI, Post: + RQS: 49 Personality Other: - Affect cons., Millons. Follow-up: + disorder (96%), Social functioning: depression (52%), level of social anxiety (20%) adjustment.

145 (18) Cohort, Psychotherapy 33 Moderate 3.7 years Pt: + Symptoms: CGI-I Pre: + Von Rad et Prospective pathology: Th: + Personality: Giessen During: + al., 1998 58% has a Ob: + test, CGI-I Post: + Heuft et al., RQS:42 Psychoanalysis 36 psychiatric 2.2 years Other: - Social functioning:+ Follow-up: + 1996, disorder. Treatment Kordy et al., satisfaction:+ 1988 Heidelberg Studie (19) Cohort, Psychotherapy 56 Moderate 1.5 years Pt: + Symptoms: Global Pre: + Rudolf et al., Prospective pathology: Th: + impression During: - 1994 Psychoneurosis Ob: - Personality: Global Post: + Berliner RQS: 32 (54%), Other: - impression Follow-up: - Psycho- Psychoanalysis 44 Personality 1.7 years Social functioning: therapie disorder (20%) PSKS-Se Studie Psychosomatic (13%) (20) Cohort, Psychoherapy 331 Moderate 4 years Pt: + Symptoms: SCL-90, Pre: + Sandell et al, Prospective pathology: Th: - Personality: Sense of During: + 2000 unspecified. Ob: - Coherence Scale Post: + STOPP- RQS: 35 Psychoanalysis 74 4.5 years Other: - Social functioning: Follow-up: + project Social Adjustment Scale, Caseness criterium: on 3 scales in worst 10% (21) Cohort, Psychoanalysis 183 Moderate 4.2 years Pt: - Symptoms: CGI Pre: + Sashin et al., retrospective pathology. Th: + Personality: CGI During: - 1975 (1959-1966) Hysteric Ob: - Social functioning: CGI Post: - Boston study character Other: - Follow-up: + RQS: 21 neurosis (some with depression) 30%, Obsessive- compulsive neurosis: 18% (22) Survey, Psychotherapy 581 Moderate 3 years Pt: - Symptoms: CGI-I Pre: - Stehle et al., retrospective pathology: Th: + Personality: CGI-I During: - 2004 60% neurosis, Ob: - Post: + RQS: 22 31% personality Other: - Follow-up: - disorder (23) Cohort, Psychotherapy 48 Severe 1 year Pt: + Symptoms: Cornell Pre: + Stevenson Prospective pathology: Th: - Index During: - & Meares, Borderline Ob: + Post: - 1992 RQS: 48 personality Other: + Follow-up: + New South disorder (100%) Wales study (24) Cohort, Psychotherapy 20 Severe Unspecified. Pt: - Symptoms: HSRS Pre: + Wallerstein, Prospective pathology: Th: + Personality: CGI-I During: - 1986, Majority Ob: - Post: + Kernberg, RQS: 54 Borderline Other: - Follow-up: - 1972, Harty, Psychoanalysis 22 Personality 1976, disorder. Menninger Project

146

(25) Cohort, Psychoherapy 138 Moderate 0.75 year Pt: - Symptoms: CGI-I Pre: + Weber et Retrospective pathology: Th: + Personality: CGI-I, During: - al., 1985 (II), unspecified. Ob: - ego strength scale Post: + Bachrach et RQS: 27 Psychoanalysis 235 2.5 years Other: - Social functioning: Follow-up: - al., 1985 social relations scale, Columbia work gratification scale study I (26) Cohort, Psychotherapy 29 Moderate 2 years Pt: - Symptoms: CGI-I Pre: + Weber et al., prospective pathology: Th: + Personality: CGI-I, During: - 1985 (III) Psychoanalysis 36 unspecified. 3 years Ob: - ego strength scale Post: + Bachrach et al., RQS: 35 Other: - Social functioning: Follow-up: - 1985 social relations scale, Columbia work gratification scale study II (27) Cohort, Psychotherapy 55 Moderate 3 years Pt: + Symptoms: DSM-III-R Pre: + Wilczek et prospective pathology: Th: - CPRS-S-A During: - al., 2004 53% DSM-III-R Ob: + Personality: KAPP, Post: - RQS: 39 Axis I Other: - KSP Follow-up: + 11% DSM-III-R Axis II ¹ Psychoanalytic psychotherapy (1-2sessions/week), psychoanalysis (3-5 sessions/week). ² Approximation mean duration of therapy. ³ Pt=patient, Th=therapist, Ob=Independent observer.

of them were surveys (sent to either therapists or patients). The fact that the studies were retrospective accounted for many of the low quality scores. Other factors with a negative effect on the RQS of studies were: not stating inclusion and exclusion criteria for patient selection, not describing the characteristics of the patient sample or the treatment applied, no intention-to-treat analyses, no independent or blind assessments, not defining or describing dropout, and moderate quality of assessment instruments (i.e. only a five-step Likert scale addressing overall therapy success). Of the remaining nineteen studies that met the quality criterion, only one was an RCT (Giessen-Bloo, 2006). Two made an attempt at a matched control group (Holm-Hadulla et al., 1997, and Leichsenring, 2005), and the other sixteen were single or multiple cohort studies. Examples of cohort studies meeting most of the quality criteria were the studies of Brockman et al. (2003), Grande et al. (2006) and Leichsenring (2005). The studies were well designed and prospective, patient samples were clearly described, treatments were described, treatment adherence was controlled for (although not in Brockman et al., 2003), assessment instruments were well qualified, there were multiple assessors (although none of the studies included blind assessment), follow-up results were included and the Leichsenring study attempted to compare the sample with a control group.

147 9.4.3 LPT Effect Sizes (ES)

Tables 9.2.1-9.2.3 present the ESs for psychoanalysis and psychotherapy in moderate and in severe pathology. As can be seen in table 9.2.1, the majority of the studies regarding psychotherapy in moderate pathology reported large ESs, both at treatment termination (65%) and at follow-up (69%). The weighted mean ESs were large at both termination (0.78) and follow-up (0.94). We calculated mean ESs separately for symptom reduction and personality change (scales marked ‘s’ and ‘p’, respectively in the table; scales marked ‘–‘ were not included because they provided global assessments). All ESs at treatment termination, as well as at follow-up,

Table 9.2.1: Psychotherapy ESs* in moderate pathology. Pre-post Pre-follow-up Mean es/study Study, N Scale P/S** Small Medium Large Small Medium Large Post Follow-up (years) Brockman, 2002/03. n=31 SCL-90 (GSI) s 1.37 1.28 IIP (relationships) p 1.19 Grande, 2006. n=27 SCL-90 (GSI) s 0.92 1.04 0.80 0.95 IIP (relationships) p 0.67 0.85 (1 year) Holm-Hadulla, 1997. n=37 Global judgement patient - 1.13 1.05 Global judgement therapist - 0.96 Keller, 1998. n=111 BSS scale s 2.10 2.10 (6 years) Leichsenring, 2005. n=36 SCL-90 (GSI) s 1.34 1.38 IIP (relationships) p 1.28 1.85 1.64 1.88 Life satisfaction p 1.55 1.81 (1 year) GAS - 2.39 2.48 Sandell, 2000. n=331 SCL-90 s 0.60 Sense of Coherence Scale p 0.34 0.46 Social Adjustment Scale p 0.44 (3 years) Weber, 1985 II n=138 Ego strength p 0.59 Social Relations p 0.43 0.51 Work gratification p 0.50 Weber, 1985 III, n=29 Ego strength p 0.11 0.11 Wilczek, 2004. n=36 KAPP p 0.30 KSP p 0.40 GAF - 0.87 CPRS Anxiety s 0.99 0.82 CPRS Depression s 1.23 (0.5 years) CPRS Obs. Compul. s 1.10 Percentage of ES 14% 21% 65% 25% 6% 69% Weighted Mean ES (n=776) 0.78 0.94 Mean duration follow-up (3.2 years) * Small = < 0.5. Medium = 0.5-0.8. Large = > 0.8. ** P = scale mostly related to personality aspects, S = scale mostly related to symptoms, - = global outcome.

148 were included in this calculation. The weighted mean ESs were 1.05 for symptom reduction and 0.57 for personality change. as can be seen in table 9.2.2, approximately half of the studies dealing with psychotherapy in severe pathology report large ESs, both at treatment termination (46%) and follow-up (56%). The weighted mean ES at termination (0.94) was large, as was the follow-up ES (1.02). However, the latter is a result from only one study. As the majority of the studies dealt with personality aspects, we did not calculate separate ESs for symptom reduction. The weighted mean ES for personality change, including all ESs at termination and at follow-up, was 1.09. as can be seen in table 9.2.3, which looks at psychoanalysis in moderate pathology, 50% and 80% of the studies reported large ESs at treatment termination and at follow-up respectively. Both the weighted mean ES at termination (0.96) and the ES at follow-up (1.18) were large. We made separate calculations

Table 9.2.2: Psychotherapy ESs* in severe pathology. Pre-post Pre-follow-up Mean es/study Study, N Scale P/S** Small Medium Large Small Medium Large Post Follow-up (years) Clarkin, 2001. n=23 Self-injurious behaviour: number of incidents s 0.15 Medical Risk of incidents - 0.37 0.33 Physical condition after incidents - 0.46 Giesen-Bloo, BPDSI p 1.85 2006. n=42 EuroQol p 0.64 WHOQOL p 1.16 1.12 Psycho- and personality pathplogy p 0.84 Monsen, 1995. n=25 MMPI (depression) p 0.67 0.67 MMPI (anxiety) p 0.78 0.78 MMPI (ego resilience) p 1.45 MMPI (ego strength) p 1.15 MMPI (social withdrawal) p 0.92 0.92 1.20 1.02 Affect consciousness p 2.46 2.30 (5.2 years) Morey PD scale (MMPI) 0.91 (passive-agressive, borderline, antisocial, compulsive, paranoid) p Morey PD scale (MMPI) 0.54 (avoidant, dependent, schizoid) p Morey PD scale (MMPI) 0.49 (histrionic, narcissistic) p Percentage of ES 27% 27% 46% 11% 33% 56% Weighted Mean ES (n=90) 0.94 1.02 Mean duration follow-up (5.2 years) * Small = < 0.5. Medium = 0.5-0.8. Large = > 0.8. ** P = scale mostly related to personality aspects, S = scale mostly related to symptoms, - = other category outcome.

149 Table 9.2.3: Psychoanalysis ESs* in moderate pathology. Pre-post Pre-follow-up Mean es/study Study, N Scale P/S** Small Medium Large Small Medium Large Post Follow-up Grande, 2006. n=32 SCL-90 (GSI) s 1.54 1.58 IIP (relationships) p 1.22 1.30 1.38 1.44 (1 year) Rudolf, 1994. n=44 Somatic anxiety s 1.36 Depression s 0.94 Somatic complaints s 0.80 1.13 Social anxiety s 0.78 Regressive attachment p 0.62 Sandell, 2000. n=74 SCL-90 s 1.55 Sense of Coherence Scale p 1.18 1.06 Social Adjustment Scale p 0.45 (3 years) Weber, 1985II, n=36 Ego strength p 0.39 0.39 Percentage of ES 12.5% 37.5% 50% 20% 0% 80% Weighted Mean ES (n=150) 0.96 1.18 Mean duration follow-up (2.6 years) * Small = < 0.5. Medium = 0.5-0.8. Large = > 0.8. ** P = scale mostly related to personality aspects, S= scale mostly related to symptoms.

Table 9.3.1: Therapist opinions of psychotherapy success* rates. Study/N Scale post follow-up Holm-Hadulla, 1997. n=37 CGI Symptoms 89% n.a. Keller, 1998. n=111 Global assessment of 95% n.a. overall health state Leichsenring, 2005. n=36 CGI 84% n.a. Leuzinger, 2003. n=194 CGI 67% n.a. Rad, von,1998. n=33 CGI 30% 55% (3.5 years)** Rudolf, 1994. n=56 PSKB 44% n.a. Weber, 1985 II, n=138 CGI 67% n.a. Weber, 1985 III, n=29 CGI 66% n.a. n=634 Weighted mean % 70% * Success = at least moderate improvement on the scale concerned. ** Independent observers.

of mean ESs for symptom reduction and for personality change (scales marked ‘s’ and ‘p’ respectively in the table). All ESs at treatment termination and at follow-up were included in this calculation. The weighted mean ESs were 1.23 for symptom reduction and 0.83 for personality change. We did not find ES data for psychoanalysis in severe pathology.

150 Table 9.3.2: Patient opinions of psychotherapy success* rates. Study/N Scale post follow-up Brockman, 2002/03. n=31 GSI of SCL-90 60% n.a. Grande, 2006. n=27 GSI of SCL-90 56% 56% (1 year) Holm-Hadulla, 1997. n=37 CGI Symptoms 85% n.a. Keller, 1998. n=111 Global assessment of 93% n.a. overall health state Leichsenring, 2005. n=36 GSI of SCL-90 77% n.a. Leuzinger, 2003. n=194 CGI 64% n.a. Rudolf, 1994. n=56 PSKB 55% n.a. Sandell, 2000. n=222 No longer meeting 33% n.a. (n= only cases**) caseness criterium** n=714 Weighted mean 69% * Success = at least moderate improvement on the scale concerned. ** Case = scores in worst 10% on SCL, SOC, SAS.

Table 9.3.3: Therapist opinions of psychoanalysis success* rates. Study/N Scale post follow-up Erle, 2003 II, n=92 CGI 66% n.a. Leuzinger, 2003. n=207 CGI 60% n.a. Rudolf, 1994. n=44 PSKB 64% n.a. Von Rad, 1998. n=23 CGI 57% 30% (3.5 years) Weber, 1985 II,n=77** CGI 91% n.a. Weber, 1985 III, n=36 CGI 100% n.a. n=479 Weighted mean 69% * Success= at least moderate improvement on the scale concerned. ** Only end clinic patients.

9.4.4 LPT success rates

9.4.4.1 Success rates: symptom-related or general assessments Tables 9.3.1-9.3.4 present the success rates for psychoanalysis and psychotherapy in moderate pathology on the basis of therapist and patient opinions about symptom reduction and overall health state. as can be seen in tables 9.3.1-9.3.2, in moderate pathology psychotherapy, success rates at treatment termination ranged between 30% and 95% (weighted mean: 70%) according to therapists, and between 33% and 93% (weighted mean: 69%) according to patients. A lack of data meant that follow-up success rates could not be calculated.

151 Turning to psychotherapy success in severe pathology, Monsen et al. (1995) reported a reduction of DSM Axis I disorders of 75% at treatment termination and 83% at follow-up, as rated by independent observers (mean: 79%). In addition, 76% of their patients did not meet the caseness criterion of the GSI of the SCL-90 at follow-up. as can be seen in tables 9.3.3-9.3.4, in psychoanalysis for moderate pathology, success rates at treatment termination ranged between 57% and 100% (weighted mean: 69%) according to therapists, and between 66% and 81% (weighted mean: 73%) according to patients. Lack of data precluded the calculation of follow-up success rates.

9.4.4.2 Success rates: personality-related assessments Tables 9.4.1-9.4.3 present the success rates for psychoanalysis and psychotherapy in moderate pathology, based on therapist and patient assessments of personality-related changes.

Table 9.3.4: Patient opinions of psychoanalysis success* rates. Study/N Scale post follow-up Grande, 2006. n=32 GSI-SCL-90 81% 72% (1 year) Leuzinger, 2003. n=207 CGI 73% n.a. Rudolf, 1994. n=44 PSKB 75% n.a. Sandell, 2000. n=65 No longer meeting 66% n.a. (n=only cases**) caseness criterium** n=348 Weighted mean 73% * Success = at least moderate improvement on the scale concerned. ** Case = scores in worst 10% on SCL, SOC, SAS.

Table 9.4.1: Therapist opinions of psychotherapy success* rates. Study/N Scale post Mean/study Holm-Hadulla, 1997. n=37 CGI Conflicts 60% CGI Reality 49% 55% Leichsenring, 2005. n=36 Interpersonal capacities 53% Capacity to work 53% Capacity to enjoy 53% Capacity to deal with 79% conflicts 60% Range (n=73) 49% - 79% Weighted mean 57% * Success = at least moderate improvement on the scale concerned.

152 Table 9.4.2: Patient opinions of psychotherapy success* rates. Study/N Scale post Mean/study Brockman, 2002/03. n=31 IIP 32%* 32% Grande, 2006. n=27 IIP 50% 50% Holm-Hadulla, 1997. n=37 CGI Conflicts 84% CGI Reality 65% 75% Keller, 1998. n=111 Emotional condition 94% Physical condition 66% Global health 51% Satisfaction job 74% Satisfaction partner 75% 72% Rudolf, 1994. n=56 Somatic problems 60% Mental problems 41% Relational problems 20% 40% Range (n=262) 20% - 94% Weighted mean 59% * Success = at least moderate improvement on the scale concerned.

Table 9.4.3: Patient opinions of psychoanalysis success* rates. Study/N Scale post Mean/study Grande, 2006. n=32 Reliable Change Index 72% 72% IIP Rudolf, 1994. n=56 Somatic problems 80% Mental problems 44% Relational problems 41% 55% Range (n=88) 41% - 80% Weighted mean 61% * Success = at least moderate improvement the scale concerned.

In psychotherapy for moderate pathology (tables 9.4.1-9.4.2), the therapist success rates at treatment termination (weighted mean: 57%) were comparable to those stated by the patients (weighted mean: 59%). With regard to psychotherapy in severe pathology, Giesen-Bloo et al. (2006) reported a success rate of 43% at treatment termination. This was the percentage of patients that achieved the reliable change criterion for Borderline Personality Disorder measured by the BPDI-IV. At follow-up, Stevenson & Meares (1992) found a reduction of 30% in patients meeting the criteria for BPD. In the study of Monsen et al. (1995) 78% of all patients with Axis II diagnoses no longer met the criteria for this diagnosis at treatment termination, and 74% no longer met them at follow-up. The weighted mean of the success rates in these 3 studies was 51% (according to independent observers).

153 there were only two studies of psychoanalysis (table 9.4.3), each represent- ing patient opinions, resulting in a weighted mean of 61%. They dealt with moderate pathology. There were no studies dealing with therapist assessments of personality changes in moderate pathology, and no studies regarding severe pathology.

9.4.4.3 Overall success rates We calculated overall success rates by pooling all rates. In psychotherapy for moderate pathology, the success rates were 67% at termination and 55% at follow-up. In psychotherapy for severe pathology (only 4 studies available: Wallerstein, 1986; Giesen-Bloo et al., 2006; Stevenson & Meares, 1992; and Monsen et al., 1995), the overall success rates were 61% at termination, and 61% at follow-up as well. in psychoanalysis for moderate pathology, the pre-post success rate was 70% and the pre-follow-up success rate was 54%. In psychoanalysis for severe

Table 9.5: Outcomes of excluded studies. Psychotherapy Changes in scores Pre Post Follow-up Pre-Post Pre-F-U Heinzel, 1998 Physical health* 3.20 2.20 2.13 31% 33% General well being* 4.32 2.25 2.16 48% 50% Mental health* 4.45 2.26 2.18 49% 51% Social functioning* 3.67 2.41 2.14 34% 42% Dührssen, 1986 Symptoms (GB) 73** 30*** 41% Personality (GI) 8** 4*** 50% Personality (FPI) 12** 4*** 33% Friedman, 2005 GAF(0-100) 60.36 76.96 28% Success rates Scale Pre-Post Mean/study Stehle, 2004. n=581 CGI general 88% CGI symptoms 85% 81% CGI personality 69% Hartmann, 2004. n=448 CGI success 55% Mean success rate (n=1.029) 69% * scale: 1 = very good, 5 = very bad. ** number of pre-treatment symptoms. *** number of improved symptoms.

154 Psychoanalysis Changes in scores Pre Post Follow-up Pre-Post Pre-F-U Dührssen, 1986 Symptoms 73* 26** 36% Personality (GI) 8* 4** 50% Personality (FPI) 12* 3** 25% Success rates Scale Pre-post Mean/study Erle, 2003. n=161 CGI 74% Sashin, 1975. n=130 CGI Symptoms (discomfort) 61% CGI Sexual adjustment 51% CGI Restriction of life functioning 62% CGI Interpersonal relationships 41% 54% CGI Insight 46% CGI Work productivity 43% CGI Global 75% Hartmann, 2004. n=263 CGI 69% Mean success rate (n=554) 67% * number of pre-treatment symptoms. ** number of improved symptoms. pathology, only one study was found (Wallerstein, 1986). The author reported a 59% success rate at treatment termination.

9.4.5 Outcomes of lower quality studies

Table 9.5 contains the outcomes from studies not meeting the quality criterion. in psychotherapy, pre-post success rates ranged from 55% to 88%, with a mean of 69%. In psychoanalysis, the range was 41% to 75% (mean: 67%). Freedman et al. (1999) and Hartman & Zepf (2004) used a Mean Effectiveness Score (MES). The authors describe this as, in essence, a treatment satisfaction scale, with a maximum of 300 points, 150 points indicating no effectiveness and below 150 indicating negative effectiveness. Freedman et al. (1999) found a mean score of 209; Hartman & Zepf (2004) found a mean score of 237 in psychotherapy and 244 in psychoanalysis.

9.4.6 Summary of results

We summarize all the findings of this review in table 9.6.

155 Table 9.6: Summary of results. 1. Overall Effect Sizes and Success Rates Psychotherapy Psychoanalysis Mean Effect % large Overall Mean Effect % large Overall Size Effect Sizes success rate Size Effect Sizes success rate Moderate pathology (n=9) (n=11) (n=4) (n=8) Termination 0.78 (n=6) 65% 67% (n=11) 0.96 (n=3) 50% 70% (n=8) Follow-up 0.94 (n=5) 69% 55% (n=2) 1.18 (n=2) 80% 54% (n=2) Severe pathology (n=3) (n=4) (n=1) Termination 0.94 (n=3) 46% 61% (n=3) n.a. n.a. 59% (n=1) Follow-up 1.02 (n=1) 56% 61% (n=2) n.a. n.a. n.a.

2. Effect Sizes and Success Rates for symptom reduction and personality change Psychotherapy Psychoanalysis Mean Effect Success rate Success rate Mean Effect Success rate Success rate Size according to according to Size according to according to patient therapist patient therapist Moderate pathology (n=9) (n=8) (n=8) (n=4) (n=4) (n=7) Symptoms 1.05 (n=6) 70% (n=8) 69% (n=8) 1.23 (n=3) 73% (n=4) 69% (n=6) Personality 0.57 (n=7) 59% (n=5) 57% (n=2) 0.83 (n=4) 61% (n=2) n.a. Severe pathology (n=3) (n=1) (n=3*) Symptoms n.a. 76% (n=1) 79% (n=1*) n.a. n.a. n.a. Personality 1.09 (n=2) n.a. 51% (n=3*) n.a. n.a. n.a. n = number of studies n.a. = not available * ratings made by independent observers

9.5 Discussion

We conducted a systematic search of the literature from 1970 onwards for outcome-intervention studies of LPT effectiveness in ambulatory, adult patients. We found twenty-seven relevant studies. We classified eight of them (30%) as ‘lower quality’ studies using a quality criterion based on a self-developed research quality control list. We compiled the results of the nineteen ‘higher quality’ studies by pooling their data. Firstly, we calculated overall effect sizes and success rates. Secondly, we broke down the results according to symptom reduction/personality change, and therapist/patient opinion.

Psychotherapy In moderate pathology, the overall success rates (67% at termination and 55% at follow-up), the percentage of large ESs (65% at termination and 69% at follow-up) and the weighted mean ESs (0.78 at termination and 0.94 at follow-up)

156 indicate substantial effectiveness. This was maintained for years after treatment termination. In severe pathology, taking into consideration that the results mainly related to personality pathology, the percentage of ESs (46% at termination and 56% at follow-up) and the weighted mean ESs (0.94 at termination and 1.02 at follow-up) were certainly not inferior to those in moderate pathology. However, the results in severe pathology should be interpreted cautiously, as they are based on only three studies (follow-up data in only one study), one of them presenting as many as nine ESs. These ESs are likely to be highly corre- lated since they present dimensions from only two separate instruments. When differentiating between symptom reduction and personality changes, it appears that the effects for moderate pathology were more pronounced for the first (ES 1.05) than for the second (ES 0.57). Furthermore, there seemed to be general agreement between patients and therapists about success rates. They both thought the effect on personality change was lower (59% and 57% respectively) than that on symptom reduction (70% and 69% respectively). as only a few studies presented data on severe pathology, our findings were mainly suggestive. We found a large weighted ES for personality change (1.09) and no ES information about symptom reduction. As with the results for moderate pathology, success rates for personality change seemed lower than for symptom reduction.

Psychoanalysis In moderate pathology, the overall success rates (70% at termination and 54% at follow-up), the percentage of large ESs (50% at termination and 80% at follow-up) and the weighted mean ESs (0.96 at termination and 1.18 at follow- up) indicated substantial effectiveness, which was maintained for years after treatment termination. We did not find ESs with regard to psychoanalysis in severe pathology. Only one study provided information about success for severe pathology (Wallerstein, 1986): 59%. after differentiating between symptom reduction and personality changes, we found that, in moderate pathology, the effects were more prominent in symptom reduction (ES: 1.23) than in personality changes (ES: 0.83). As the results are based on only four studies, the results must by interpreted cautiously. Patients and therapists concurred in their opinions about success rates, and the assessment of changes in personality (61% and 51% respectively) seemed to be lower than the assessment of symptom reduction (73% and 69% respectively). The effects on personality change, however, are based on only three studies.

Comparison with outcomes of ‘lower quality’ studies Thirty percent of the retrieved studies did not meet our quality criterion. As only a few of them provided information that could be compared with the results of the higher quality studies, it is not possible to infer strong conclusions. The data suggest that mean success rates do not greatly differ from those in ‘higher quality’ studies. The mean psychotherapy success rate was 69% in the lower quality studies and 67% in the higher quality studies. For psychoanalysis, the success rates were 67% and 67% respectively. Our findings concur with those of Lipsey

157 and Wilson (1993), who concluded on the basis of 27 meta-analyses that the effect sizes of studies of varying methodological quality do not differ significantly.

Additional comments We made no direct comparison with shorter psychodynamic psychotherapies. We cannot therefore compare LPT with short-term therapies. However, in the literature, ESs for short psychodynamic psychotherapies varied from 0.16 (Svartberg & Stiles, 1991) to 1.10 (Crits-Christoph, 1992), 0.71 (Anderson & Lambert, 1995) and 1.46 (Leichsenring & Leibing, 2003). The ESs found in general psychotherapies were 0.68 (Smith and Glass, 1977) and 0.85 (Smith, Glass & Miller, 1980). These outcomes were mostly based on symptom-related assessments. It seems that long-term psychotherapy is as effective as shorter psychotherapies for symptom reduction. The added value probably lies in the resulting personality changes. Rudolf et al. (1994 and 2004) found that long- term psychoanalytic therapy performed better than moderate-length psychoanalytic therapy (M=60 sessions) in terms of structural changes of personality. The results of a study of Kopta et al. (1994) showed that improvements in personality structure take longer than symptom reduction. And Perry et al. (1999) estimated the length of treatment needed if patients were no longer to meet the criteria for personality disorder. They found that 50% of the patients would recover after 1.3 years (or 92 sessions) and 75% by 2.2 years (or 216 sessions). The results of our review indicate similar to somewhat lower success rates in personality changes. In our opinion, even a moderate effect on personality characteristics may be more important clinically than a large effect on symptoms related to quality of life, social functioning and vulnerability to relapse in the long run. Another issue is the durability of the results. The general literature contains indications that the effects of short-time therapies are more short-lived (e.g. Gloaguen et al., 1998; Hollon et al., 2005; de Maat et al., 2006). In contrast, our results suggest that the effects of LPT persist for years after treatment termination. As a final remark about the comparison of short-term and long-term psychotherapies, we would wish to recall that that the patients selected for short-term treatment in short-term trials differ in a number of respects from patients who need long-term treatment. This selection bias means that the results merit cautious interpretation. another comment should be made about the comparability of psychotherapy and psychoanalysis results. In our opinion, our findings do not allow for valid comparisons of the effectiveness of psychotherapy and psychoanalysis. Little is known about the comparability of the psychotherapy patient group and the psychoanalysis patient group in terms of all relevant pre-treatment characteristics. Since the groups are not randomized, it is highly likely that the personality traits of patients receiving psychotherapy and those receiving psychoanalysis differ in important aspects.

Limitations Our study presents several limitations. Firstly, our review is almost entirely based on cohort studies. Some will consider this to be a major flaw because, in

158 the hierarchy of empirical evidence, RCTs are the gold standard and present the strongest case. We do not dispute the strengths of RCTs and the fact that they provide the strongest possible scientific evidence. We are aware of the fact that the conclusive power of cohort studies is still a matter of debate. However, we are among those who consider well-designed cohort studies to be the highest possible level of evidence in this field. It has been extensively argued (Seligman, 1995; Wallerstein, 1999; Leichsenring, 2004; Westen et al., 2004; de Maat et al., 2007) that it is almost impossible to conduct randomized controlled trials in the field of long-term therapies (be they psychological or somatic). Treat- ments for borderline personality disorders present somewhat fewer difficulties in this respect. Leichsenring (2005) has cogently argued that, when studying long-term treatments, it is not possible to maintain a waiting list of untreated patients, to apply manualised treatments or to create equal treatment conditions. He points out that the personality characteristics of patients who opt for long-term psychoanalytic therapy differ from those of patients preferring other therapies. Randomization, if at all feasible, violates the therapy-patient match. so criticism of study designs in this field should focus on the quality of the cohort studies and not on the fact that they are not RCTs. Some feel cohort studies tend to overestimate treatment effect (Sacks, Chalmers and Schmidt, 1982; Kunz and Oxman, 1998). However, some studies contradict this point of view. Concato et al. (2000) compared the results of RCTs to those of cohort studies dealing with the same interventions in five different somatic disorders. Analyses of seventy-two meta-analyses of RCTs, twenty-four meta-analyses of cohort studies or case-control studies and six meta-analyses involving both designs brought to light that the results of the observational studies are remarkably close to those of RCTs. The authors concluded: ‘The popular belief that only randomized, controlled trials produce trustworthy results and that all observational studies are misleading does a disservice to patient care, clinical investigation, and the education of health care professionals.’ Shadish et al. (2000) did not find a significant correlation between the degree of clinical representativeness (e.g. RCTs versus naturalistic studies) and the size of effects reported in studies of psychotherapy. Benson & Hartz (2000) compared the results of observational studies with those of randomized, controlled trials reported between 1985 and 1998 dealing with two or more treatments or interventions for the same condition. The authors concluded: ‘We found little evidence that estimates of treatment effects in observational studies reported after 1984 are either consistently larger than or qualitatively different from those obtained in randomized, controlled trials.’ secondly, it will be clear that our review pools data from studies conducted in different countries and settings at different moments in time and in different patients, and applying non-manualised therapies (that despite being psychoanalytic, undoubtedly have different features). We have tried to enhance the comparability of the studies included by restricting our search to individual therapy in adult, ambulant patients, by distinguishing between moderate and severe pathology and by applying a quality criterion. Nevertheless, the studies

159 included are still obviously characterised by a clinical heterogeneity that is not accounted for and that is not assessed statistically. Our results must therefore be interpreted with caution. For the same reason, comparing pre-post findings with pre-follow-up findings is hazardous, since they mostly refer to different patients. thirdly, most of the studies included treated patients with different diagnoses and with high levels of comorbidity. They did not always break down their results according to diagnostic categories. Some studies provided limited demographic and clinical information about the patients. This represents a limit on conclusions about the effectiveness of LPT in specific disorders. However, it does enhance the representativeness of the patient population in general LPT clinical practice, which is precisely the strength of naturalistic effectiveness studies. Another consideration regarding the diagnostic aspects of this review is that our description of studies as dealing with ‘moderate’ or ‘severe pathology’ could be misleading. Even moderate pathology in LPT patients is not simple, single disorder pathology by any standard. These patients suffer from severe disorders and high comorbidity, and they have often tried briefer therapies, apparently to no avail (Doigde et al., 2002). fourthly, the research quality of the included studies is limited. Unfortu- nately, it is evident that the quality of a review depends heavily on the quality of the studies included. In an effort to safeguard quality, we devised a 24-item Research Quality Checklist inspired by the criteria put forward by Leichsenring (2004) and by the Cochrane Collaboration (Cochrane Reviewers Handbook, Cochrane Collaboration, 2004). We defined an admittedly arbitrary cut-off point and the qualities of the list have not yet been tested. The rigorousness of our quality check could, therefore, be amenable to amelioration. Thirty percent of the studies we found in the literature did not meet our quality criterion. Nevertheless, the quality of the studies that met our criterion varied widely. Flaws consisted of a lack of pre-treatment data (retrospective design), an over- general description of the patient population (e.g. only stating that the sample consisted of patients treated in a given period), no diagnoses mentioned, no detailed information about the therapies (e.g. only saying that the study dealt with psychoanalysis or psychotherapy, but not stating mean number of sessions or duration), over-general outcome criteria (e.g. only a five-step Likert scale for ‘therapy success’), no independent, let alone blind, assessors, no information about dropouts (e.g. only data for completers), and no confounder analyses. Only two studies tried to provide some information about control groups (Leichsenring, 2005; Holm-Hadulla et al., 1997). fifthly, the measurements of success rates were quite simple in many studies. The reliability of global assessments of success based on patient and therapist opinions are questionable. Some studies showed that global assessment scales provide better differentiation in terms of the severity of the illness and treatment results (Ottoson, 1960), but they do not provide the same information as detailed questionnaires, they are simple instruments, and often demand considerable clinical competence. Furthermore, success in our review was rated predominantly by therapists and patients, and only seldom

160 by independent observers. Some might argue that they tend to overestimate the effects of therapy, as they are too involved to judge objectively. Harty et al. (1976) performed an analysis of the findings of the Menninger Founda- tions’ Psychotherapy research project and found that both therapists (65%) and patients (54%) rated therapy success higher than independent judges (38%). The authors conclude: ‘(….) the disparity seemed partly due to differing frames of reference, but may also have selected the participants’ need to feel that their efforts had been worthwhile’. In our review, therapists and patients seem to be in agreement about therapy success rates. This concurs with the results of Harty et al. (1976), but we have too little data to make comparisons with opinions of independent observers. sixthly, the statistic analyses we could perform were rather simple. ESs are a reasonably robust evaluation of treatment effects but we were limited to a simple calculation of the average of the ESs reported in the individual studies. For lack of the original data, we could not calculate mean ESs based on the original means and standard deviations. Furthermore, some of the pooled results are based on only a few studies, and in some cases the results of only one study are mentioned. It is clear that findings are less convincing when founded on fewer studies. finally, the decisions about differentiation according to symptom reduction and personality changes may be debated. Success rates for symptom reduction were often based on Clinical Global Improvement scales, that do not necessarily differentiate between symptom improvement and changes in personality. We classified them as symptom reduction measures.

Conclusions LPT effectiveness has been extensively investigated, mainly by cohort studies of varying quality. Summarising the available empirical evidence suggests that LPT is an effective treatment modality, with moderate to large effects on both symptom reduction and personality changes. The effects are stronger in symptom reduction than in personality change, but moderate changes in personality may be more significant clinically in terms of quality of life and relapse prevention. Patients and therapists do not seem to differ in their opinions about therapy success. The results of LPT seem to be maintained in the years after treatment termination. Further research with cohort studies can and should be improved in several aspects.

References

Alexander, F. (1937). Five year report of the Chicago Institute for Psychoanalysis, 1932- 1937. Institute for Psychoanalysis. Chicago. Anderson, E.M., Lambert, M.J. (1995). Short-term dynamically orientated psycho therapy: A review and meta-analysis. Clinical Psychology Review, 15, 6: 503-514. Antonovsky, A. (1987). Unraveling the Mystery of Health. San Francisco, CA: Jossey- Bass. Bachrach, H., Galatzer-Levy, R., Skolnikoff, A., Waldron, S. (1991). On the efficacy of psychoanalysis. Journal of the American Psychoanalytic Association, 39: 871-916.

161 Bachrach, H., Weber, J., Solomon, M. (1985). Factors associated with the outcome of psychoanalysis (clinical and methodological considerations): Report of the Colom- bian Psychoanalytic Center Research Project (IV). International Review of Psycho- Aanalysis, 12: 379-389. Bateman, A., Fonagy, P. (1999). The effectiveness of partial hospitalisation in the treatment of borderline personality disorder: a randomized controlled trial. American Journal of Psychiatry, 156: 1563-1569. Benson, K, Hartz, A.J. (2000). A comparison of observational studies and randomized, controlled trials. New England Journal of Medicine, 342: 1878-86. Bieber, I. (1962). Homosexuality: a psychoanalytic study. New York, Basic Books Brockman, J., Schlüter, T., Brodbeck, D., Eckert, J. (2002) Die Effekte psychoanalytisch orientierter und verhaltenstherapeutischer Langzeittherapien. Ein vergleichende Studie aus der Praxis niedergelassener Psychotherapeuten. Psychotherapeut, 6: 347-355. Brockman, J., Schlüter, T., Eckert, J. (2003). Therpieziele, Zieländerungen und Zieler- reichung im Verlauf psychoanalytisch orientierter und verhaltenstherapeutischer Langzeittherapie. Psychotherapie und Psychosomatischen Medizin, 53: 163-170. Clarkin, J.F., Foelsch, P.A., Levy, K.N., Hull, J.W., Delaney, J.C., Kernberg, O.F. (2001). The development of a psychodynamic treatment for patients with borderline personality disorder: a preliminary study of behavioural change. Journal of Personality Disorders, 15(6): 487-495. Cochrane Collaboration, Cochrane Reviewers Handbook. Available at: http://cochrane. com. Accessed May, 2004. Cohen, J. (1988). Satistical power analysis for behavioural sciences. Hillsdale, NJ: Lawrence Erlbaum. Concato, J., Shah, N., Horwitz, R.I. (2000) Randomized controlled trials, observational studies and the hierarchy of research designs. New England Journal of Medicine, 342 (25), 1887-1892. Coriat, I.H. (1917) Some statistical results of the psychoanalytic treatment of the psy- choneuroses. Psychoanaltic Review, 4: 209-216. Cremerius, J. (1962). Die Beurteilung des Behandlungserfolges in de Psychotherapie. Springer. Berling Heidelberg New York. Crits-Chritoph, P. (1992) The efficacy of brief dynamic psychotherapy: a meta-analysis. American Journal of Psychiatry, 149: 2: 151-158. Derogatis, L.R., Lazarus, L. (1974). SCL-90-R, Brief Symptom Inventory, and matching clinical rating scales. In: The Use of Psychological Testing for Treatment Planning and Outcome Assessment. Ed. M.E. Maruish. Hillsdale, NJ: Lawrence Erlbaum, 217- 48. Doidge, N. (1997). Empirical evidence for the efficacy of psychoanalytic psychotherapies and psychoanalysis: An overview. Psychoanalytic Inquiry Supplemental, 102-150. Doidge, N. (2001). Is Psychoanalysis Effective? TEN, 3(2): 41-47. Doidge, N., Simon, B., Brauer, L., Grant, D.C., First, M., Brunshaw, J., Lancee, W.J., Stevens, A., Oldham, J.M., Mosher, P. (2002a). Psychoanalytic patients in the U.S., Canada, and Australia:I. DSM-III-R disorders, indications, previous treatment, medications and length of treatment. Journal of the American Psychoanalytic Associa- tion, 50(2): 575-614. Doidge, N., Simon, B., Lancee, W.J., First, M., Brunshaw, J., Brauer, L., Grant, D.C., Stevens, A., Oldham, J.M., Mosher, P. (2002b). Psychoanalytic patients in the U.S., Canada, and Australia: II. A DSM-III-R validation study. Journal of the American Psychoanalytic Association, 50(2): 615-627. Dührssen, A. (1986). Dynamische Psychotherapie, Psychoanalyse und analytische Grup- penpsychotherapie im Vergleich. Zeitschrift für Psychosomatischen Medizin, 32: 161-180. Erle, J.B., Goldberg, D.A. (2003). The course of 253 analyses from selection to outcome. Journal of the American Psychoanalytic Association, 51(1): 257-293.

162 Feldman, H. (1968). Results of psychoanalysis in clinical case assignements. Journal of the American Psychoanalytic Association, 17: 2740300. Fenichel, O. (1930). Statistischer Bericht über die therapeutische Tätigkeit 1920-1930. In: Radó S, Fenichel O, Müller-Braunschweig C (Hrsg). Zehn Jahre Berliner Psy- choanalytisches Institut Poliklinik und Lehranschalt. Internationaler Psychoana- lytischer Verlag, Wien, S 13-19. Fonagy, P. (1999). An open door review of outcome studies in psychoanalysis. http:// www.ipa.org.uk/R-outcome.htm. Fonagy, P., Moran, G.S. (1990). Studies on the efficacy of child psychoanalysis. Journal of Consulting and Clinical Psychology, 58 (6): 684-95. Freedman, N., Hoffenberg, J.D., Vorus, N., Frosch, A. (1999). The effectiveness of psychoanalytic psychotherapy: the role of treatment duration, frequency of sessions, and the therapeutic relationship. Journal of the American Psychoanalytic Association, 47(3): 741-772. Freud, S. (1919). Wege der psychoanalytischen Therapie, Ergänzungsband, 243. Friedman, R.C., Bucci, W., Christian, C., Drucker, P., Garrison, W.B. 3rd. (1998). Private psychotherapy patients of psychiatrist psychoanalysts. American Journal of Psy- chiatry, 155(12): 1772-1774. Friedman, R.C., Garrison, W.B. 3rd, Bucci, W., Gorman, B.S. (2005). Private psychotherapy patients of Senior Psychoanalysts: an effectiveness study. Journal of the American Academy of Psychoanalysis and Dynamic Psychiatry, 33(4): 583- 610. Giessen-Bloo, J., Dyck, van, R., Spinhoven, P., Tilburg, van, W., Dirksen, C., Asselt, van, T., Kremers, I., Nadort, M., Arntz, A. (2006). Outpatient psychotherapy for borderline personality disorder. Archives of General Psychiatry, 63: 649-658. Gloaguen, V., Cottraux, J., Cucherat, M., Blackburn, I. (1998) A meta-analysis of the effects of cognitive therapy in depressed patients. Journal of Affective Disorders, 49: 59-72. Graham, S.R. (1960). The effects of psychoanalytically oriented psychotherapy on levels of frequency and satisfaction in sexual activity. Journal of Clinical Psychology,16: 94-5. Grande, T., Dilg, R., Jakobsen, T., Keller, W., Krawietz, B., Langer, M., Oberbracht, C., Stehle, S., Stennes, M., Rudolf, G. (2006). Differential effects of two forms of psychoanalytic therapy: results of the Heidelberg-Berlin study. Psychotherapy Research, 16(4): 470-485. Guy W. (1976). ECDEU Assessment Manual for Psychopharmacology. Revised DHEW Pub. (ADM). Rockville, MD: National Institute for Mental Health. Hamburg, D.A., Bibring, G.L., Fisher, C., Stanton, A.H., Wallerstein, R.S., Weinstock, H.I., Haggard, E. (1967). Report of ad hoc committee on central fact gathering data of the American Psychoanalytic Association. Journal of the American Psychoanalytic Association, 15: 841-61. Hartmann, S., Zepf, S. (2004). Einflüsse auf die Symptombesserung in der Psychothera- pie bei Patienten mit unterschiedlichen Beschwerdebildern. Psychotherapie und Psychosomatischen Medizin, 54: 445-456. Harty, M., & Horwitz, L. (1976). Therapeutic outcome as rated by patients, therapists, and judges. Archives of General Psychiatry, 33: 957-961. Heinicke, C.M. (1965). Frequency of psychotherapeutic sessions as a factor affecting the child’s developmental status. Psychoanalytic study of the child, 20: 42-98. Heinzel, R., Breyer, F., Klein, T. (1998). Ambulante analytische Einzel- und Gruppenpsy- chotherapie in einder bundesweiten katamnestischen Evaluationsstudie. Grup- penpsychotherapie und Gruppendynamik, 34: 135-152. Heuft, G., Seibuchler-Engec, H., Taschke, M., Senf, W. (1996). Langzeitoutcome ambu- lanter psychoanalytischer Psychotherapien und Psychoanalysen. Ein textinhaltsana- lytische Untersuchung von 53 Katamneseinterviews. Forum der Psychoanalyse, 12: 342-355.

163 Hollon, S.D., DeRubeis, R-J., Shelton, R.C., Amsterdam, J.D., Salomon, R.M., O’Reardon, J.P., Lovett, M.L., Young, P.R., Haman, K.L., Freeman, B.B., Gallop, R. (2005) Pre- vention of Relapse following cognitive therapy vs medications in moderate to severe depression. Archives of General Psychiatry, 62: 417-422. Holm-Hadulla, R., Kiefer, L., Sessar, W. (1997). Zur Effektivität tiefenpsychologisch fundierter Kurz- und Psychotherapien. Psychotherapie, Psychosomatischen Medizin und Psychologie, 47: 271-278. Horowitz, L.M., Strauss, B., & Kordy, H. (2000). Inventory of Interpersonal Problems (2nd ed.). Göttingen: Beltz. Jones, E. (1936). Report of the clinic work: 1926-1936. Clinic of Psychoanalysis. Lon- don. Keller, W., Westhoff, G., Dilg, R., Rohner, R., Studt, H.H., and the study group on empirical psychotherapy research in analytical psychology. (1997). Efficacy and cost effectiveness aspects of outpatient Jungian psychoanalysis and psychotherapy, a catamnestic study. Berlin: Department of Psychosomatics and Psychotherapy, Uni- versity Medical Center Benjamin Franklin, Free University of Berlin. At: http://www. uni-saarland.de Kernberg, O., Burstein, E.D., Coyne, L., Appelbaum, A., Horwitz, L., Voth, H. (1972). Psychotherapy and psychoanalysis-The final report of the Menninger Foundations’s Psychotherapy Rsearch Project. Bullentin of the Menninger Clinic, 36, 1-275. Kessel, L., Hyman, H.T. (1936). The value of psychoanalysis as a therapeutic procedure. Journal of the American Medical Association, 107: 326. Klein, H. (1960). A study of changes occurring in patients during and after psychoanalytic treatment. In: Hoch PH, Zubin J (eds). Current approaches to psychoanalysis. New-York. Grune & Stratton, pp. 151-175. Knapp, P.H., Levin, S., McCarter, R., Werner, H., Zetzel, E. (1960) Suitability for psychoanalysis: a review of 100 supervised analytic cases. Psychoanalytic Quarterly, 29: 459-477. Knight, R.P. (1941). Evaluation of the results of psychoanalytic therapy. American Jour- nal of Psychiatry, 98: 434-446. Kopta, S.M., Howard, K.I., Lowry, J.L., Beutler, L.E. (1994). Patterns of symptomatic recovery in psychotherapy. Journal of Consulting and Clinical Psychology, 62: 1009-1016. Kordy, H., Rad, von, M., Senf, W. (1988). Time and its relevance for successful psychotherapy. Psychotherapy and Psychosomatics, 49: 212-222. Kunz, R., Oxman, A.D. (1998). The unpredictability paradox: review of empirical comparisons of randomized and non-randomized clinical trials. British Medical Journal, 317: 1185-90. Leichsenring, F., Leibing, E., (2003). The effectiveness of psychodynamic therapy and cognitive behavior therapy in the treatment of personality disorders: A meta-analysis. American Journal of Psychiatry, 160: 1223-1232. Leichsenring, F. (2004). Randomized controlled versus naturalistic studies: a new research agenda. Bulletin of the Menninger Clinic, 68,2: 137-151. Leichsenring, F. (2005). Are psychodynamic and psychoanalytic therapies effective? A review of empirical data. International Journal of Psychoanalysis, 86: 841-68. Leichsenring, F., Biskup, J., Kreische, R., Staats, H. (2005). The Göttingen study of psychoanalytic therapy: first results. International Journal of Psychoanalysis, 86: 433-55. Leuzinger-Bohleber, M., Stuhr, U., Rüger, B., Beutel, M. (2003). How to study the ‘quality of psychoanalytic treatments’ and their long-term effects on patients well-being: a representative, multi-perspective follow-up study. International Journal of Psy- choanalysis, 84: 263-290. Leuzinger-Bohleber, M., Target, M. (2001). Outcomes of psychoanalytic treatment. Per- spectives for therapists and researchers. Ed. by M. Leuzinger-Bohleber, M. Target. Whurr Publishers, London and Philadelphia.

164 Leuzinger-Bohleber, M., Stuhr, U., Rüger, B., Beutel, M. (2003). How to study the ‘quality of psychoanalytic treatments’ and their long-term effects on patients well-being: a representative, multi-perspective follow-up study. International Journal of Psy- choanalysis, 84: 263-290. Lipsey, M.W., Wilson, D.B. (1993). The efficacy of psychological, educational, and behavioural treatment: Confirmation from meta-analyses. American Psychologist, 48, 12: 1181-1209. Luborsky, L., Stuart, J., Friedman, S., Diguer, L., Seligman, D.A., Bucci, W., Pulver, S., Krause, E.D., Ermold, J., Davidson, W.T., Woody, G., Mergenthaler, E. (2001). The Penn Psychoanalytic treatment collection: a set of complete and recorded psychoana- lyses as a research resource. Journal of the American Psychoanalytic Association, 49(1): 217-234. Maat, S., Dekker, J., Schoevers, R., Jonghe, F., de. (2006). Relative efficacy of psychotherapy and pharmacotherapy in the treatment of depression: a meta-analysis. Psychotherapy Research, 16(5): 562-572. Maat, S., Dekker, J., Schoevers, R., Jonghe, F., de. (2007). The effectiveness of long-term psychotherapy: methodological issues. Psychotherapy Research, 17(1): 59-65. Monsen, J.T., Odland, T., Faugli, A., Daae, E., Eilertsen, D.E. (1995). Personality disorders:changes and stability after intensive psychotherapy focusing on affect consciousness. Psychotherapy Research, 5(1): 33-48. Monsen, J.T., Odland, T., Faugli, A., Daae, E., Eilertsen, D.E. (1995). Personality disorders:changes and stability after intensive psychotherapy: a prospective follow-up study of an outpatient psychotherapy project, 5 years after end of treatment. Scan- dinavian Journal of Psychology, 36: 256-268. Orgel, S. (1958). Effects of psychoanalysis on the course of peptic ulcer. Psychosomatic Medicien, 20: 117-125. Ottoson, J-O. (1960). Experimental studies of the mode of action electroconvulsive treatment. Acta Psychiatrica Scandinavia, 35, suppl. 145: 69-97. Perry, J.C., Banon, E., Floriana, I. (1999). Effectiveness of psychotherapy for personality disorders. American Journal of Psychiatry, 156: 1312-21. Quiroga, S. (1998). Estudio comparativo sobre eficacia terapéutica en trastornos de la alimentación-I. Factores diferenciales y determinantes comunes [Com- parative Study about Therapeutic Efficacy in Eating Disorders -I . Differen- tial Factors and Common Determinants.]. Paper presented at the IV Encuentro Capítulo Sudamericano de la Society for Psychotherapy Research, Montevideo, Sep- tember. Rad, von, M., Senf, W., Bräutigam, W. (1998). Psychotherapie und Psychoanalyse in der Krankenversorgung: Ergebnisse des Heidelberger Katamnese-Projektes. Psycho- therapie, Psychosomatischen Medizin, Psychologie, 48: 88-100. Rosenthal, R. (1991). Meta-analytic procedures for social research: applied social research methods. Newbury Park, Calif, Sage Publications. Rudolf, G., Manz, R., Öri, C. (1994). Ergebnisse psychoanalytischer Therapien. Zeitschrift zur Psychosomatischen Medizin, 40: 25-40. Rudolf, G., Dilg, R., Grande, T., Jakobsen, T. Keller, W. Krawietz, B. et al. (2004). Effektiv- ität und Effizienz psychoanalytischer Langzeittherapie: Die Praxisstudie Analytische Langzeitpsychotherapie (Effectiveness and efficiency of long-term psychoanalytic psychotherapy: The practice study of long-term psychoanalytic psychotherapy). In: Gerlach, A, Springer, A., Schlösser, A. editors. Psychoanalyse des Glaubens (Psy- choanalysis of religious belief). Giessen: Psychosozial Verlag. Sacks, H., Chalmers, T.C., Smith, H. Jr. (1982). Randomized versus historical controls for clinical trials. American Journal of Medicine; 72, 233-40. Sandell, R., Blomberg, J., Lazar, A., Carlsson, J., Broberg, J., Schubert, J. (2000). Varie- ties of long-term outcome in psychoanalysis and long-term psychotherapy. A review of findings in the Stockholm outcome of psychoanalysis and psychotherapy project (STOPP). International Journal of Psychoanalysis, 81: 921-942.

165 Sashin, J.I., Eldred, S., Amerongen, van, S.T. (1975). A search for predictive factors in institute supervised cases: a retrospective study of 183 cases from 1959-1966 at the Boston psychoanalytical society and institute. International Journal of Psychoa- nalysis, 56: 343-359. Schjelderup, H. (1955). Lasting effects of psychoanalytic treatment. Psychiatry, 18: 109- 33. Seligman, M.E.P. (1995). The effectiveness of psychotherapy. The Consumer Reports Study. American Psychologist, 50(12): 965-974. Shadish, W.R., Matt, G., Navarro, A., Phillips, G. (2000). The effects of psychological therapies under clinically representative conditions: a meta-analysis. Journal of Consulting and Clinical Psychology, 126: 512-529. Smith, M.L., Glass, G.V., & Miller, T.I. (1980). The benefits of psychotherapy. Baltimore, MD: John Hopkins University Press. Stanton A.H., Gunderson J.G., Knapp P.H., Frank A.F., Vannicelli M.L., Schnitzer R. & Rosenthal R. (1984). Effects of psychotherapy in schizophrenia: I. Design and implementation of a controlled study. Schizophrenia Bulletin, 10: 520-564. Stehle, S. (2004). Psychotherapeutische Berfustätigkeit. Ergebnisse der DGPT-Thera- peutenerhebung. In: Psychoanalyse des Glaubens. Ed: Gerlach, A., Schlösser, A., Springer, A. Bibiliothek der Psychoanalyse, Psychosozial-verlag. Stevenson, J., Meares, R. (1992). An outcome study of psychotherapy for patients with borderline personality disorder. American Journal of Psychiatry, 149(3): 358-362. Svartberg, M. & Stiles, T.C. (1991). Comparative effects of short-term psychodynamic psychotherapy: a meta-analysis. Journal of Consulting and Clinical Psychology, 59: 704-714. Wallerstein, R. (1999). Comment on Gunderson and Gabbert. Journal of the American Psychoanalytic Association, 47: 728-34. Wallerstein, R. (1986). Forty-two lives in treatment: a study of psychoanalysis and psychotherapy. New-York: Guilford Press. Weber, J.J., Bachrach, H.M., Solomon, M. (1985). Factors associated with outcome of psychoanalysis: a report of the Colombia Psychoanalytic Center research project (II). International Review of Psychoanalysis, 12: 127-141. Weber, J.J., Bachrach, H.M., Solomon, M. (1985). Factors associated with outcome of psychoanalysis: a report of the Colombia Psychoanalytic Center research project (III). International Review of Psychoanalysis, 12: 251-262. Weissman, M., Bothwell, S. (1976). Assessment of social adjustment by patient self report. Archives of General Psychiatry, 33: 1111-15. Westen, D., Novotny, C.M., Thomson-Brenner, H. (2004). The empirical status of empirically supported psychotherapies: assumptions, findings, and reporting in controlled clinical trials. Psychology Bulletin, 130: 631-63. Wilczek, A., Berber, J.P., Gustavson, J.P., Asberg, M., Weinryb, R.M. (2004). Change after long-term psychoanalytic psychotherapy. Journal of the American Psychoana- lytic Association, 52(4): 1163-1184.

166 Appendix 1

Research Quality Score (RQS) Criteria Single or Multiple Cohort study with Randomized Cohort study matched control Controlled group* Trial Points Points Points Study design 1. Randomization n.a. n.a. 35 2. Prospective design 10 10 n.a. 3. (Matched) control group n.a. 5 -10 n.a. Patient group 4. Clear inclusion/exclusion criteria 0 -1 - 2 0 -1 - 2 0 -1 - 2 5. Characteristics described 0 -1 - 2 0 -1 - 2 0 -1 - 2 6. Baseline scores comparable n.a. 0 -2 - 4 0 -2 - 4 7. Adequate sample size 0 -1 - 2 0 -1 - 2 0 -1 - 2 Intervention 8. Clear description of intervention 0 -2 - 4 0 -2 - 4 0 -2 - 4 9. Therapist experience 0 -1 - 2 0 -1 - 2 0 -1 - 2 10. Adherence therapist checked 0 -1 - 2 0 -1 - 2 0 -1 - 2 11. Equal treatment groups n.a. 0 - 2 0 - 2 Outcome data 12. Outcome criteria clear and relevant 0 -2 - 4 0 -2 - 4 0 -2 - 4 13. Quality assessment instruments 0 -2 - 4 0 -2 - 4 0 -2 - 4 14. Blind or independent assessment 0 -2 - 4 0 -2 - 4 0 -2 - 4 15. Multiple assessors 0 -2 - 4 0 -2 - 4 0 -2 - 4 Statistics 16. Adequate statistical methods 0 -1 - 2 0 -1 - 2 0 -1 - 2 17. ITT or PP analyses 0 -2 - 4 0 -2 - 4 0 -2 - 4 18. Confounders analyses 0 -1 - 2 0 -1 - 2 n.a. Drop-out 19. Drop-out defined and acceptable 0 -2 - 4 0 -2 - 4 0 -2 - 4 20. Drop-out comparable n.a. 0 -1 - 2 0 -1 - 2 Maximum RQS 52 70 83 Cut-off score review 26 35 41 Follow-up 21. Adequate length 0 -2 - 4 0 -2 - 4 0 -2 - 4 22. Multiple assessors 0 -1 - 2 0 -1 - 2 0 -1 - 2 23. Loss to follow-up defined and acceptable 0 -2 - 4 n.a. n.a. 24. Loss to follow-up defined, acceptable, comparable n.a. 0 -2 - 4 0 -2 - 4 10 10 10 Maximum RQS 62 80 93 Cut-off score review 31 40 47 * A matched control group is either an untreated matched group (with similar features as the intervention group) or a group treated with evidence-based treatment.

167 Appendix 2

List of assessment instruments AAI Adult Attachment Interview BDI Beck Depression Inventory BPDSI-IV Borderline Personality Disorder Structured Interview BSS Beschweden-Schwere-Score (BSS Impairment Severity Score) CGI-I Clinical Global Impression-Improvement CPRS-S-A Comprehensive Psychopathological Self-Rating Scale for Affective Syndromes DSM Diagnostic Statistical Manual DSQ Defence Style Questionnaire EPQ Eysenck Personality Questionnaire EuroQol European Quality of Life thermometer FKBS Fragebogen zu Konfliktbewältigungsstrategien FLZ Fragen zur Lebenszufriedenheit FPI Freiburger Persönlichkeitsinventar GAF General Assessment of Functioning GAS Goal Attainment Scale GB Giessener Beschwerde Bogen GSI of SCL-90 General Severity Index GT Giessen Test HARS-G Hamilton Anxiety Rating Scale HDRS Hamilton Depression rating Scale HSRS Health Sickness Rating Scale IIP Inventory of Interpersonal Problems INTREX Introjektfragebogen IRES Life satisfaction scale KAPP Karolinska Psychodynamic Profile KSP Karolinska Scales of Personality LSS Life Situation Survey MMPI Minesota Multiphasic Personality Inventory NART National Adult Reading Test PC Perceived Competence PSKB-Se Psychischer und Sozial-Kommunikativer Befund-Selbstbeurteilung QRS Quality of Object Relations Rating Scale SAS Social Adjustment Scale SASB Structural Aspects of Social Behaviour SCID Structural Clinical Interview SCL-90 Symptom Check List SOC Sense of Coherence Scale SOFAS Version of Global Assessment of Functioning Scale SSI Satisfaction, Success and Inprovement (combination) SSI Scale for Suicidal Ideation STAI Spielberger State and Trait Anxiety Inventory VEV Veränderungsfragebogen des Erlebens und Verhaltens WAIS-R Wechsler Adult Intelligence Scale-Revised WHOQOL World Health Organisation Quality of Life assessment

168 Chapter 10 Costs and benefits of long-term psychoanalytic therapy: changes in health care use and work impairment

Saskia de Maat, Frans Philipszoon, Robert Schoevers, Jack Dekker, Frans de Jonghe Harvard Review of Psychiatry 2007 (in press)

Abstract

Objective: Systematic review regarding the effectiveness of LPT on health care use and work impairment in adult outpatients. Method: A systematic search for studies published between 1970 and 2005. Calculation of the weighed mean changes between pre-treatment and treatment termination and between pre-treatment and follow-up. Translating the findings into financial terms. Balancing the costs of treatment against the financial gains. Results: Seven studies (N= 861) met all the inclusion and none of the exclusion criteria. The mean cost of LPT per patient was € 20.900. During the year preceding treatment termination and that preceding mean follow-up (2.9 years), the average reduction was 85% and 59% respectively for hospital days, 54% and 56% for the number of medical consultations, 70% and 19% for medication users and 61% and 67% for days of sick leave. Health care use and sick leave costs fell by € 5.584 between the year preceding the start of psychotherapy and the year preceding treatment termination (a 66% reduction). At mean follow- up (2.9 years) these costs reductions were still apparent as the reduction was € 5.371 (64%) in the year preceding follow-up. The break-even point for benefits and treatment costs is approximately three years after treatment termination. The reduction in work impairment appears to be the main factor (65%-75%) in these positive results. Conclusions: Our data suggest that LPT substantially reduces health care use and sick leave. The benefits seem to endure for years after termination and outweigh the costs of treatment approximately three years after treatment termination.

169 10.1 Introduction

In recent decades, the costs and benefits of medical care have increasingly been the subject of attention - and concern - of patients, therapists, health insur- ances and politicians alike. Psychotherapy is no exception. Relevant literature is scarce and focuses mainly on time limited (cognitive-behavioural) psychotherapies. It is unclear whether Long-term Psychoanalytic Therapy (LPaT) is cost effective. three reviews we discuss here deal predominantly with short-term psychotherapies. Mumford et al1 report that in 85% of the 58 studies included in their review the use of medical care was reduced after psychotherapy. Gabbard et al2 report the results of a review including 35 studies. They find that 88.9% of the studies suggest that psychotherapy has a beneficial economic impact. The authors conclude: ‘Psychotherapy appears to have a beneficial impact on a variety of costs (…). Much of that impact accrues from reductions in inpatient treatment and decreases in work impairment.’ Baltensperger & Grawe3 report the results of a review including 124 studies. Their study shows that psychotherapy results in cost reductions relating to health services and work impairment. One of the studies they report on4 includes a financial evaluation of the last topic and reports a net cost saving in the two years after treatment termination of DM 21,370 per patient. Importantly, these reviews illustrate the cost-effectiveness of psychotherapy but they also have some limitations. Firstly, they cover a wide variety of patient populations and disorders and this heterogeneity precludes any specific conclusion. Secondly, only a few of the primary studies deal with issues relating to work impairment. two singular studies deserve attention because of their longer treatment durations. Hall et al5 calculate the net costs of one year psychotherapy for Borderline Personality Disorder (BPD) patients (as described in Stevenson & Meares6) defining net costs as the difference between the costs of health care services during the twelve months before the start of treatment and the costs of psychotherapy plus those of health care use for a period of twelve months after treatment termination. The authors report a net cost saving of AUD 18,000 per patient, mostly due to reduced hospital admissions. Bateman & Fonagy7 assess the health care costs for BPD patients assigned to either eighteen months of partial hospital treatment or treatment as usual. On the basis of their data, we calculated the net costs of both groups in a way similar to that of the Hall study. The net cost saving was USD 14,400 for the hospital group and 6,000 for the TAU group. Both studies were restricted to health care costs for BPD patients. Although some individual studies cover certain aspects of cost-effectiveness of some psychoanalytic therapies for some patients, there is no systematic overview of the evidence for the cost-effectiveness of Long-term Psychoanalytic Therapy (LPaT). It is therefore not known whether LPaT is cost-effective. The aim of this paper is twofold. Firstly, we perform a systematic review of relevant studies published from 1970 until 2005 relating to the effects of LPaT on health care use and work impairment. Secondly, we present a financial evaluation of the costs of LPaT compared to its benefits in the same two areas.

170 10.2 Method

10.2.1 Literature search

A literature search was conducted in Pubmed, Embase, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, PsychLit en ACP Journal Club. The following subject headings were applied: Long-term psychotherapy, Psychoanalysis, Psychoanalytic psychotherapy, Psychodynamic therapy. The time limit was 1970. There were no limits set on language. Cross-references in the retrieved publications were searched. The Open Door Review8, the review by Brandl et al9 and the review by Leichsenring10 were used as additional sources of information. There was no systematic effort to find unpublished data.

The following in and exclusion criteria were applied: 6. the studies were required to be ‘outcome-intervention studies’, that provided (in addition to other outcome measures) data about effects on health care use and/or work impairment issues. 7. the studies had to be either RCTs or cohort studies. Case studies or case series were excluded. 8. the studies had to relate to individual, outpatient psychoanalytic therapies with adult patients (18 tot 65 year). Studies relating to children or the elderly, and studies conducted in clinical settings or day-care settings were excluded. 9. the studies had to deal with ‘standard’ patients in ‘standard’ psychoanalytic private practices. What this means in DSM-III-R terms is indicated by the results of a review performed by Doidge et al22 of 510 psychoanalysts in the USA, Canada en Australia, and dealing with 1718 patients: 65% had a mood disorder, 44% an anxiety disorder, and 73% a personality disorder; 3% had no DSM diagnosis at all; the modus of DSM-III-R disorders was 3. The authors mention in addition that 80% had received previous treatment. These results are corroborated by the DSM-IV findings of Friedman et al23 for 51 psychoanalysts and 551 patients: 46% had an axis I disorder, 12% an axis II disorder, and 42% had both an axis I and an axis II disorder. We included these more ‘commonly’ axis I and II disorders found in private therapists practices, but excluded the disorders that are mostly treated in health care institutions. This means we did not exclude BPD and other severe personality disorders, but we did exclude ‘heroic indications’11,12, psychotic disorders such as schizophrenia13, somatic illnesses such as juve- nile-onset insuline-dependend diabetes14, eating disorders15 and studies focusing exclusively on rare disorders such as Münchausen-by-proxy’. 10. the average length of treatment had to be at least one year, and consist of at least 50 sessions. This included psychoanalytic psychotherapy as well as psychoanalysis. 11. the methodological quality of each study was required to be sufficient, as independently assessed with the aid of a checklist (see appendix 1) by two of the authors (SdM & FdJ).

171 10.2.2 Data-extraction

All data regarding the number of sessions, health care use and work impairment were extracted from each paper. It emerged that data was only available about numbers of hospital days and medical visits, percentages of patients taking medication and numbers of sick leave days. The measurement points were: before or at start of treatment, at treatment termination and at follow-up.

10.2.3 Reduction calculations

The weighted mean reductions of all parameters across the studies were calculated by comparing the situation before the start of treatment to that at treatment termination and at mean follow up.

10.2.4 Financial evaluation

In order to assess the costs and benefits and present a financial evaluation, we applied present-day financial costing standards to the data.16,17,18,19 As these sources represent the situation in The Netherlands in 2003 and 2004, we indexed them for 2005. See Appendix 2 for all cost calculations. The average indexed cost of a psychotherapy session was taken to be € 76. We calculated the treatment costs of an ‘average LPT’ by multiplying the mean number of sessions over all studies by € 76 (see section 10.3.1 for the results). The average indexed cost of a sick leave day was considered to cost € 288,30, a day in a psychiatric hospital € 257,55, an average medical consult € 38,91, and medication use € 272,55 per year per patient (see Appendix 2). Pre-treatment costs were computed by multiplying the mean number of sick leave days, hospital days, medical consultations and medication use by the costs mentioned before. Post- treatment costs and follow-up costs were calculated similarly; mean number of sick leave days, hospital days, consultations and medication at the specific time multiplied by the same amounts mentioned. The average benefit of LPaT was established as the difference between the pre-treatment costs and the post- treatment costs, respectively the follow-up costs.

10.3 Results

10.3.1 Literature search

172 issues. The remaining 73 articles were retrieved and a further exclusion of 45 studies followed in the second selection round. Reasons for exclusion in this round were: process-outcome studies, studies focusing on children, reviews, studies on short-term therapies, studies focusing on therapists variables, studies on inpatients. Finally, 28 studies outcome studies were retrieved of LPT in patients with the indications that we aimed for (see Method section). Of these studies seven offered information on data about effects on health care use and/or work impairment issues. The seven studies we included are presented in table 10.1. the number of patients included in the studies ranged from 25 to 500 (total n=861), treated with either psychoanalysis or psychoanalytic psychotherapy. The number of sessions ranged from 80 to 640, averaging 139 in psychotherapy and 413 in psychoanalysis. The average for all the studies was 275. This means that a LPT costs € 20.900 euro per person (275 x 76 euro) on average. All studies except one (the still ongoing study of Rudolf et al20) included follow-up data. Length of follow-up varied from 1 to 6 years (mean 2.9 years). The patients in the studies of Monsen et al21 and Stevenson & Meares6 (total n = 55) suffered from severe personality disorders and were treated in health care facilities. The remaining studies dealt with ‘standard’ patients (total n =806) in ‘standard’ psychoanalytic private practices.

10.3.2 Sick leave

Six studies, relating to a total of 776 patients, report data on changes in patterns of sick leave expressed in days per annum (Table 10.2). the average pre-treatment/post-treatment reduction is 61%, the mean pre- treatment/follow-up reduction is 67%. The follow-up period ranges from 1 year to 6 years, mean duration is 3 years. The number of sick leave days is exceptionally high in the study of Stevenson and Meares6. This is not surprising since this study looks at borderline patients while most other studies look at ‘normal private psychoanalytic practice’. We have calculated the results of this table without this study that can be considered an ‘outlier’. Pre-treatment mean number of sick leave days is then 16,2, post treatment it is 8,1 (reduction of 50%) and post-treatment it is 6 (reduction compared to pre-treatment 63%). Rudolf et al20 break down the results according to the treatment modality: the rate of sick leave before the start of treatment is 24.6 days per annum in psychoanalysis and 28.3 days per annum in psychotherapy. After two years of treatment it is 6.8 d/y in psychoanalysis and 9.3 d/y in psychotherapy, which is a reduction of 72% for psychoanalysis and of 67% for psychotherapy. The author also includes for comparative purposes sick leave data for the general German population in the years relevant for his study: 17.3 days per annum in 1996 and 11.4 days per annum in 2002, which is a reduction of 34%. Keller et al26 compare the results of their patient sample with the general population of a German health care insurance company (Barmer Ersatz Kasse). Extrapolated to 100 patients (based on their sample of 47 patients) they find a duration of work disability days of 41,6 (1 year before treatment) and 13,5 (1 year after treatment), showing a

173 - a at in at

- of - is -

of 25

that

and etc. insur

after than II

they

had

clinic

drop-

35 and treat

subject

but 31% 7 Data that ended 78% educa meas

was

68%

group patients

Institut fulfil

sampled, different

untrained to less

axis care

started

Berlin and German

concluded in leave leave

30 general

had

found

Psychology

that

differ

patients question Results

replied, that participated

the (from not and 979 patients by

etc.)

not

treatment.

follow-up.

Outpatient

addition

patients All

90-94. group recruited. sick sick continuously 67

specified

improved in say remarks

treatment not

with of of

unsuccessfully

AOK

of at do cases

Erkrankungen,

health

(including In

were

of German

Associations orientations

patients

outcome Relative

between regarding a 77 referrals insured

patients

the the

do treatment.

not

all

study participated 21

the reduction

DGIP), completers, into

study.

analysts of received

remaining All

been patients study. Jungian,

Analytical participated.

100 patients

company.

and

participated questionnaires.

individual 87-88 first

psychoanalysis, that patients sample sample occupation

37 of

psychoanalyst).

between

General

were Authors

PD, Only of and

screened.

patients were regarding

regarding

with

Of short-term

article.

LPT

111

Majority therapies

30 groups compared 25

a company.

significantly 62 of (55) had standard and

patients

broken

differences

Psychogene

this members data

months.

psychoanalytic (Freudian, therapies between therapists. 979 returned No groups. of therapists population not insurance therapy, continued, criteria 75% termination. were period 6 high were outs). therapists) therapy according other From between Psychoanalytic (DGPT only previously mention hospitalisation. least documented of results Berlin. initially results clinic. (Jungian (n=174) 25% Random employed 75% fur Random ended Australian ment Society treatment. Norwegian 72%. Catamnestic ance ures in naires. tion Pt Pt Pt/IO Pt/IO Pt Pt Pt/IO Pt/IO -- -- Pt -- HCI Pt/IO HCI HCI

Pt Pt/IO Pt/IO HCI

Assessment HD: MC: MU: HD: MC: MU: SL: HD: MC: MU: SL: HD: MC: MU: SL: HD: MC: MU: SL: SL: design

Study Retrospective Prospective Prospective Retrospective Retrospective years.

years year years years

Follow-up 5,2 6 1 2 5 follow-up

and

months

months Time

months months

29-36 41 12 25,4 duration

mean) Treatment Mean

¹ sessions

number 80² 102³ 193 (psychoanalysis) 78 (psychotherapy) 166 (total 261 289 - 35 home. 44,5 28,6 29.4

unable - at begin

than assist days

level

more age: age: age age

Patient female. college married. previously female. married. female. married. government unem employees. more

further further owned low

work

demographics psychoanalysis group. No demographics follow-up: treatment: to Mean mentioned. Mean years. 63% ance. 20% education. 7% years. 24% ployed. 16% hospitalised. Mean years. Mean years. 69% given. 30% 76% 74% 73% 56% 4% financial ‘workers’ 62% No demographics Psychotherapy group hospital and - - - - per - in of the

disor- disor dura per disor-

the

than

II I

20% sickest, severe major Patients

disorder, mention more than

in practice practice

disorder disorder

years

subjects

assessed

problems:

factors absence dysfunction’

the

Mean

10 unemploy schizofrenia). Axis affective Axis anorexia had

of Diagnosis patients had

also ‘persisting

(28%

(52%

least.

dysfunctional somatoform psychoanalysis psychotherapy

years.

in the Borderline in were initially patients Institute. personality and der childhood healthy affective 12% 92% stress der 49% disorder, LPT 4 that than sonality (private der). ment, or relationships). 96% sonality and social (e.g. tion Unspecified. Authors patients). 28% disorders. 34% symptoms. Unspecified (private patients) or and

of tudies included in the review. patients S (unspecified Number psychotherapy)

(psychoanalysis (psychotherapy) (psychotherapy) (psychoanalysis) (psychotherapy)

psychotherapy) 62 and 30 25 36 11 500 psychoanalysis et et

1998 1998 1995

Meares, Authors

Table 10.1: al., Heinzel Keller al., Dührsen, 1986 Stevenson & 1992 Monsen al.,

174 - a at in at

- of - is - of

of 25 that and etc. insur after than II

they in

had

clinic

drop-

35 and treat

subject

but 31% 7 Data that ended 78% educa meas was

68%

group patients

Institut fulfil sampled, different

untrained to less

axis care started

Berlin and German

concluded in leave leave

30 general had

found

Psychology

that differ patients question Results

replied, that participated the (from not and 979 patients by

etc.) not treatment. follow-up.

Outpatient

addition patients All

90-94. group recruited. sick sick continuously 67

specified improved in say remarks

treatment not

with of of unsuccessfully

AOK of at do cases

Erkrankungen,

health

(including In

were

of German

Associations orientations

patients

outcome Relative

between regarding a 77 referrals insured patients the the

do treatment.

not

all

study participated 21

the reduction

DGIP), completers, into study.

analysts of received

remaining All been patients study. Jungian,

Analytical participated.

100 patients

to company.

and

participated questionnaires.

individual 87-88 first

psychoanalysis, that patients sample sample occupation

37 of

psychoanalyst).

between

General were Authors

PD, Only of and screened. patients were regarding

regarding with

Of short-term article.

LPT

111

Majority therapies

30 groups compared 25

a company.

significantly 62 of (55) had standard and

patients

broken

differences

Psychogene

this members data

months.

Pt Pt/IO Pt/IO HCI

Assessment HD: MC: MU: HD: MC: MU: SL: HD: MC: MU: SL: HD: MC: MU: SL: HD: MC: MU: SL: SL: design

Study Retrospective Prospective Prospective Retrospective Retrospective years.

years year years years

Follow-up 5,2 6 1 2 5 months months Time

months months

29-36 41 12 25,4 of

mean) Mean

¹ sessions number 80² 102³ 193 (psychoanalysis) 78 (psychotherapy) 166 (total 261 289 - 35 home. 44,5 28,6 29.4

unable - at begin

than assist days

level

more age: age: age age

Patient female. college married. previously female. married. female. married. government unem employees. more

further further owned low

work

disor- disor dura per disor- the

than

II I

20% sickest, severe major Patients

disorder, mention more than in practice practice

disorder disorder

years

subjects

assessed

problems:

factors absence dysfunction’ the

Mean

10 unemploy schizofrenia). Axis affective Axis anorexia had of Diagnosis patients had also ‘persisting

(28%

(52%

least.

dysfunctional somatoform psychoanalysis psychotherapy

years.

(psychoanalysis (psychotherapy) (psychotherapy) (psychoanalysis) (psychotherapy)

psychotherapy) 62 and 30 25 36 11 500 psychoanalysis et et

1998 1998 1995

Meares, Authors

al., Heinzel Keller al., Dührsen, 1986 Stevenson & 1992 Monsen al.,

175 in

not no

- severely study did and found

between

received that

401 article

returned

patients

therapy

psycho

of were 207

were

found

This 194 report

Patients

and

questionnaire

ended study

patients

multicenter

were

the Patients

psychoanalysis 90-93.

that baseline. 154

Authors differences

at interviewed, with

questionnaire.

psychoanalysis differences (74%). sample. deals a psychotherapy. differ were between between ill. Catamnestic patients Germany. therapy. Naturalistic HCI HCI Pt/HCI -- HCI --

HCI Pt/HCI

relatives)

HD: MC: MU: HD: MC: MU: SL: SL: sources,

Retrospective Prospective referral

records, years

Ongoing 2.9years 6 medical

months

48 Ongoing company (including

mean) mean)

observer Insurance

640 297 (psychoanalysis) 240 (psychotherapy) 440 (total (psychoanalysis) 70 (psychotherapy) 191 (total 275 Care

44, 37,2

time

Health age: age:

female. part college female. married. full Therapist Independent

Patient

Beutel

Pt: Th: IO: HCI: Mean years. 71% working. 30% working. 82% education. Mean years. 69% 52% 56% - Heinzel, well-

more severely

months

practice months

patients their treatment.

years.

was

5 of personality psychoneu Keller, 12

25,4

patients). 75% roses 52% disorder. Symptoms than report being compromised before 56% Unspecified (private during during

means

the

weekly weekly

users

from

(psychoanalysis) (psychotherapy) (psychoanalysis) (psychotherapy) days consultations

once twice

74 80 23 20 861 leave

guess, guess,

et &

mean medical medication hospital

sick 2003,

best best extrapolated

MC: MU: HD: SL: ³ ² ¹ Beutel 2002 Rudolf 2004 Total Rastin, al.,

176 Table 10.2: Sick leave from work (days/year). Study Pre Post Follow-up (years) Reduction Dührsen 32,4 no data 8,0 (5 years) Post: no data (1986) n=62 F-up: 75% 289 sessions Keller et al. 16 no data 8,0 (6 years) Post: no data (1998) n=47 F-up: 50% 166 sessions Heinzel et al. 14,5 8,5 5,9 (2 years) Post: 41% (1998) n=500 F-up: 59% 261 sessions Stevenson& 134,1 no data 41,1 (1 year) Post: no data Meares (1992) F-up: 69% n=30, 80sessions Beutel&Rastin 10,3 6,3* 4,0 (6 years) Post: 39%* (2002) n=94 F-up: 61% 440 sessions Rudolf 26,3 8,0** no data Post: 70%** (2003) n=43 F-up: no data 191 sessions Mean days 20,8 8,1 6,9 Range (10,3-134,1) (6,3-8,5) (4,0-41,1) Reduction (days, %) 12,7 (61%) 13,9 (67%) (base=pre-treatment) * ‘post’ means ‘1 year in treatment’; ** ‘post’ means ‘2 years in treatment’.

reduction of 68%. In the general population of BEK this is respectively 16 and 15 days in the same time frame, a reduction of 0,06%. Table 10.2 does not include the data of Monsen et al21 relating to patients with ‘severe personality disorders’. Instead of the number of sick leave days, the authors state the percentage of patients (n=21; 102 sessions) who were unemployed or on insured long-term sick leave: 55% at the start of treatment, 20% at treatment termination and 19% at follow-up (5 years). This is a sustained reduction of 45%. Before treatment, at termination and at follow-up the percentage of patients with skilled work or a job requiring at least three years education above senior high school was 15%, 25% and 53% respectively. The percentages of patients with a high monthly net income are respectively 10%, 20% and 48%. The authors point out that: ‘During the treatment period, the increase in income for this sample was 9.6% greater than the average increase for personal taxpayers in Oslo21. Apparently, these patients became not only healthier but also wealthier.

177 10.3.3 Days in hospital

Five studies, dealing with a total of 719 patients, report data on changes in hospitalisation rates, expressed in days per annum (Table 10.3). The follow-up period ranges from 1 year to 6 years, the mean duration is 2.6 years. The mean pre-treatment/post-treatment reduction is 85%, the mean pre- treatment/follow-up reduction 59%. The mean pre-post reduction is striking and seems to have largely disappeared at follow-up. However, this is explained by the fact that the study of Stevenson and Meares6, which has the largest number of hospital days at pre-treatment, presents no post treatment data. We have calculated the results of this table without this study that can be considered an ‘outlier’. Pre-treatment mean number of days in hospital is then 4,4 days, post treatment it is 1,2 days (reduction of 27%) and post-treatment it is also 1,2 days (reduction compared to pre-treatment remains 27%). for comparative purposes, Dührssen24 states the mean number of days in hospital per annum of all insured persons in Germany. In the period corre-

Table 10.3: Days in hospital (days/year). Study Pre Post Follow-up (years) Reduction Dührsen 10,0 no data 1,9 (5 years) Post: no data (1986) n=62 F-up: 81% 289 sessions Keller et al. 8,0 no data 1,0 (6 years) Post: no data (1998) n=58 F-up: 87% 165 sessions Heinzel et al. 3,4 1,2 1,2 (2 years) Post: 65% (1998) n=510 F-up: 65% 261 sessions Stevenson& 86,1 no data 44,1 (1 year) Post: no data Meares (1992) F-up: 49% n=30, 80 sessions Rudolf 3,4 1,3* no data Post: 62%** (2003) n=59 F-up: no data 297 sessions Mean days 7,8 1,2 3,2 Range (2,4-86,1) (1,2-1,3) (1,0-44,1) Reduction (days, %) 6,6 (85%) 4,6 (59%) (base=pre-treatment) * ‘post’ means ‘2 years in treatment’

178 sponding to the year preceding the start of treatment in her study, it was 4 days per annum In the period corresponding to the year preceding follow-up in her study, it was 3.5 days per annum. It can be concluded that before treatment the percentage of hospitalization days of the patients in this study was 150% higher than the average insured person; after treatment it was 46% lower.

10.3.4 Medical consultations

Three studies6,25,31 dealing with a total of 632 patients, report data on changes in outpatient medical consultations, expressed in number of visits per annum (Table 10.4). the follow-up period ranges from 1 year to 6 years, the mean follow-up duration is 2.6 years. The mean pre-treatment/post-treatment reduction is 54%; the mean pre-treatment/follow-up reduction is 56%. Predictably, in the study of Stevenson and Meares6 looking at borderline patients, the number of medical consultations is exceptionally high. We have calculated the results of this table without this study that can be considered an ‘outlier’. Pre-treatment mean number of medical consultations is then 5,4, post treatment it is 3,3 (reduction of 39%) and post-treatment it is 2,9 (reduction compared to pre-treatment 46%).Heinzel et al25 break down the results according to the type of visit. The reduction in the number of visits to general practitioners is 40% at treatment termination and 52% at follow-up. The reduction in the number of visits to medical specialists is 35% at treatment termination and 40% at follow-up. The

Table 10.4: Medical consultations (visits/year). Study Pre Post Follow-up (years) Reduction Beutel&Rastin 6,7 3,7* 4,0 (6 years) Post: 45%* (2002) n=111 F-up: 40% 438 sessions Heinzel et al. 5,1 3,2 2,7 (2 years) Post: 37% (1998) n=491 F-up: 47% 261 sessions Stevenson& 42,0 no data 6,0 (1 year) Post: no data Meares F-up: 86% (1992) n=30 80 sessions Mean days 7,1 3,3 3,1 Range (5,1-42,0) (3,2-3,7) (2,7-6,0) Reduction (days, %) 3,8 (54%) 4 (56%) (base=pre-treatment) * ‘post’ means ‘1 year in treatment’.

179 data of Keller et al26 (n=111; 165 sessions) regarding medical consultations are not included in table 4, because the authors do not provide figures regarding the initial situation. They do, however, report the number of medical visits (women 11 per annum, men 10 per annum) at follow-up (six years after treatment termination). They find that fifty percent of the patients reported ‘less frequent’ medical visits during the year preceding follow-up than during the year preceding the start of treatment. For comparative purposes, they refer to two studies27,28 in private practice. The mean number of consultations for women in both studies is 15 and 14 respectively, for men it is 12 and 11 respectively. Table 10.4 also does not include the data of Monsen et al21 (n=21; 102 sessions) regarding the use of health and social services, expressed as a percentage of patients with at least 10 contacts per annum (‘frequent users’). The percentage of frequent users during the year preceding the start of treatment, treatment termination and follow-up (5 years) is 70%, 20% and 30% respectively for somatic services, 30%, 0% and 24% for psychiatric services, and 10%, 0% and 0% for social services.

10.3.5 Medication users

Three studies20,21,26 dealing with a total of 190 patients, report data on medication users, expressed as a percentage of patients taking medication (Table 10.5).

Table 10.5: Medication use (percentage of patients taking medication). Study Pre Post Follow-up Reduction (years) Keller et al. 56% no data 45% (6 years) Post: no data (1998) n=111 F-up: 19,7% 165 sessions Rudolf 37,4% 11,8%* no data Post: 68%* (2003) n=59 F-up: no data 297 sessions Monsen et al. 50% 25% 14% Post: 50% (1995) n=20 F-up: 72% 102 sessions Mean (%) 49,6% 15% 40,3% Range (37-56) (11-25) (14-45) Reduction (%) 34,5% (70%) 9,3% (19%) (base=pre-treatment) * ‘post’ means ‘2 years in treatment’.

180 Mean follow-up period is 5.8 years. The mean pre-treatment/post-treatment reduction in medication users is 70%, and the mean pre-treatment/follow-up reduction 19%. The mean pre-post reduction is striking and seems to have largely disappeared at follow-up. However, this is explained by the fact that the largest study, that of Keller et al26, which has the largest percentage of medication users, presents no post-treatment data. Rudolf et al20 break down the results according to the treatment modality: the percentage of medication users before the start of treatment is 30.4% in psychoanalysis and 45.8% in psychotherapy. After 2 years of treatment, it is 15.6% in psychoanalysis and 7.4% in psychotherapy. This means a reduction of 49% in psychoanalysis and of 84% in psychotherapy. Table 10.5 does not include the data of Stevenson and Meares6. The authors do not report the percentage of medication users but the mean number of different drugs used per day: 3.8 in the year preceding therapy and 0.63 in the year succeeding it. This is a reduction of 83%.

10.4 Financial evaluation of the pre-post changes

Table 10.6a summarises the absolute pre-treatment/post-treatment reductions in sick leave days, hospital days, medical consultations and medication users. Applying present-day financial standards to these data (see method section 2.4 and Appendix 2) resulted in a financial evaluation shown in table 10.6b.

Table 10.6a: Pre-post changes in sick leave days, hospital days, medical consultations and medication. Sick leave & medical use Pretreatment n Posttreatment n Reduction per patient sick leave days/year 20,8 776 8,1 637 12,6 hospital days/year 7,8 719 1,2 569 6,6 medical visits/year 7,1 632 3,3 602 3,8 % medication users/year 50% 190 15,1% 79 34,5% Total

Table 10.6b: Monetary evaluation of pre-post changes. Costs per patient Pretreatment n Posttreatment n Reduction costs costs sick leave days/year € 5.994 776 € 2.347,19 637 € 3.646,43- hospital days/year € 2.007 719 € 311,74 569 € 1.694,82- medical visits/year € 278 632 € 128,10 602 € 149,43- % medication users/year € 135 190 € 41,27 79 € 93,89- Total € 8.413 € 2.828,29 € 5.584,57-

181 the mean costs of health care use and sick leave during the year preceding the start of treatment were € 8.413 per person. On average these costs were reduced to € 2.828 per person (a reduction of 66%) during the year preceding treatment termination. In that year, in other words, the savings amounted to € 5.584 per person. The reduction in sick leave was the largest contributor (65% of the savings).

10.5 Financial evaluation of the pre -follow-up changes

Table 10.7a summarises the absolute pre-treatment/follow-up reductions in sick leave days, hospital days, medical consultations and medication users. Applying present-day financial standards to these data (see method section 2.4 and Appendix 2) resulted in a financial evaluation shown in table 10.7b. the costs of health care use and sick leave during the year preceding follow- up (mean 2.9 years), compared to those during the year preceding the start of treatment, fell by 64%, i.e., € 5.371 per person. The reduction in sick leave accounted for 75% of the savings. tables 10.6 and 10.7 indicate that the benefits (i.e. the cost reductions) remained stable over time: 66% at treatment termination and 64% 2.9 years

Table 10.7a: Pre-follow-up changes in sick leave days, hospital days, medical consultations and medication. Sick leave & medical use Pretreat- Follow-up Reduction per patient ment n mean duration n in years sick leave days/year 20,8 776 3,0 6,9 733 13,9 hospital days/year 7,8 719 2,6 3,2 660 4,6 medical visits/year 7,1 632 2,7 3,1 632 4,0 % medication users/year 49,6% 190 5,8 40,3% 131 9,3% Total 2,9

Table 10.7b: Monetary evaluation of pre-follow-up changes. Pretreatment Follow-up Reduction Costs per patient costs n mean duration costs n in years sick leave days/year € 5.994 776 3,0 € 1.988,18 733 € 4.005,44- hospital days/year € 2.007 719 2,6 € 823,70 660 € 1.182,85- medical visits/year € 278 632 2,7 € 120,03 632 € 157,49- % medication users/year € 135 190 5,8 € 109,75 131 € 25,42- Total € 8.413 2,9 € 3.041,66 € 5.371,19-

182 later. The relative and absolute importance of the reductions in sick leave days increased during the years after treatment termination. Assuming that the benefits are similar in each year during these 2.9 years, and since the total costs of treatment averaged € 20.900 per person (see section 10.3.1), it can be calculated that the break-even point is reached by the savings achieved during the year before treatment termination and the three years thereafter.

10.6 Discussion

There is a fierce discussion in progress about the cost-effectiveness of psychotherapy and a scarcity of studies examining the cost effects of Long-term Psychoanalytic Therapy (LPaT). This systematic review examined the literature relating to the effects of LPaT on the use of health care and work impairment. It was found that, at treatment termination and at mean follow-up (2.9 years), the average reduction in hospital days was 85% and 59% respectively. The reduction in the number of medical consultations was 54% and 56%, 70% and 19% in medication users and 61% and 67% in sick leave days. Applying present-day financial standards (explained in section 10.2.4 and Appendix 2) to these data showed that, during the year preceding the start of psychotherapy, health care use and sick leave cost an average of € 8,413 per person. The average cost reduction per person during the year preceding treatment termination and during the year preceding follow up is 66% (€ 5,584) and 64% (€ 5,371) respectively. The most important economic gains result from the reduction in absenteeism, followed by the reduction of hospital days. Furthermore, there is evidence that the benefits are stable. As the average cost of LPaT is € 20.900 per person, the break-even point is reached approximately three years after treatment termination. Our review has several limitations. Most of the studies we included are not randomized controlled trials but cohort studies. However, apart from borderline patients, decisive RCTs in LPaT research present insurmountable, method- inherent feasibility problems.29 One of them is that the most informative control conditions (no treatment, waiting list or placebo) are not feasible in long-term treatment studies. Another problem is that comparing LPaT to alternative long- term treatments (if available at all) provides ‘RCT-worthy’ information only if a difference is found. If it is not found, the study might well be considered a (well- designed) cohort study examining two cohorts. Furthermore, psychoanalysis and psychotherapy groups present severe limitations as control groups for each other, as their baseline characteristics are likely to be different, influencing the indication for psychoanalysis or psychotherapy. Similarly, the comparison with general populations of health care insurance companies holds flaws, since these populations also are very different at baseline from the patient groups with whom they are compared. Furthermore, the strict design of RCTs requires naturalistic studies to demonstrate that a form of therapy works in the field.30 Consequently, cohort studies, that are a better reflection of actual clinical practice, provide the best available evidence as far as LPT is concerned.

183 Another limitation is that, although we used a checklist to assess the methodological qualities of the studies we included, it cannot be denied that the quality of the studies in our review was varied. Four of the included studies are retrospective24,25,26,31 and only three are prospective.6,20,21 Furthermore, some studies do not adequately describe treatments or diagnoses24,25 or do not use independent assessors.25 Four studies6,20,24,26 obtain data from objective sources such as insurance companies or medical records; the other three21,25,31 rely on data provided by patients or therapists. In the light of these methodological shortcomings, the data in this review should be interpreted cautiously. Prospective, adequately designed cohort studies in this field are necessary. Nevertheless, despite the methodological variation, the results of the studies in this review all point in the same direction. not all the studies provide both post-treatment and follow-up data. Our pre-treatment, post-treatment and follow-up calculations therefore all include different studies. The studies present no measures of variance to their mean outcomes, which obscures a broader perspective on the findings. Furthermore, there is one study (Stevenson and Meares6 ) that can be considered outlier. This study included only patients with severe personality disorders and because of its large outcomes it increases the variance of the review outcomes. Excluding the outlier diminishes the mean reductions in sick leave days, hospitalisation days and medical consultations. In our opinion these findings support the hypothesis that the sicker the patients are, the bigger the gains can be. In addition, we found only seven suitable studies, which is a rather small sample for a review. On the other hand, the number of patients in our review is large. Another shortcoming is that most studies do not provide separate data for psychotherapy and psychoanalysis. Consequently, we could not differentiate between the costs and benefits of standard psychoanalysis and those of psychoanalytic psychotherapy. Nor do the studies provide data about dose-response relationships, and they therefore provide no answer to the question of whether there is an optimal LPaT duration for a maximum economic benefit. As we made no comparison with studies of time-limited psychotherapies, we cannot address the question whether short-time treatments may yield similar results. In this context we can only refer to the general literature, which contains indications that the effects of short-time therapies are short-lived [e.g.32,33]. Our study also suffers from the limitation that, although we furthered homogeneity by restricting ourselves to LPaT, there remained some heterogeneity. furthermore, because the data mostly use naturally occurring termination of the therapies (and the follow-up thereafter), it is possible that a selection bias occurs. Healthier patients are likely to be able to end therapy after a few years, while more sicker patients or those not getting better might not be included in the samples (for instance because they are still in treatment). This might bias the results towards stronger treatment effects. some patients were still in therapy at the time of follow-up. This means that some of the sustained benefits of LPaT could be due to this effect. although all studies show improvement in outcomes regarding sick leave and health care costs, it is evident that the lack of comparable control groups

184 complicates the interpretation of the data. We can not present hard evidence that the improvements would not have occurred over time in the lives of these patients. In addition, positive group averages do not mean that some individual patients might not improve, or even deteriorate. finally, as many patients in the studies used medication alongside their psychotherapies, it is likely that a part of the effects could be subscribed to the pharmacotherapy. Our financial evaluation also has limitations. Firstly, we applied present-day financial standards in the Netherlands (explained in section 10.2.4 Financial evaluation, and in Appendix 2) to the results of studies performed in other times in other countries. Our financial evaluations do not therefore reflect the ‘there and then’ costs and benefits of the treatments. As costs and gains indubitably vary across countries, our conclusions cannot be generalised to other states. However, as we provide the gross numbers for changes in health care use and sick leave, local financial standards can easily be applied to the data. Secondly, a number of factors might lead to either overestimation or underes- timation of the benefits. We cannot exclude the possibility that LPaT results do not last for three years after termination. However, we found indications that they steadily increase even after that period of time.25,31 In the first scenario, we overestimate LPaT benefits; in the second we underestimate them. In addition, we may underestimate the treatment costs by not including indirect costs, e.g., time loss and travelling costs for patients and supervision or intervision costs for therapists. On the other hand, it seems a fair assumption that well treated patients do better in their careers, and therefore earn higher incomes, pay extra taxes and consume more.21 We therefore most probably underestimate LPaT benefits by not including (because of a lack of data) factors such as the detrimental effects of mental problems on work performance. Finally, our calculations refer almost exclusively to employed patients. The LPaT benefits in patients living from social benefits would of course not include any reduction in sick leave but it may be argued that at least some of them would regain the capacity to participate in social and economic intercourse. the study of Monsen et al21 supports this assumption as the level of employment rises from 35% pre therapy to 50% at treatment termination and 62% at follow-up. The study of Doidge et al22 shows that 15% of the patients in analysis in Australia went from unemployment to employment (4 years into treatment) and in the U.S. there was an 4.5% increase in employment (4 years into treatment). Our calculation regarding the break-even point of treatment costs and benefits is based on our findings that the benefits at treatment termination and at follow-up (mean follow-up 2,9 years) seem to be similar. Of course, our assumption is that they are also the same in all three years between treatment termination and follow-up. This might not be the case, clearly influencing the point of break even. However, until now we have no information on the course of the costs between these two measurement points. finally, our estimations of the costs of medication use are based on the average costs of medication during a certain year. As it is unlikely that the patients presented in the studies of this review present the ‘average patient’,

185 there could be an under or overestimation of these costs. It might be helpful to calculate the costs based on the number of medications that patients used, but there were not enough data to do so. In addition, again for lack of data we presented only the number and percentages of patients that used medication, which is a very rough estimation of true changes in medication costs. to our knowledge, our review is the first one to summarise the available data regarding the effects of LPaT on health care use and work impairment in adult, LPaT outpatients, and translate them into financial terms. Despite its limitations, we consider it a valid, albeit preliminary, evaluation. It can be concluded that patients treated with LPaT present enduring reductions in the use of health care services and sick leave days, and that these reductions probably offset the costs of treatment in three years after termination. Needless to say, what really matters in life, being unutterable in economic terms, remains undiscussed.

References

1. Mumford E, Schlesinger HJ, Glass GV, et al. A new look at evidence about reduced cost of medical utilization following mental health treatment. Am J Psychiatry 1984; 141: 1145-1158. 2. Gabbard GO, Lazar SG, Hornberger J, et al. The economic impact of psychotherapy: a review. Am J Psychiatry 1997; 154: 147-155. 3. Baltensperger C, Grawe K. Psychotherapie unter gesundheitsökonomischem Aspekt. Zeitschr Klin Psych Psychoth 2001; 30: 10-21. 4. Zielke M. Wirksamkeit der stationärer Verhaltenstherapie. Weinheim:Psychologie Verlag Union, 1993. 5. hall J, Caleo S, Stevenson J, et al. An economic analysis of psychotherapy for Bor- derline Personality Disorder patients. J Ment Health Pol Econ 2001; 4: 3-8. 6. stevenson J, Meares R. An outcome study of psychotherapy for patients with borderline personality disorder. Am J Psychiatry 1992; 149: 358-362. 7. Bateman A, Fonagy P. Health service utilization costs for Borderline Personality Disorder patients treated with psychoanalytically oriented partial hospitalisation versus general psychiatric care. Am J Psychiatry 2003; 160: 169-171. 8. fonagy P, Kachele H, Krause R, et al. An Open Door Review of Outcome Studies in Psychoanalysis. London: International Psychoanalytical Association, 1999. 9. Brandl Y, Bruns G, Gerlach A, et al. Psychoanalytische Therapie: eine Stellungnahme fur die wissenschaftliche Öffentlichkeit und für den Wissenschaftlichen Beirat Psy- chotherapie. Forum Psychoanalyse 2004; 20: 13-125. 10. Leichsenring F. Are psychodynamic and psychoanalytic therapies effective? A review of empirical data. Int J Psychoanalysis 2005; 86: 841-868. 11. Wallerstein R. Forty-two lives in treatment: A study of psychoanalysis and psychotherapy. New York: Guilford Press, 1986. 12. Wallerstein R. The Psychotherapy Research Project of the Menninger Foundation: An overview. J Cons Clin Psychology 1989; 57: 195-205. 13. Stanton AH, Gunderson JG, Knapp PH, et al. Effect of psychotherapy in schizophrenia: I. Design and implementation of a controlled study. Schizophr Bull 1984; 10: 520-563. 14. Fonagy P, Moran GS. Studies of the efficacy of child psychoanalysis. J Cons Clin Psychology 1991; 58: 684-695. 15. Quiroga SE, Zonis R, Zukerfeld R. Conducta alimentaria, peso corporal y psycopato- logia en mujeres ingresantes a la Universidad de Buenos Aires. Revista del instiuto de investigaciones de la facultad de psicologia 1998; 3.

186 16. Koningsveld EAP, Zwinkels WS, Mossink JCM, et al. Maatschappelijke kosten van arbeidsomstandigheden van werknemers in 2001 (Social costs related to employee working conditions in 2001). Doetinchem: Elsevier; 2003. 17. Ministerie van Sociale Zaken en Werkgelegenheid in aanvulling met de meest recente loongegevens van CBS, over 2004 (Ministry of Social Affairs and Employment com- plemented with recent salary data of 2004). Source: Centraal Bureau voor de Sta- tistiek (Central Bureau of Statistics): Arbeidsrekeningen, 2004). 18. Oostenbrink JB, Bouwmans CAM, Koopmanschap MA, et al. Manual for Costing: Methods and Standard Costs for Economic Evaluations in Healthcare (in Dutch) iMTA (Institute for Medical Technology Assessment), Erasmus Medical Center Rot- terdam) by order of the Health Care Insurance Board (CVZ), 2004. 19. Data and facts 2005. Organisation of Pharmaceutical Statistics The Netherlands (in Dutch) Stichting Farmaceutische Kengetallen, The Hague, 2005. 20. Rudolf G, Dilg R, Grande T, et al. Effektivität und Effizienz psychoanalytischer Langzeittherapie: Die Praxisstudie Analytische Langzeittherapie. In: Gerlach, A, Schlösser A-M Springer A, editors. Psychoanalyse des Glaubens. Giessen: Psychoso- zial-Verlag 2004. p. 515-528. 21. Monsen J, Odland T, Faugli A, et al. Personality disorders and psychosocial changes after intensive psychotherapy: a prospective follow-up study of an outpatient psychotherapy project, 5 years after end of treatment. Scand J Psychology 1995; 36: 256-268. 22. Doidge N, Simon B, Brauer L, et al. Psychoanalytic patients in the U.S., Canada, and Australia: I. DSM-III-R disorders, indications, previous treatment, medications, and length of treatment. JAPA 2002; 50: 575-614. 23. Friedman RC, Garrison WB, Bucci W, et al. Factors affecting change in private psychotherapy patients: an effectiveness study. J Am Acad Psychoanal Dyn Psychiatry 2005; 33: 583-610. 24. Dührssen A. Dynamische Psychotherapie, Psychoanalyse und analytische Grup- penpsychotherapie im Vergleich. Zeitschr Psychosom Med 1986;32:161-180. 25. Heinzel R, Breyer F, Klein T. Ambulante analytische Einzel- und Gruppenpsycho- therapie in einer bundesweiten katamnestischen Evaluationsstudie. Gruppenpsy- choth Gruppendynamik 1998; 34: 135-152. 26. Keller W, Westhoff G, Dilg R, et al. Study and the study group on empirical psychotherapy research in analytical psychology. Berlin: Department of Psychosomatics and Psychotherapy, University Medical Center Benjamin Franklin, Free University of Berlin, 1998. 27. Hoffmeister J, Hoeltz J, Schön D, et al. Nationaler Untersuchungs-Survey und regionale Untersuchungs-Surveys der DHP (Deutsche Herzkreislauf-Praeventions- studie). DHP Forum 3; Heft 1, 1988. 28. Schacht E, Schwartz FH, Kerek-Bodden HE. Die EvaS-Studie. Eine Erhebung über die ambulante medizinische Versorgung in der BRD. Köln: Zentralinsitut für die kassenärztliche Versorgung, 1989. 29. Maat SCM de, Dekker JJ, Schoevers RA, et al. The effectiveness of long-term psychotherapy: Methodological research issues. Psychotherapy Research, 2007, 17 (1): 59-65. 30. Leichsenring F. Randomized controlled versus naturalistic studies: a new research agenda. Bull Menninger Clinic 2004; 68: 137-151. 31. Beutel M, Raskin M. Long-term treatments from the perspectives of the former patients. In: Leuzinger-Bohleber M, Target M, editors. Outcomes of psychoanalytic treatment: Perspectives for therapists and researchers. London/Philadelphia: Whurr Publishers 2002. p. 130-142. 32. Gloaguen V, Cottraux J, Cucherat M, et al. A meta-analysis of the effects of cognitive therapy in depressed patients. J Affect Disorders 1998; 49: 59-72. 33. Hollon SD, Jarrett RB, Nierenberg, AA, et al. Psychotherapy and medication in the treatment of adult and geriatric depression: which monotherapy or combined treatment? J Clin Psychiatry 2005; 66: 455-468.

187 Appendix 1

List for Quality assessment of Cohort Research

A. Study design 1. Prospective study 2. Matched control group 3. Pre-post assessments

B. Patients 4. Clear inclusion and exclusion criteria 5. Clear diagnostic description of patients 6. Adequate size of patient sample

C. Interventions 7. Clear description of treatment 8. Trained professionals 9. Adherence

D. Assessments 10. Clear outcome measures 11. Reliable and valid measure instruments

E. Data 12. More than one assessment source 13. Adequate statistical methods 14. ITT or Per-protocol analyses

F. Results 15. Adequate and clear description of the results 16. Acceptable Pre-Post drop-out 17. Check for possible confounders

G. Follow-up 18. Minimum follow-up of 1 year 19. Acceptable drop-out in follow-up 20. More than one assessment source

188 Appendix 2

2003 2004 2005 Consumer inflation index CPB 1,2% 1,8% ambulatory consult in an institution € 71,54 € 72,40 € 73,70 ambulatory consult psychiatrist € 76,00 € 76,91 € 78,30 average costs of a therapy session € 73,77 € 74,65 € 76,00 productivity per hour € 34,98 € 35,40 € 36,04 costs of productivity loss per 8 hour day € 279,84 € 283,20 € 288,30 costs day in psychiatric hospital € 250,00 € 253,00 € 257,55 costs of a GP consult € 20,20 € 20,44 € 20,81 costs of a medical specialist consult € 56,00 € 56,00 € 57,01 average costs of a medical consult € 38,10 € 38,22 € 38,91 costs medication million Euro € 589,00 € 599,60 total amount of patients using medication million patients 2,2 2,2 average costs of medication per patient € 267,73 € 272,55

189

Chapter 11 General discussion

The results of this thesis are summarized by the answers to the eleven research questions formulated in the introduction. Paragraphs 11.1-11.11 address the eight questions regarding short-term (psychoanalytic) psychotherapy. Para- graphs 11.12-11.18 address the three questions regarding long-term psychoanalytic therapy.

Short-term (psychoanalytic) psychotherapy for depression

11.1 What is the efficacy of short-term psychotherapy compared to that of pharmacotherapy in ambulatory psychiatric outpatients presenting a major depressive disorder (Q.01, chapter 2)?

The main question of this meta-analysis, based on RCTs published between 1980 and 2005, was to determine the relative efficacy of time-limited (max. 6 months) psychotherapy and pharmacotherapy in the treatment of ambulatory, psychiatric outpatients diagnosed with unipolar, non-psychotic major depressive disorder. An important limitation of many previous reviews and meta-analyses concerning this topic may be the striking methodological and clinical heterogeneity of the studies included. Clinical heterogeneity refers to differences in patient samples, treatment protocols and treatment settings between studies. To enhance the clinical homogeneity among the studies in this meta-analysis, strict inclusion criteria regarding patient samples, diagnoses and treatment settings were applied. Ten studies were included, considering 1.233 patients (640 treated with pharmacotherapy and 593 treated with psychotherapy). Mean HDRS score across all studies was approximately 21 (by our definition moderately depressed). One study considered IPT; the remaining studies applied Cognitive-Behavioural Therapy (CBT). The dropout rate in pharmacotherapy (28.43%) was significantly higher than in psychotherapy (23.60%). The (HDRS measured) remission rates of psychotherapy (38%) and pharmacotherapy (35%) at treatment termination did not differ. This held for both chronic and non-chronic, mild or moderate depression. Both treatments performed better in mild (HDRS<20) than in moderate (HDRS>20) depression. At follow-up, relapse in pharmacotherapy (56.56%) was larger than in psychotherapy (26.51%).

191 11.2 What is the efficacy of short-term psychotherapy compared to that of short-term psychotherapy plus pharmacotherapy in ambulatory, psychiatric outpatients presenting a major depressive disorder? (Q.02, chapter 3)

The main question of this meta-analysis, based on RCTs published between 1980 and 2005, was to determine the relative efficacy of time-limited (max. 6 months) psychotherapy and combined therapy (psychotherapy and antidepressants) in the treatment of ambulatory, psychiatric outpatients diagnosed with unipolar, non-psychotic major depressive disorder. The heterogeneity problem discussed in 11.1 applied here likewise. Seven studies were included, considering 903 patients (459 treated with psychotherapy and 444 treated with combined therapy). Mean HDRS score across all studies was approximately 21 (by our definition moderately depressed). One study considered psychoanalytic therapy; the remaining studies applied (a form of) CBT. Dropout rates for combined therapy (25%) and for psychotherapy (24%) did not differ statistically. The remission rates (HDRS measured) showed that the combined therapy (46%) did better than psychotherapy (34%). However, these results depend on severity and chronicity. In mild, non-chronic depression combined therapy (42%) and psychotherapy (37%) performed equally. No data on mild, chronic depression were found. In moderate, non-chronic depression remission rates did not differ either (combined therapy 44% and psychotherapy 39%). In moderate, chronic depression combined therapy (48%) was superior to psychotherapy (32%).

11.3 What is Short-term Psychoanalytic Psychotherapy (SPSP)? (Q.03, chapter 4)

SPSP is a face-to-face, six-month, individual psychotherapy consisting of sixteen sessions (first eight weekly, then eight fortnightly sessions). SPSP’s primary goal is to cure depression. The secondary goal is to reduce a patient’s vulnerability to depression. The latter is conceived as the shaping or altering of internal relationships, especially the relationship the patient has with himself (IntraPersonal Relationship, IPR). Taking into account SPSP’s restricted number of sessions, it is all too obvious that personality change will be limited. Setting, frame and contract are ingredients of the supportive approach. The therapist’s attitude is expected to be basically interested, accepting, affirmative, empathic, concerned, helpful, patient, actively expectant, perseverant, honest, transparent and flexible. The therapist is supposed to promote, maintain and, if necessary, restore a supportive therapeutic relationship, to the best of his abilities. Admittedly, up to now the supportive technique has been less well articulated than the interpretive one. Important aspects of it are containing (Bion, 1962), holding (Winnicott, 1965), mirroring (Kohut, 1977), developmental help (Fonagy and Target, 1996) and the judicious blending of homeostatic (‘maternal’) and disruptive (‘paternal’) attunement.

192 spsp unfolds as a discourse in which nine levels are distinguised. At the first level the focus is on the patient’s physical and psychological complaints and symptoms. At the second level the focus shifts from symptoms and complaints to life circumstances, which somehow influence or have influenced the depression. At the third level the focus shifts from life circumstances to one or more external, interpersonal relationships. The fourth level discusses one or more relational patterns in the patient’s life. Level five addresses an internal issue: the patient’s attitude in life. At the sixth level the focus shifts to an epigenetic explanation of level five. It is about how past relationships persist in the patient’s actual life. The seventh level concerns the relationship the patient maintains with herself as the consequence of identification with internal-interpersonal relationships. At the eighth and ninth level the focus shifts to how problems discussed at the levels four to seven manifest themselves in the relationship with the therapist. SPSP often starts at level one and in many cases rises to level five. Levels six and seven are reached considerably less often. Level eight may be reached for some time with some patients but can hardly be worked through fully. Level nine is unattainable in the short time frame of SPSP.

11.4 How may adequate, psychoanalytically defined support contribute to personality change? (Q.04, chapter5)

The aim of this publication was to propose a psychoanalytic definition of support and to elaborate on its role as an agent of change in psychoanalytic therapies (be they short or long). In the classic, psychoanalytical views, interpretation leads to insight, which in turn results in personality change, if all goes well. These views are not challenged but it is contended that a second, largely neglected process also plays an important part: support may lead to personality change, which in turn may result in insight. essential to this proposition is the notion that internal relationships represent vital personality aspects. Their alteration is crucial to personality change. The longstanding interpretation versus relationship controversy is reformu- lated as an interpretation versus support debate. The proposed psychoanalytic definition of support is: the proper gratification of unmet developmental needs, as they appear in the archaic aspects of the therapeutic relationship. Then, the controversy comes down to divergent views on the relative roles of interpretation and gratification. the importance of differentiating between adequate and inadequate support is stressed. Some ideas about the putative mode of action of support are presented, suggesting that it may reside in its power to evoke, in the patient, the experience of a ‘dissonance’ between two incongruous aspects of the therapeutic relationship: a ‘malignant’ an a ‘benign’ one. Both are simultaneously felt in the present, the benign one being determined by the present, the malignant one by the past. If the benign aspect of the relationship prevails over the malignant one and if it becomes internalised, internal relationships, especially ‘how I relate to Myself’ (the IntraPersonal Relationship, IPR), may change for the better.

193 11.5 In ambulatory psychiatric patients presenting a major depressive disorder, what is the efficacy of SPSP compared to pharmacotherapy (Q.05), the efficacy of SPSP compared to SPSP plus pharmacotherapy (Q.06), and the efficacy of SPSP plus pharmacotherapy compared to Pharmacotherapy (Q.07)? (Chapter 6)

This study assessed the relative efficacy of SPSP, pharmacotherapy and their combination in the treatment of ambulatory, psychiatric outpatients diagnosed with unipolar, non-psychotic major depressive disorder. The study was based on the data of three consecutive RCTs, performed and published by the Depres- sion Research Team of JellinekMentrum in Amsterdam. Based on the results of the original trials, three hypotheses were formulated: Combined therapy is more efficacious than pharmacotherapy (1) and SPSP (2), and SPSP is more efficacious than pharmacotherapy (3). the results showed that, as intended, the patient population was mildly to moderately depressed (mean HDRS 18.7). The drop-out in pharmacotherapy was 38%, in combined therapy it was 28% and in SPSP it was 25%. The dropout rates for the three conditions did not differ significantly. Drop-out rates of both SPSP and pharmacotherapy were lower in the combined therapy condition than in mono-therapy conditions (10% versus 25% for SPSP and 25% versus 38% for pharmacotherapy). the first hypothesis was confirmed, the other two were rejected. Accord- ing to the independent observers (HDRS), the main criterion in this study, there was no significant difference in remission rates between SPSP (31%) and pharmacotherapy (24%) (which answers Q.05), nor between combined therapy (40%) and SPSP (31%) (which answers Q.06). However, combined therapy (40%) outperformed pharmacotherapy (24%) (which answers Q.07). Secondary criteria showed that, according to the therapists (CGI-S), there was no significant difference between SPSP and combined therapy, and both were superior to pharmacotherapy. According to patients’ opinions on symptom improvement (SCL-D), combined therapy was superior to pharmacotherapy and to SPSP, the latter outperforming pharmacotherapy. According to patients’ evaluations of quality of life (QLDS), there was no significant difference between SPSP and combined therapy, nor between SPSP and pharmacotherapy, but combined therapy was better than pharmacotherapy. Figure 11.1 presents the results graphically. The symbol > indicates a statistically significant difference at the p < .05 level. NS means no statistically significant difference

11.6 What are the answers to the questions Q.05-07 at the level of two HDRS factors and three SCL-90 subscales of: Depression, Anxiety and Somatic complaints? (Q.08, chapter 7)

This study presented a second mega-analysis, using a similar design as the one in chapter 6. However, the relative efficacy of SPSP, pharmacotherapy and

194 > (HDRS) > (CGI-S) > (SCL-D) > (QLDS) Pharmaco CT Therapy

NS (HDRS) NS (HDRS) NS (CGI-S) > (CGI-S) SPSP > (SCL-D) > (SCL-D) NS (QLDS) NS (QLDS)

Figure 11.1: Comparative efficacies of combined therapy (CT), Short Psychoanalytic Supportive Psychotherapy (SPSP) and Pharmacotherapy. their combination was now assessed examining HDRS factor levels and three SCL subscale levels. In this study, a factor analysis of the mega-analysis sample yielded two main factors, which were labelled ‘Mental’ and ‘Somatic’. A mega- analysis was performed at these HDRS levels and at those of three SCL subscales: Depression (SCL-D), Anxiety (SCL-A) and Somatic complaints (SCL-S). Three hypotheses were formulated: 1. spsp outperforms pharmacotherapy on the ‘Mental’ factor of the HDRS (1.a), the SCL-Depression (1.b) and the SCL-Anxiety (1.c); 2. Pharmacotherapy is better than SPSP on the HDRS ‘Somatic’ factor (2.a) and the SCL-Somatic complaints (2.b); 3. Combined therapy outperforms both monotherapies on all sub-dimensions (3.a-j).

At the level of the HDRS ‘Mental’ factor, we found there were no differences between SPSP and pharmacotherapy (1.a rejected), nor between SPSP and combined therapy (3.a rejected), but the latter did outperform pharmacotherapy (3.b confirmed). At the level of the HDRS ‘Somatic’ factor, no differences were found between SPSP and pharmacotherapy (2.a rejected), nor between pharmacotherapy and combined therapy (3.c rejected), but the latter did outperform SPSP (3.d confirmed). At the level of the SCL-D, no differences were found between SPSP and pharmacotherapy (1.b rejected), and combined therapy outperformed both (3.e-f confirmed). At the level of the SCL-A, SPSP equalled combined therapy (3.g rejected) and both were superior to pharmacotherapy (1.c and 3.h confirmed). At the level of the SCL-S, no differences were found between the two treatments (2.b and 3.i-j rejected). in short it could be stated that the first hypothesis is rejected (two out of three comparisons rejected), as is the second (two out of two comparisons rejected), and the third hypothesis seems undecided so far (five of ten comparisons rejected). Most differences were found between combined therapy and

195 NS (HDRS Somatic) > (HDRS Mental) > (SCL-Depression) > (SCL-Anxiety) NS (SCL-Somatic) Pharmaco CT Therapy

> (HDRS Somatic) NS (HDRS Somatic) NS (HDRS Mental) NS (HDRS Mental) > (SCL-Depression) SPSP NS (SCL-Depression) NS (SCL-Anxiety) > (SCL-Anxiety) NS (SCL-Somatic) NS (SCL-Somatic) Figure 11.2: Comparative efficacies of combined therapy (CT), Short Psychoanalytic Supportive Psychotherapy (SPSP) and Pharmacotherapy at HDRS factor level and SCL subscale level. pharmacotherapy, favouring the first. The difference between the SCL-D outcomes of the this mega-analysis and the previous one, can be explained by the fact that the first considered remission rates and the second mean end scores. figure 11.2 presents the results graphically. The symbol > indicates a statistically significant difference at the p < .05 level. NS means no statistically significant difference.

11.7 Comments on the results

Regarding the investigation of short-term psychotherapy for depression (in ambulatory, psychiatric patients), it can be concluded that the studies in the meta-analyses mostly concerned cognitive behavioural psychotherapy (CBT). Therefore, we refer to these psychotherapies as CBT hereafter. The results of the meta-analyses show that CBT and pharmacotherapy are equally efficacious at treatment termination. At follow-up, CBT seems to outperform pharmacotherapy, revealing lower relapse rates. Combined therapy and CBT are also equally efficacious, except for moderate, chronic depression. In that depression subtype combined therapy outperforms psychotherapy. In all, the differences seem to be small. A meta-analysis comparing combined therapy and pharmacotherapy has not been performed in this thesis. However, a recent meta-analysis (Pampallona et al., 2005) concluded that combined therapy was superior to pharmacotherapy. pharmacotherapy, CBT and their combination are established treatments of depression. SPSP adds to this an alternative with totally different characteristics. SPSP is a new branch on the STPP tree (Short-Term Psychodynamic Psychotherapy, Markowitz et al., 1998). SPSP has been developed for treating

196 depression. The emphasis is on supportive attitude, relationship and techniques to counter regression and foster psychological growth. Although adequate, psychoanalytically defined support could be a therapeutic factor in personality change, it is obvious this change is limited in a short-term treatment such as SPSP. In treatments of longer duration support may well have more changing power. viewing the results of two mega-analyses investigating SPSP, it may be tentatively stated that the efficacy of SPSP in the treatment of psychiatric outpatients with mild to moderate depression is comparable to that of pharmacotherapy and to that of combined therapy. Combined therapy outperforms pharmacotherapy. The results vary when HDRS factors and SCL subscales are taken into consideration, but pharmacotherapy never outperforms SPSP or combined therapy. In all, the differences seem to be small. It may come as a surprise that SPSP is more potent in reducing anxiety than pharmacotherapy, with SSRI’s and SNRI’s being first choice anxiolytic drugs. The strong emphasis on support in SPSP might be an explanation for this effect. Making a ‘historical’ comparison (not a head-to-head comparison) between short-term, cognitive behavioural psychotherapy and SPSP, it may be tentatively concluded that their efficacies are comparable. First, their remission rates seem similar (31% for SPSP, 34%-38% for CBT). Second, both equal combined therapy and pharmacotherapy in efficacy, except,in moderate chronic depression, it would appear. Combined with pharmacotherapy, they both seem superior to pharmacotherapy. These views are corroborated by the findings of Leichsenring (2001). In a meta-analysis of six studies comparing STPP and CBT in depression, he concludes that in 58 of the 60 comparisons (97%) ‘(…) no significant difference could be detected between STPP and CBT/BT concerning the effects in depressive symptoms, general psychiatric symptomatology and social functioning.’ the question may arise whether there is a need for yet another short-term therapy for depression. But as ‘patients presenting a major depressive disorder’ still constitute a considerably heterogeneous category, it seems unlikely that just a few methods will suffice to treat all of them adequately. On the contrary, it seems plausible that some patients will profit from pharmacotherapy, others from CBT and still others from SPSP (or from their combinations). Admittedly, matching patient and therapy is still a vexing riddle. it can thus be concluded that there are several effective short-term methods for treating depression today, but their short-term efficacy seems limited. Remission rates seldom reach fifty percent. There is no indication that SPSP is better in this respect. For many patients the short-term treatments do not seem to be enough. In addition, remission, if achieved, may be short lived after treatment termination. Is success obtained by short-term therapy ephemeral? Pharmacotherapists seem convinced, with the result that continued and maintenance pharmacotherapy constantly gets longer. The NICE guidelines (2004) for instance, recommend to continue medication for 2 years when a patient has experienced two or more depressive episodes in a recent time period. An important shortcoming of the research into time-limited psycho-

197 therapy for depression is that few studies investigate their long-term effects. Based on some of the studies that do, there are indications that many patients relapse some time after treatment termination (at follow-up). Although we did not investigate this topic extensively, we did find relapse percentages of 27% - 57% in the meta-analysis of this thesis comparing psychotherapy and pharmacotherapy. In their meta-analysis of treatment for depression, Westen and Morrison (2001) find that, on average, only 37% of the patients who completed treatment and initially improved remain so at follow-up. They conclude: “By any standard, it is difficult to construe these data as evidence for the hypothesis that these treatments show genuine efficacy for the treatment of depressive disorders” (p. 886). Over a period of 15 years, recurrence rates may rise as high as 85% (Mueller et al., 1999). Koelen et al. (submitted) performed relapse analyses on the Mentrum RCT considering 6 months combined therapy (SPSP and pharmacotherapy). They showed that of the remitted patients, only 17.7% had stayed so two years later. Although not explicitly investigated, it is not to be expected that relapse rates in both monotherapies (SPSP or pharmacotherapy) are any better than in combined therapy. the Nemesis cohort described by Spijker et al. (2000) found that, among community living persons, 50% of the initial depressions were cured (with or without treatment) within 3 months and 20% had a chronic course. It is likely that the latter group ends up in secondary health care and that these patients suffer from more severe, chronic and treatment resistant depressions. Although consecutive treatment steps in refractory patients may have beneficial effects, remission rates appear to diminish with each step in the sequential treatment protocol (see the study Sequenced Treatment Alternatives to Relieve Depres- sion (STAR-D, Rush et al., 2006)). Depression, especially when more severe and chronic, seems a though ‘enemy’ to fight. This may partially explain why, despite the enormous efforts in developing treatments and investigating their efficacies, short-term treatments for psychiatric outpatients have modest and, probably, short-lived effects.

11.8 Comments on the methodology

11.8.1 Study design

The conclusions of the first part of the thesis are based on RCT’s. RCTs are rightfully at the high end of the hierarchy of scientific evidence. Within reasonable limits, this method ensures, like no other, that chance is the only alternative explanation for the differences found between the studied groups. Factors possibly (co)determining results (‘confounders’) are neutralised as much as possible. Besides these considerable advantages randomized trials have, they present some well known limitations. A major one is selection bias. RCT’s leave patients with serious comorbidity, such as drug dependence, suicide intentions or severe personality disorders out of scope. In fact, the majority of suitable patients do not end up in RCTs, due to all in- and exclusion criteria that have

198 to be met. Keitner et al. (2003), e.g., mentioned that only 14.5 % of eligible depressed patients eventually took part in an RCT. The RCTs mentioned in this thesis (in the meta-analyses as well as in the mega-analyses) suffered from this same bias. Therefore its conclusions are limited to what is probably a small part of the depressed population. Finally, RCTs work with protocols and manuals and so ensure a controlled situation. This may not reflect day-to-day clinical practice. It may be reasoned that in clinical practice, the results of time limited treatments are poorer than those measured in RCTs. First, in clinical practice there are more patients with comorbid disorders, suicidal intentions and psychotic features. Second, patients enrolling in an RCT receive more attention, for instance at all measurement points. Finally, the awareness of participating in a study might motivate both therapists and patients alike to ‘work’ harder. the mega-analyses were based on three trials performed by the same research group in the same population. As they used an identical research design in similar study populations, high levels of clinical and methodological homogeneity can be expected. One limitation of mega-analysis is that the comparison of SPSP to pharmacotherapy was not based on directly randomized subjects. However, both groups stemmed from the same population and the studies were performed only a short time after each other. In addition, their baseline characteristics did not differ significantly. the conclusions regarding the relative efficacies of SPSP and (cognitive behavioural) psychotherapy were based on comparing the findings of a mega- analysis based on three Mentrum trials to the results of two meta-analyses based on literature. This too is a ‘historical’ comparison with the evident drawback of lacking randomization. Yet we tried to enhance the homogeneity between the mega-analysis and the meta-analyses, to the best of our abilities. We therefore assume cautious conclusions may be drawn.

11.8.2 Patients and treatments

The two meta- and the two mega-analyses aimed at furthering clinical homogeneity. The meta-analyses pursued this by applying rather strict inclusion criteria regarding patients (only adult, ambulatory, psychiatric outpatients with major depression), treatments (only formal psychotherapy and adequate pharmacotherapy) and treatment setting (ambulatory). The mega-analyses were applied in a rather homogeneous study sample, resulting from the three RCTs featuring the same selection criteria. However, homogeneity has advantages and disadvantages. Its strength is that it allows conclusions on relatively specific treatments for specific patients to be drawn. Its limitation is that it narrows down the number of suitable trials and, more importantly, homogeneity limits the generalizability of the results. the study sample of the mega-analyses consisted of employed people, living alone, with an education higher than average. They suffered from mild to moderate depression. Generalizing these findings to other populations (with lower educational levels or with more severe depression) may be hazardous. So

199 this thesis focuses on patients with mild to moderate depression, with severe depression remaining out of scope.

11.8.3 Measurements of effects

The primary outcome of efficacy in the clinical trials was the HDRS, reflecting the opinion of the ‘independent observer’. Most of the studies in the meta- analyses used this as main criterion. They did not assess social functioning or quality of life, nor patient opinion. Therefore the results of the meta-analyses focused solely on this criterion, which is important, but does not include the opinions of clinicians and patients. A methodological strength of the Men- trum trials lies in the use of therapists’ and patients’ opinions, in addition to those of independent observers. There are two reasons why this is considered a strength. First, evidence-based medicine values the opinions of patients and clinicians highly, alongside those of independent observers. Secondly, a recent systematic review by Bagby et al. (2004) showed that the Hamilton depression scale, which is widely used in depression research, has considerable psychometrical and conceptual flaws. the definition of the categories mild, moderate and severe depression is debatable. According to clinical lore, mild and moderate depression respectively range from 12-14 to 18-20, and from 18-20 to 24-26 17-item HDRS points. This is in accordance with what is mostly found in literature. Unfor- tunately, Hamilton himself did not define cutting scores for his scale. The result is that there is no generally accepted definition of mild, moderate and severe depression. In the meta-analyses we defined mild depression as a HDRS score < 20, after Thase et al. (1997), who distinguished between ‘more severe’ depression (HDRS>20) and ‘less severe’ depression (HDRS<20) and other authors (Elkin et al., 1989; Hollon et al., 1992). We defined moderate depression as HDRS scores between 20 and 25, and severe depression as HDRS scores over 25. Of course these cut-off scores are somewhat arbitrary and create artificial boundaries, where a continuum is in fact reality. The patient sample of the mega-analyses showed a HDRS mean of 18.7 points, which is almost exactly in the middle of the cutting points range (18-20) differentiating between mild and moderate depression. Depending on the exact cut-off score, 50% (HDRS≤ 18), 61% (HDRS≤ 19), and 70% (HDRS≤ 20) of the mega-analysis sample was mildly depressed. Only 8.5% of the sample screened for the RCTs had HDRS scores over 24 points and were therefore excluded. in both meta-analyses, we based our conclusions concerning depression severity on mean baseline scores of the studies, not on individual patient data. We are aware of the fact that this is, again, a rather rough division of a spectrum. Nevertheless, our results did not seem to diverge from findings based on individual patients (Hollon et al., 1992; Hautzinger et al, 1999). a final strength of our mega-analysis is the inclusion of the treatment effect on quality of life, which is, it may be said, the ultimate goal of treatment.

200 11.8.4 Statistics

In the meta-analysis we drew a distinction between mild and moderate and between chronic and non-chronic depression. Following the recommendations of the Cochrane Collaboration (Cochrane Reviewers Handbook 4.2.2, 2003), we based the distinctions on clinical judgement, guided by the data the studies provided concerning depression severity and chronicity. According to the Cochrane Reviewers Handbook: ‘Decisions concerning what should and should not be combined are inevitably subjective, and are not amenable to statistical solutions but require discussion and clinical judgement.’ therefore, we did not divide the samples by performing ‘a priori’ statistical analyses of subgroup patterns. Of course this can lead to ‘seeing’ subgroups where one wants to find them. However, our ‘a posteriori’ tests for homogeneity showed that there was sufficient homogeneity, providing some statistical support for our divisions. We did not test for publication bias in the meta-analyses. Also we restricted ourselves to dichotomous outcome measures. in the mega-analyses there were many statistical comparisons. It could be argued that a Bonferroni correction is needed to prevent type I statistical errors. To this end a significance level of for instance 0.016 (0.5/3 as there are three sources of information) might be taken as threshold. If this correction is applied, approximately half of the significant differences disappear. This means all the more that our results must be interpreted with caution.

11.9 Clinical implications

One should keep in mind that the clinical implications suggested hereafter regard psychiatric outpatients with a mild to moderate depressive disorder. The differences in efficacy between the short-term treatment modalities discussed in this part of the thesis are not impressive, if existing at all. This means that in clinical practice the well informed patient may choose the treatment he prefers in a first step. After all, the wishes and expectations of patients play a large role in the practice of EBM. If the patient has no preference, the therapeutic anam- nesis may be of some help: has the patient been treated previously and what was the effect? If this does not lead to a decisive answer, the therapist may take into consideration that combined therapy seems to outperform psychotherapy in moderate, chronic depression (not in other types of depression) and that psychotherapy seems to outperforms pharmacotherapy at follow-up (not at treatment termination). Other possibly relevant data are that combined therapy seems the first choice treatment according to the SCL Depression subscale. As monotherapy is to be preferred over combined therapy, other things being equal, it may tentatively be concluded that SPSP is the first choice treatment in regard to the HDRS ‘Mental’ factor and the SCL Anxiety subscale. Regarding the HDRS ‘Somatic’ factor, it could be pharmacotherapy. However, therapists treat patients, not scale factors of subscales. If psychotherapy is chosen as treatment, other considerations than efficacy must decide between CBT and STPP

201 (SPSP being one of the latter). Therapist expertise is an example. After all, most patients, if given the possibility, choose a therapist, not a therapy. the reasoning above does not take into account the costs of treatment. Of course this is an important aspect of mental health care, not addressed in this thesis. Given the findings that the differences between psychotherapy (including SPSP), combined therapy and pharmacotherapy seem small, it may be argued that it is recommendable to start with a monotherapy, since the costs are lower than those of a combined therapy. severe depression has been left out of scope in this thesis. Severity seems an important factor in therapy success; the severer (HDRS>20) the depression, the more combined therapy seems to outperform monotherapy. For severe depression (HDRS>25) the clinical guidelines prescribe pharmacotherapy or combined therapy as a first step (CBO, 2005, p.56). short-term depression treatments, including SPSP, do not seem to be enough for many patients to achieve complete remission (an important condition for stable remission and relapse prevention). Therefore, a ‘stepped care’ approach may be warranted. Second step options are not investigated in this thesis. Nevertheless, some speculations about them may be allowed. If initial psychotherapy is not successful enough, the options would be: applying another ‘dose’ (in frequency and/or duration) of the same psychotherapy, changing the therapist and/or the form of psychotherapy, combining psychotherapy with pharmacotherapy or switching to pharmacotherapy. If initial pharmacotherapy is not successful enough the options would be: applying another dose of the same drug, adding another drug to the first one (‘augmentation’), switching to another drug, combining pharmacotherapy with psychotherapy or switching to psychotherapy. If initial combined therapy is not successful enough the options would be: changing the pharmacotherapy and/or the psychotherapy. The CBO guidelines (2005, p.56) hold similar recommendations. Although a stepped care approach may yield higher remission rates, there are indications that these remission rates decrease with each step. The study on Sequenced Treatment Alternatives to Relieve Depression (STAR-D, Rush et al., 2006) investigated one (n=3,671) to four (n=123) sequential treatment steps for depression. The results showed that ‘when more treatment steps are required, lower acute remission rates (….) and higher relapse rates during the follow-up phase are to be expected.’

11.10 Public health implications

According to Smit (2006) depression affects 738.000 people per year in the Netherlands. The author has calculated that this leads to a loss of 158.000 healthy life years. He contends that the economic costs for Dutch society can run up to/amount to 1.3 billion euro per year, mainly as a result of work impairment costs. The WHO predicts that in the coming decades depression will rank among the most prevalent psychiatric disorders in the world. These facts under- line the importance of depressive disorders in public health. Research on the

202 causes, risk factors and treatments of depression, and perhaps more importantly, on prevention of depression is highly relevant. Depression prevention is a relatively new research area. Promising results show that prevention is clearly a possibility (Beekman, 2004, 2006, Cuijpers et al., 2003 and 2005, Schoevers, 2006). It may well be more cost-effective than treatment once the disorder has fully developed (Smit, 2006). the relevance of our findings applies to secondary care of depression. The majority (54%) of depressed patients is treated in primary care facilities (Bijl & Ravelli, 1998). Early and effective treatment (as long as necessary to gain remission) is probably of significant importance to prevent a chronic course of the depressive disorder and to optimise social functioning. After all, there are strong indications that remission is associated with a better prognosis at follow-up (Rush et al, 2006).

11.11 Implications for future research

So far, SPSP has been investigated in five RCTs. However, these trials were performed by one and the same research group within JellinekMentrum (Amster- dam, The Netherlands). It has not been investigated by other research groups, in other treatment facilities or in patients presenting other disorders than depression. A head-to-head comparison with CBT is underway (Driessen et al., submitted). In this trial not only the relative efficacy of CBT and SPSP will be investigated. In addition, there will be a focus on finding out which therapy suits which patient better and whether, in a stepped-care approach, the addition of pharmacotherapy is helpful to psychotherapy-refractory patients. it is recommended that SPSP be investigated more broadly. It would be interesting to find out whether SPSP can be applied, e.g., in primary care. Second, it is important to direct further research towards long-term effects of SPSP (and other short-term treatment modalities), focussing, e.g., on relapse rates. Third, SPSP seems to fit perfectly in a stepped care approach but this idea is presently not vindicated. It would be interesting, e.g., to find out whether SPSP, if not successful enough, should be followed by Long-term Psychoana- lytic Supportive Psychotherapy (LPSP). Other options would be to investigate whether some kind of maintenance treatment (i.e. booster sessions) is helpful to prevent relapse. Fourth, the data presented in this thesis regard SPSP outcome research. In order to gain in depth insight, process research complementing the existing efficacy research is recommended. Much has been said but little has been elucidated regarding ‘the mighty placebo’ in biological and ‘the common factor’ in psychological therapies. There is growing evidence that the therapist’s personality, attitude and relational competence play a major part in therapeutic success. These factors may come down to just one, admittedly complex, factor of change: adequate support, i.e., adequate gratification of basic needs. Support may prove an overriding factor in psychiatry and psychotherapy. The IPR, its importance as a personality characteristic and its changeability by therapeutic means, are relatively new, clinical concepts to be tested in future research. It

203 is advisable that, as a first step, a measuring instrument for the IPR should be developed, indubitably a taxing undertaking. fifth, although there are several options for treatment, we do not yet know which patients benefit from which treatment. It is recommendable to investigate which patient characteristics predict success of treatments such as CBT, SPSP, pharmacotherapy and their combinations. An example of such research is the investigation of sub-dimensional levels of assessment instruments. It may be informative and of potential clinical use to know whether patients with high initial scores on sub-dimensions may profit more from the treatments that work better in these sub-dimensions. Another example is distinguishing between severity and chronicity. last, as preventing is always to be preferred to curing, research in the prevention and early treatment of depression is top priority.

Long-term Psychoanalytic Therapy

11.12 What study type provides the best available evidence regarding the effectiveness of long-term psychotherapy? (Q.09, chapter 8)

This paper reflected on the characteristics of Randomized Clinical Trials (RCTs) and cohort studies. For good reasons evidence-based medicine ranks RCTs higher than cohort studies. However, the acceptability, feasibility and decisive power of RCTs in Long-Term Psychotherapy (LPT) may be questioned. The only essential and unchangeable difference between RCTs and cohort studies is that the first always and the latter never include randomized control groups. Randomization is the only way to neutralize change affecting factors other than the investigated treatment. However, randomization nearly always presents practical and ethical problems for LPT research. It could be postulated that the RCT design in itself can also be divided into a kind of ‘hierarchy’. The most informative RCTs (i.e. with highly conclusive power regarding the effectiveness of a treatment) compare the treatment with no treatment, a waiting list or a placebo treatment. After this level, comparisons can be made with Treatment As Usual (TAU), treatment in a low dose, another treatment with strong efficacy evidence and another treatment that is assumed (but without strong evidence) to be effective. The ‘lower’ RCT levels are more feasible but less informative. The control conditions that are most informative are so unacceptable to patients that decisive RCTs are, in most cases, unfeasible. Cohort studies are as capable of meeting all quality criteria of intervention research as RCTs, except for randomization. The advantage of cohort studies is their greater capability to reflect day-to-day clinical practice, their limitation concerns the internal validity. Their decisive power is determined by their methodological quality and knowledge of the natural course of the investigated disorders. The knowledge of the natural course of disorders suitable for LPT treatment is limited but

204 existent. In short, the most informative RCTs present insurmountable, method- inherent feasibility problems in LPT research. Therefore, cohort studies provide the best available evidence.

11.13 What is the effectiveness of Long-term Psychoanalytic Therapy (LPaT) in clinical terms? (Q.10, chapter 9)

This paper reported a systematic review of studies, published between 1970 and 2005, relating the effects of Long-term Psychoanalytic Therapy (LPaT) on clinical variables. In order to enhance clinical homogeneity, it is restricted to individual, ambulatory treatment in adult patients with mostly ‘regular’ indications for long-term psychoanalytic therapy. In addition it distinguishes between two LPaT types: long-term psychotherapy and psychoanalysis. Before inclusion in the review, the quality of potential studies was systematically assessed, using an explicit quality criterion. Nineteen studies with a total of 2.030 patients were included, all but one cohort studies. Patients suffered from a wide range of disorders, with high rates of co-morbidity of personality and mood disorders. Effectiveness data were pooled, both for symptom reduction and personality change. Psychotherapy yielded large mean ESs (0.78 at termination, 0.94 at follow-up) and high mean overall success rates (67% at termination, 55% at follow-up) in patients presenting moderate pathology. The mean ES was larger for symptom reduction (1.05) than for personality change (0.57). In severe pathology, the results were similar, but it was only possible to calculate the mean ES for personality change (1.09). Psychoanalysis too achieved large mean ESs (0.96 at termination, 1.18 at follow-up) and high mean overall success rates (70% at termination, 54% at follow-up) in patients presenting moderate pathology. Again, the mean ES for symptom reduction was larger (1.23) than for personality change (0.83). Insufficient data regarding severe pathology was available. Success assessments by therapists and patients were in broad agreement. They thought there was more symptom reduction than personality change. Afterwards, the results of studies meeting the quality criterion were compared with those of lower quality studies, finding no large differences. In short, LPaT effectiveness has been extensively investigated, mainly by cohort studies of varying quality. The data suggest that it is an effective treatment for a large range of pathologies, with moderate to large effects that last for years after treatment termination.

11.14 What is the effectiveness of Long-term Psychoanalytic Therapy (LPaT) in costs and benefits terms? (Q.11, chapter 10)

This paper reported a systematic review of studies, published between 1970 and 2005, relating the effects of LPaT on health care use and work impairment. In addition, it presented a financial evaluation of the costs of LPT in these two areas, compared to its benefits. Changes in health care use and

205 work impairment were translated into data regarding numbers of hospital days, medical visits, percentages of patients taking medication and numbers of sick leave days. They were compared with the costs of treatment. Before inclusion in the review, the quality of potential studies was systematically assessed, using an explicit quality criterion. Seven studies (N= 861) met the inclusion criteria. The results showed that the mean cost of LPaT per patient is € 20.900. Average yearly cost reduction for days in hospital was 85% at treatment termination and 59% at follow-up (mean 2.9 years). For medical consultations costs it was 54% at termination and 56% at follow-up, for medication costs 70% and 19% respectively and for sick leave costs 61% and 67% respectively. Average yearly health care use and sick leave costs fell by € 5.584 at treatment termination (a 66% reduction). At follow-up (mean 2.9 years) these reductions still amounted to € 5.371 (64%). The break-even point for benefits and treatment costs was approximately three years after treatment termination. The reduction in work impairment appeared to be the main factor (65%-75%) in these positive results. In short, the data suggest that LPaT substantially reduces health care use and sick leave. The benefits seem to endure for years after termination. They outweigh the costs of treatment approximately three years after treatment termination.

11.15 Comments on the results

Thus far, the literature has paid insufficient attention to the limited acceptability of control conditions to patients and the subsequent problems for RCTs, especially in long-lasting therapies. Where RCTs are often impracticable, cohort studies provide the best available evidence. Neither RCTs nor cohort studies can meet all quality criteria because of inherent limitations of randomization (at the expense of external validity) and lack of randomization (at the expense of internal validity). taking an overall view on the answers to the questions 09-11, it may tentatively be stated that, according to the available evidence, long-term psychoanalytic therapies are effective treatments for a broad range of moderate to severe psychiatric pathologies. Treatment is intensive and long, but it seems that the results are stable for years after treatment termination. The costs of treatment are high, but after a relatively short time period the benefits outweigh the costs, even when only reductions in health care costs and work impairment are taken into account. the studies included in the two reviews regard psychoanalysis as well as long-term psychoanalytic psychotherapy. Comparing the results for these treatments is hazardous. Since the groups were not randomized, there is no guarantee whatsoever that the treatment is the only factor determining the result. If this serious limitation is borne in mind, it must be tentatively concluded that we found no big differences between the two treatments. Possible explanations for these findings are: 1. there were no differences; 2. the differences were not found due to lack of adequate assessment instruments; 3. the differences were

206 obscured by unknown baseline differences in the patient samples, especially in personality. Direct comparisons of short-term therapies and long-term therapies also suffer from the limitation that the groups are not randomized. It can be taken for granted that patients selected for short-term treatments differ from patients who need long-term treatment in a number of respects. Yet, yielding to the temptation to make unwarranted comparisons, we may mention that, in literature, ESs for short psychodynamic psychotherapies vary from 0.16 (Svart- berg & Stiles, 1991) to 1.10 (Crits-Christoph, 1992), 0.71 (Anderson & Lambert, 1995) and 1.46 (Leichsenring & Leibing, 2003). Lumping all psychotherapies together, Smith and Glass (1977) found an 0.68 ES, while Smith, Glass & Miller (1980) found 0.85. These outcomes were mostly based on symptom-related assessments. from a ‘historical’ (i.e. not direct) comparison between short-term therapies and LPaT, it may cautiously be concluded that long-term psychotherapy seems as effective in symptom reduction as shorter psychotherapies. The added value, if there is any, is to be found in personality change. Literature in this context shows the following. Rudolf et al. (1994 and 2004) reported that long-term psychoanalytic therapy performs better than moderate-length psychoanalytic therapy (M=60 sessions) in terms of structural changes of personality. Kopta et al. (1994) reported that improvements in personality structure take longer than symptom reduction. Perry et al. (1999) estimated that 50% of the patients with a personality disorder would recover, i.e., no longer would meet the DSM criteria for personality disorder, after 1.3 year of treatment (or 92 sessions) and 75% after 2.2 years (or 216 sessions). It may be reminded that the studies in our reviews used psychoanalytic, not DSM success criteria and that patients no longer meeting DSM personality disorder criteria still may be significantly disturbed from a psychoanalytic point of view. personality characteristics are strongly associated with quality of life, social functioning and vulnerability to relapse. In the long run, limited personality changes therefore may be clinically more important than large effects in symptom reduction.

11.16 Comments on methodology

11.16.1 Study design

The two reviews of LPaT were almost entirely based on cohort studies, the second level of evidence in the hierarchy of Evidence Based Medicine (EBM). Although high in external validity, the conclusive power of cohort studies is still a matter of debate. Many feel that they tend to overestimate the effects of treatments, as there is no correction for potential confounders. First, it cannot be ruled out that the changes found in a cohort are effectuated over time and not (just) by the treatment. Second, the comparability of treated and untreated groups, or of two treated groups, may be doubted as they lack randomization.

207 However, many studies have contradicted this point of view, finding no significant differences between the results of cohort studies and RCTs (Concato et al, 2000; Shadish et al., 2000; Benson & Hartz, 2000). Of course, the strength of cohort studies depends on their methodological quality. Thirty percent of the studies we found in the literature did not meet our quality criterion, and the quality of the studies that did still varied widely. Flaws consisted of a lack of pre-treatment data (retrospective design), an over- general description of the patient population (e.g. only stating that the sample consisted of patients treated in a given period), no diagnoses mentioned, no detailed information about the therapies (e.g. only stating that the study dealt with psychoanalysis or psychotherapy, but not mentioning the mean number of sessions or treatment duration), over-general outcome criteria (e.g. only a five-step Likert scale for ‘therapy success’), no independent, let alone blind, assessors, no information about drop-outs (e.g. only data for completers), and no confounder analyses. another flaw limits the strength of the evidence presented here. The two reviews pool data from studies applying non-manualized therapies, conducted in different patients at different moments in time and in different countries and settings. The efforts done in order to enhance the comparability of the included studies (hence the homogeneity of the review) are: restricting the search to individual therapy in adult, ambulant patients, applying a quality selection criterion and distinguishing between moderate and severe pathology and between psychoanalytic psychotherapy and psychoanalysis. Nevertheless, the included studies still were characterised by a clinical heterogeneity that was not accounted for and was not assessed statistically. In addition, comparing pre-post findings with pre-follow-up findings must be done tentatively, since they often refer to different patients. two strong points of the reviews are their external validity and their reporting follow-up data. First, they investigated treatment methods conforming to norms that hold in daily practice. Therefore they reported on the effectiveness of day-to-day clinical practice. Second, they presented data on long-term follow-up, mostly based on patients’ and incidentally on independent observers’ opinions.

11.16.2 Patients and treatments

The patient samples included in the two reviews presented a high percentage of affective and personality disorders. However, in many included studies the information about the patients was limited. Of course, all studies regarded patients ‘suitable for long-term psychoanalytic treatment’. It may be trusted that psychoanalysts thoroughly and thoughtfully consider the pathological and healthy aspects of their patients before starting treatment. The patient indeed has to be ‘sick enough to need it and healthy enough to stand it’. However, the diagnostic concepts psychoanalysts use and the indication criteria they apply differ from present day psychiatric categories. Intrapsychic conflict, developmental inhibition, ego strength, frustration tolerance and reality testing, to

208 name some examples, are beyond the scope of the DSM system. Many persons not analytically trained, understandably wonder what type of pathology LPaT treats, when expressed in DSM terms. The Psychodynamic Diagnostic Manual (PDM, Alliance of Psychoanalytical Organizations, 2006) is a recent publication that proposes conceptual and research foundations for a psychodynamically based classification system for mental health disorders. However, it is not yet systematically used or reported in research. a guideline for psychoanalysis indication, jointly framed by the American Psychiatric Association and the American Psychoanalytic Association (1985), states: Psychoanalysis is indicated for patients whose chronic DSM disorders or symptoms are based on intrapsychic conflict and developmental inhibition, for whom less intensive treatment would not work or would prolong or cause needless suffering, and who are ‘analysable’. The review of Doigde et al. (2002) examines whether this guideline is implemented in daily practice. These authors performed a survey among analysts in the U.S.A, Canada and Australia. They collected data on DSM disorders, previous treatment histories, use of medication and length of treatment, regarding 1.718 patients in regular, private practice, psychoanalytic treatment (psychoanalysis). Of these patients 75% had a DSM-III-R Axis II diagnosis (mean number of disorders 1.28, mostly Borderline, Narcissistic, Masochistic and Obsessive Compulsive) and 93% had an Axis I diagnosis (mean number of disorders 4.17, mostly mood disorder and anxiety disorder). The authors also mentioned that more than 80% of the patients had sought previous treatment, apparently to no avail. These previous treatments encompassed brief psychotherapy, supportive therapy, medication, family and marital therapies, and also long-term psychotherapy. Nevertheless it may be stated that the diagnostic indistinctiveness in the two reviews of this thesis pleads for cautious conclusions. the diagnostic indistinctiveness of the studies parallels the vague description of their treatments. There is not just one psychoanalytic therapy, there are many, and they are not rooted in just one psychoanalytic theory but in many. They are mostly not manualized and certainly not protocolized. Although this may mean freedom, variety, creativity and richness in therapeutic practice, it is a drawback for research. Certain is that the question how to select an appropriate, specific therapy for a suitable, specific patient is far from answered. The study of Giessen-Bloo et al. (2006) is a serious attempt at tackling that question. In this study the authors compared two fairly well described treatments (Schema-focused therapy and Transference Focused Therapy) for a well described patient group (diagnosed with Borderline Personality Disorder).

11.16.3 Measurements of effects

11.16.3.1 On effectiveness in clinical terms The assessment of success is quite straightforward in many of the included studies. Global assessment scales are used most frequently. These instruments trust raters’ clinical competence but their reliability is questionable. Some studies showed that they provide better information in terms of the severity

209 of the illness and treatment results (Ottoson, 1960), but they certainly do not provide the same details more specified questionnaires do. in our review, therapists and patients fairly agreed on success rates. This concurs with the results of Harty et al. (1976). However, we had too little data to compare the therapists’ and patients’ opinions with those of independent observers. Indeed, therapeutic success was predominantly rated by therapists and patients, seldom by independent observers. Some argue that therapists and patients tend to overestimate the effects of therapy. They would be too person- ally involved to judge objectively. Harty et al. (1976) performed an analysis of the findings of the Menninger Foundations’ Psychotherapy research project and found that both therapists (65%) and patients (54%) indeed rated therapy success higher than independent judges (38%). The authors conclude: ‘(….) the disparity seemed partly due to differing frames of reference, but may also have selected the participants’ need to feel that their efforts had been worthwhile’. the review on LPaT effectiveness in clinical terms (chapter 9) distinguished symptom reduction from personality change. However, this distinction was not always clearly made in the included studies. For example, Clinical Global Impression scales do not differentiate between symptom improvement and personality change. We decided to classify them as symptom reduction measures but this is a plainly debatable decision. personality change was measured in very different ways. Psychoanalytic treatments aim at personality change, but there seems to be little consensus as to what this is in the first place and how to measure its change. It cannot be denied that the impressive list of different assessment instruments limits the comparability of the outcome data. the therapy success data were mostly gathered at naturally occurring therapy termination (and at follow-up). This may entail a positive selection bias. Healthier patients enter therapy and those profiting from it are more likely to end therapy after a few years. Sicker patients (‘too ill’ is an exclusion criterion in almost all research designs) and those not getting better might not be included in the samples (for instance because the latter are still in treatment). This might bias the results towards stronger treatment effects. In addition, a limited number of patients were still in therapy at the time of ‘follow-up’. This may have a small, positive influence on the sustained benefits of LPaT.

11.16.3.2. On effectiveness in costs and benefits terms All included studies show improvement in sick leave and health care costs. However the lack of control groups evidently complicates the interpretation of the data. It cannot be excluded that improvements would have occurred ‘spontaneously’ over time in the lives of patients. In addition, positive group averages do not mean that some individual patients might not improve, or even deteriorate. Finally, as some patients in the studies used medication alongside their psychotherapy, part of its effect should probably be attributed to pharmacotherapy. Our financial evaluation too presented limitations. First, we applied present- day Dutch financial standards on the results of studies performed in other times

210 in other countries. Our financial evaluations did not therefore reflect the ‘there and then’ costs and benefits of the treatments. However, as we provided the gross numbers for changes in health care use and sick leave, local financial standards can easily be applied to the data. Second, we may have underestimated the treatment costs by not including indirect costs, e.g., time loss and travelling costs for patients and supervision or intervision costs for therapists. On the other hand, it also seems a fair assumption that well treated patients do better in their careers, and therefore earn higher incomes, pay extra taxes and consume more. We therefore most probably have underestimated LPaT benefits by not including (because of a lack of data) factors such as the detrimental effects of mental problems on work performance. Third, our calculations referred almost exclusively to employed patients. The LPaT benefits in patients living on social benefits would of course not include any reduction in sick leave, but it may be argued that at least some of them would regain the capacity to participate in social and economic intercourse. The study of Monsen et al. (1995) supports this assumption as the level of employment rises from 35% pre therapy to 50% at treatment termination and 62% at follow-up. The study of Doidge et al. (2002) shows that 15% of the patients in analysis in Aus- tralia went from unemployment to employment (4 years into treatment) and in the U.S. there was a 4.5% increase in employment (4 years into treatment). Fourth, our calculation regarding the break-even point of treatment costs and benefits was based on our findings that the benefits at treatment termination and at follow-up (mean follow-up 2.9 years) seem to be similar. Our assumption was that they are also the same in all three years between treatment termination and follow-up. This might not be the case, clearly influencing the break-even point. It may be important to take medication costs into consideration, but we did not have enough data at our disposal to do so. For lack of more precise data, we only presented the number and percentages of patients using medication.

11.16.4 Statistics

The statistical analyses we performed were not very sophisticated. The data of the included articles simply did not allow more. ESs are a reasonably robust evaluation of treatment effects, but our results are limited to a simple average calculation of the ESs reported in the included studies. Without the original data, it was not possible to calculate mean ESs based on the original means and standard deviations.

11.17 Clinical implications

LPaT demands much from patients and therapists. Over a long period of time, they must work together intensively, in a time-consuming endeavour that progresses slowly. Their endurance will be tested heavily. For some patients and some therapists that is more than they are able or willing to invest. There- fore, the indication for LPaT should be assessed expertly and carefully. How-

211 ever, for those who persevere the rewards seem worthwhile. They appear to remain stable for years after treatment termination. The costs of treatment are high, but after some years the benefits outweigh the costs. lpaT consists of psychoanalysis and long-term psychoanalytic psycho therapy. They differ in setting (vis-à-vis versus couch) and session frequency (1-2 versus 3-5 sessions a week) (see Introduction, figure 1.1.). Perhaps more importantly, the two treatment modalities can be characterized by locating them on a continuum extending from a ‘purely interpretive’ to a ‘purely supportive’ pole. It appears then that they overlap considerably: psychoanalysis may be quite supportive, psychotherapy quite interpretive (see introduction, figure 1.2.). Small wonder the differential indication is often difficult to establish. An indication basically tries to answer three questions: What is necessary? What is possible? What does the patient want? More often than not, the answers to the last two questions are decisive in difficult cases. In doubtful cases, it seems not a bad idea to begin with a less intensive therapy and, if necessary and possible, to switch to a more intensive one. After all, diagnoses and indications may change during treatment. It is not exceptional that a psychoanalysis is converted into a psychotherapy, or a psychotherapy into a psychoanalysis. lpaT is certainly no panacea. Who is suitable? The American Psychoanalytic Association describes him/her as follows: ‘The person best able to undergo psychoanalysis is someone who, no matter how incapacitated at the time, is basically, or potentially, a sturdy individual. This person may have already achieved important satisfactions—with friends, in marriage, in work, or through special interests and hobbies—but is nonetheless significantly impaired by long-standing symptoms: depression or anxiety, sexual incapacities, or physical symptoms without any demonstrable underlying physical cause. One person may be plagued by private rituals or compulsions or repetitive thoughts of which no one else is aware. Another may live a constricted life of isolation and loneliness, incapable of feeling close to anyone. A victim of childhood sexual abuse might suffer from an inability to trust others. Some people come to analysis because of repeated failures in work or in love, brought about not by chance but by self- destructive patterns of behaviour. Others need analysis because the way they are—their character—substantially limits their choices and their pleasures.’ In short, the person described above presents a chronic incapacity to work, love and play.

11.18 Public health implications

Some people meet with fateful incidents in their lives. If they dysfunction because of it, they readily will be recognized as ‘real’ clients or patients deserving treatment. Others appear unable to cope with normal or relatively normal life events, e.g., becoming an adult, growing older, experiencing difficulties at work or at home, loosing beloved ones. If they dysfunction because of it, they run the risk of being dismissed, by professionals and the public, as ‘well worried’ over ‘problems of living’ only. It may be overlooked that they

212 present ‘an enduring pattern of inner experience and behaviour that markedly deviates from the expectations of the individual’s culture, is pervasive and inflexible, has an onset in adolescence or early adulthood, is stable over time and leads to distress or impairment’, the very DSM-IV definition of personality disorder (APA, 1994, p. 629). Their problem is their inability to cope with normal problems. Their burden and that of their surroundings and the (financial) costs for society are not easy to calculate. They deserve therapy. Freud (1905) called them ‘dauernd existenzunfähigen Kranken’. For them, he said, psychoanalytic therapy is developed. It is not meant for recent but for long-standing problems.

Die psychoanalytische Therapie ist an dauernd existenzunfähigen Kranken und für solche geschaffen worden, und ihr Triumph ist es, dass sie eine befriedigende Anzahl von solchen dauernd exitenzfähig macht. (Freud, 1905)

This thesis argues that LPaT is an effective treatment for a large range of long-standing problems. Alternatives are currently scarce to non-existent, except for borderline pathology (Dialectical Behaviour Therapy, Transfer- ence Focused Therapy, Schema Focused Therapy, hospitalization programs). LPaT costs much but it is not expensive if the costs of persistent psychological problems are taken into account. In our present society, LPaT is increasingly affordable to financially affluent people only, just as it was in Freud’s times. If expertly and carefully indicated, it should be accessible to the public at large. in the clinical implications section (see 11.9), the suggestion was to begin with a less intensive LPaT in doubtful cases and, if necessary and possible, to switch to a more intensive one. This of course is a stepped care approach that holds when the doubt concerns short- or long-term therapy. The adage then could be: ‘Short if possible, long if necessary’.

11.19 Implications for future research

LPaT research can and needs to be improved and extended on several points. First, cohort studies should clearly describe their patient samples in terms of demographic characteristics and DSM-IV diagnoses, describe the treatments applied and the experience and supervision of the therapists, report dropout rates and perform secondary confounders analyses, and they should have a prospective design. Cohort studies should also include independent observers for measuring success and response. The quality checklist used in the reviews of this thesis could be used as a guideline in this regard. second, LPaT studies should explicitly state the goals of therapy in terms of both descriptive and structural pathology. The descriptive pathology system (DSM diagnoses) and the psychodynamic (structural) pathology system (e.g. presented in the Psychodynamic Diagnostic Manual (PDM, 2006) are used

213 alongside each other. We consider both systems important. However, further research is necessary to discover their differences, similarities and correlations. LPaT studies should preferably use both systems and measure treatment effects in both. third, psychoanalytic researchers should develop and use validated measurement instruments that can assess these changes. This means we should also develop a better understanding of what is called ‘personality and personality change’ and how to measure it. fourth, regarding the stepped care principle, further dose-response inves- tigations are necessary. In psychoanalytic therapies, three therapy variables play an essential role: number of sessions/frequency of sessions, duration of therapy and the setting (lying on the couch or sitting vis-à-vis). Theoretically it is possible to investigate the contribution of each of these variables, albeit a formidable research effort. However, taking into account the number of sessions only, dose-response research can further insight in what is an optimum dose for which patient. fifth, as mentioned in 11.17, the differential indications for psychoanalysis and psychoanalytic psychotherapy are difficult to make. Patients considered fit for therapy that aims at furthering insight, could be offered both psychoanalysis as well as (insight promoting) psychoanalytic psychotherapy. It would be informative to investigate how many of them are sufficiently treated with the less intensive modality. Furthermore, it should be investigated which patients profit from which treatment. sixth, epidemiological research can improve the interpretations and value of cohort studies. Knowledge of the ‘natural’ course of certain disorders and knowledge of ‘untreated’ people with long-standing problems can provide the necessary control groups for cohort studies. Of course, the problem with this research is that not many people experiencing enduring problems seek no treatment at all. Those who don’t, might differ in some personality characteristics from those who do. finally, further research should aim at expanding the evidence and insight in the financial costs and benefits of LPaT in terms of work related issues (unemployment, sick leave, working below capacities) and health care costs (medication, medical consultations, hospitalization etc.).

Final remarks

This thesis investigated the efficacy of short-term psychotherapy (among which Short-term Psychoanalytic Supportive Psychotherapy), pharmacotherapy and their combination in the treatment of psychiatric outpatients suffering from depression. The results show that these treatments do not differ much from each other, yield modest results and probably have short-lived effects. A stepped care approach therefore seems mandatory. Guidelines propose various short-term options in case of disappointing first choice treatments. Changing from psychotherapy to pharmacotherapy or vice versa and changing from monotherapy to

214 combined therapy are examples. As a rule, second steps result in additional suc- cesses, but their numbers are limited. In addition, there is little evidence that could guide clinicians in choosing between the several second step alternatives. There is some evidence that combined therapy is more effective for severer (and chronic) depressions, but this seems to recommend it as a first rather than a second step. Among all these uncertainties, the indubitable, saddening fact stands that the efficacy of each next short-term step declines steeply. Depression, certainly in secondary care, where patients are likely to suffer from severer and chronic depressions, is a difficult disorder to conquer. Is there an alternative? In my opinion, long-term psychotherapy is a serious candidate. a multitude of factors determine the occurrence, persistence, recovery and recurrence of depression. Among them, personality factors and their counter- parts, life circumstances, play a major role. The proposition that depression is a stress related disorder, is a shorthand for saying that it results from an imbalance between a person’s coping capacities (his strengths and vulnerabilities) and his living conditions (their supportive and taxing aspects). Is this only true for depression? Probably not. A larger range of so-called stress related disorders, such as anxiety disorders and somatoform disorders may well result from the same imbalance. Long-term psychotherapy aims at personality change: reducing vulnerability and increasing strength. It does not target life circumstances directly. It trusts that after a successful therapy the person will be in a better position to cope with them or to alter them. this thesis presents data regarding the effectiveness of long-term psychoanalytic therapy (LPaT). It appears that it yields long-standing results in a variety of disorders, among them depression. It also shows that many patients have sought treatment before, apparently to no avail. It seems that the added value of LPaT lies in the changes it may bring about in personality aspects. As mentioned before, LPaT is not a panacea. It can alleviate but not eradicate the damaging influence of an unresolved past on the present, not even in patients most suitable for this treatment type. Freud was explicit on this issue, but many of his followers seem to have forgotten his warning, so much so that Gaskill (1980) found it appropriate to caution once more against ‘the myth of perfectibility’. In the same vein, it may be reminded that LPaT in itself does not lead to happiness. A successful LPaT increases the chances of a person to feel happy in appropriate circumstances, and it increases the chances that life circumstances become indeed appropriate for being happy. a sensible stepped care approach in the psychological treatment of depression seems to start with short-term treatments (psychotherapy, pharmacotherapy or their combination), primarily targeting symptoms and complaints and to leave it at that if patient and clinician consider it will do. If the first short-term steps appear insufficient, long-term psychotherapy aiming at personality characteristics is recommendable. This adage ‘short-term if possible, long if necessary’ may be applicable in a large range of so-called stress related disorders.

215 References

Anderson, E.M., Lambert, M.J. (1995). Short-term dynamically orientated psychotherapy: A review and meta-analysis. Clinical Psychology Review, 15, 6: 503-514. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders. Fourth edition (DSM-IV). Washington: APA. American Psychoanalytic Association (APSAA)(1985).http://skepdic.com//psychoan. html Bagby RM, Ryder AG, Schuller DR, Marshall MB. (2004) The Hamilton Depression Rating Scale: Has the Gold Standard Become a Lead Weight? American Journal of Psychiatry; 161, 2163-2177. Beekman, A.T.F. (2004). Psychiatric epidemiology, on observation and experiment (in Dutch), Maandblad Geestelijke Volksgezondheid, 59 (7/8): 587-599. Beekman, A.T.F., Cijpers, P., van Marwijk, H.W.J., Smit, F., Schoevers, R.A., Hosman, C. (2006). Preventie van psychiatrische stoornissen. Nederlands Tijdschrift voor Geneeskunde, 150, 419-423. Benson, K, Hartz, A.J. (2000). A comparison of observational studies and randomized, controlled trials. New England Journal of Medicine, 342: 1878-86. Bijl, R.V., & Ravelli, A. (1998). Psychiatrische morbiditeit, zorggebruik en zorgbehoefte: resultaten van de Netherlands Mental Health Survey and Incidence Study (NEM- ESIS). TSG, vol. 76, 446-457. Bion, W.R. (1962). Learning from experience. Heinemann Medical Books, 1984, Lon- don. CBO Multidisciplinaire richtlijn voor depressie 2005. [email protected] Cochrane Collaboration, Cochrane Reviewers Handbook. Available at: http://cochrane. com. Accessed May, 2004. Concato, J., Shah, N., Horwitz, R.I. (2000) Randomized controlled trials, observational studies and the hierarchy of research designs. New England Journal of Medicine, 342 (25), 1887-1892. Crits-Chritoph, P. (1992) The efficacy of brief dynamic psychotherapy: a meta-analysis. American Journal of Psychiatry, 149:2: 151-158. Cuijpers, P. (2003). Examining the effects of prevention programs on the incidence of new cases of mental disorders: the lack of statistical power. American Journal of Psychiatry, 160(8): 1385-1391. Cuijpers, P., van Straten, A., Smit, F.(2005). Preventing the incidence of new cases of mental disorders: a meta-analytic review. Journal of Nervous and Mental Disease, 193 (2): 119-125. Doidge, N., Simon, B., Brauer, L., Grant, D.C., First, M., Brunshaw, J., Lancee, W.J., Stevens, A., Oldham, J.M., Mosher, P. (2002). Psychoanalytic patients in the U.S., Canada, and Australia:I. DSM-III-R disorders, indications, previous treatment, medications and length of treatment. Journal of the American Psychoanalytic Associa- tion, 50(2): 575-614. Driessen, E., Van, H.L. Schoevers, R.A., Cuijpers, P., van Aalst, G., Don, F.J., Hen- driksen, M., Kool, S., Molenaar, P., Peen, J., Dekker, J.J.M. Cognitive behavioral Therapy versus Short Psychodynamic Supportive Psychotherapy in the outpatient treatment of depression: a randomized controlled trial. Submitted for publication. Elkin, I., Shea, M.T., Watkins, J.T., Imber, S.D., Sotsky, S.M., Collins, J.F., Glass, D.R., Pilkonis, P.A., Leber, W.R., Docherty, J.P., Fiester, S.J., Parloff, M.B. (1989). National Institute of Mental Health Treatment of Depression Collaborative Research Program. Archives of General Psychiatry, 46, 971-982. Fonagy, P., Target, M. (1996). Playing with reality I: Theory of mind and the normal development of psychic reality. International Journal of Psychoanalysis, 77: 217- 234. Freud, S. (1905). Über psychotherapie, Ergänzungsband, 115.

216 Gaskill HS. (1980) The closing phase of the psychoanalytic treatment of adults and the goals of psychoanalysis ‘The myth of perfectibility’. International Journal of Psy- choanalysis, 61(1): 11-23. Giessen-Bloo, J., Dyck, van, R., Spinhoven, P., Tilburg, van, W., Dirksen, C., Asselt, van, T., Kremers, I., Nadort, M., Arntz, A. (2006). Outpatient psychotherapy for borderline personality disorder. Archives of General Psychiatry, 63: 649-658. Harty, M., & Horwitz, L. (1976). Therapeutic outcome as rated by patients, therapists, and judges. Archives of General Psychiatry, 33: 957-961. Hautzinger, M., de Jong-Meyer, R., Treiber, R.,Rudolf, G.A.E., Thien, U. (1996) Wirk- samkeit Kognitiever Verhaltenstherapie, Pharmakotherapie und deren Kombination bei nicht-endogenen, unipolaren Depressionen. Zeitschrift für Klinische psychologie, 25(2): 130-145. Hollon, S.D., DeRubeis, R.J., Evans, M.D., Wiemer, M.J., Garvey, M.J., Grove, W.M., Tuason, V.B. (1992). Cognitive therapy and Pharmacotherapy for Depression. Singly and in combination. Archives of General Psychiatry, vol. 49, 774-781. Jonghe, F. (2005). Kort en krachtig. Kortdurende Psychoanalytische Steungevende Psychotherapie. Amsterdam, Benecke N.I. Keitner, G.I., Posternak, M.A., Ryan, C.E. (2003). How many subjects with major depressive disorder meet eligibility requirements of an antidepressant efficacy trial? Jour- nal of Clinical psychiatry, 59, 598-607. Koelen, M.A.J., Dekker, J., Peen. J., Schoevers, R., de Jonghe, F., Molenaar, P. Factors associated with two-year course after combined treatment for depression: a follow- up study. Submitted for publication. Kohut, H. (1977). The restoration of the self. International Universities Press, New York. Kopta, S.M., Howard, K.I., Lowry, J.L., Beutler, L.E. (1994). Patterns of symptomatic recovery in psychotherapy. Journal of Consulting and Clinical Psychology, 62: 1009-1016. Kordy, H., Rad, von, M., Senf, W. (1988). Time and its relevance for successful psychotherapy. Psychotherapy and Psychosomatics, 49: 212-222. Leichsenring, F., Leibing, E., (2003). The effectiveness of psychodynamic therapy and cognitive behavior therapy in the treatment of personality disorders: A meta-analysis. American Journal of Psychiatry, 160: 1223-1232. Leichsenring, F. (2001). Comparative effects of short-term psychodynamic psychotherapy and cognitive-behavioral therapy in depression: a meta-analytic approach. Clinical Psychology Review, 21: 401-19. Markowitz, J.C., Svartberg, M., Swartz, H.A. (1998). Is IPT Time-Limited Psychody- namic Psychotherapy? Journal of Psychotherapy Practice Research, 7: 185- 195. Monsen, J.T., Odland, T., Faugli, A., Daae, E., Eilertsen, D.E. (1995). Personality disorders:changes and stability after intensive psychotherapy focusing on affect consciousness. Psychotherapy Research, 5(1): 33-48. Mueller, T.I., Leon, A.C., Keller, M.B., Solomon, D.A., Endicott, J., Coryell, W., War- shaw, M., Maser, J.D. (1999). Recurrence after recovery from major depressive disorder during 15 years of observational follow-up. American Journal of Psychiatry, 156: 1000-1006. NICE Clinical Guideline (2004). Management of depression in primary and secondary care. National Institute of Clinical Excellence. Ottoson, J-O. (1960). Experimental studies of the mode of action electroconvulsive treatment. Acta Psychiatrica Scandinavia, 35, suppl. 145: 69-97. Pampallona, S., Bollini, P., Tibaldi, G., Kupelnick, B., Munizza, C. (2004) Combined pharmacotherapy and psychological treatment for depression. A systematic review. Archives of General psychiatry; 61: 714-719. Perry, J.C., Banon, E., Floriana, I. (1999). Effectiveness of psychotherapy for personality disorders. American Journal of Psychiatry, 156: 1312-21.

217 PDM Taksforce. (2006), Psychodynamic Diagnostic Manual. Silver Spring, MD: Alli- ance of Psychoanalytic Organizations. Rudolf, G., Manz, R., Öri, C. (1994). Ergebnisse psychoanalytischer Therapien. Zeitschrift zur Psychosomatischen Medizin, 40: 25-40. Rudolf, G., Dilg, R., Grande, T., Jakobsen, T. Keller, W. Krawietz, B. et al. (2004). Effektiv- ität und Effizienz psychoanalytischer Langzeittherapie: Die Praxisstudie Analytische Langzeitpsychotherapie (Effectiveness and efficiency of long-term psychoanalytic psychotherapy: The practice study of long-term psychoanalytic psychotherapy). In: Gerlach, A, Springer, A., Schlösser, A. editors. Psychoanalyse des Glaubens (Psy- choanalysis of religious belief). Giessen: Psychosozial Verlag. Rush, J., Irivedi, M.H., Wisniewske, S.R., Nierenberg, A.A., Stewart, J.W., Warden, D.W., Niederehe, G., Thase, M.Th., Lavori, W., Lebowitz, B.D., McGrath, P.J., Rosen- baum, J.F., Sackheim, H.A., Kupfer, D.J., Luther, J. Fava, M. (2006). Acute and longer-term outcomes in depressed outpatients requiring one or several treatment steps: a STAR*D report. American Journal of Psychiatry, 163, 1905-1917. Schoevers, R.A., Smit, F., Deeg, D.J.H., Dekker, J., Beekman, A.T.F. (2006). Preven- tion of late-life depression in primary care; do we know where to begin? American Journal of Psychiatry, 163, 1611-1621. Shadish, W.R., Matt, G., Navarro, A., Phillips, G. (2000). The effects of psychological therapies under clinically representative conditions: a meta-analysis. Journal of Consulting and Clinical Psychology, 126: 512-529. Shafer, A. (2006). Meta-analysis of the factor structures of four depression questionnaires: Beck, CES-D, Hamilton, and Zung. Journal of Clinical Psychology, 62(1): 123-146. Smit, F. (2006). Prevention of Depression. Trimbos-instituut: Utrecht. Smith, M.L.,& Glass, G.V.(1977). Meta-analysis of psychotherapy outcome studies. American Psychologist, 32: 752-760. Smith, M.L., Glass, G.V., & Miller, T.I. (1980). The benefits of psychotherapy. Baltimore, MD: John Hopkins University Press. Spijker, J.A., de Graaf, R., Bijl, R., e.a. (2002). Duration of major depressive episodes in the general population. Results from the Netherlands Mental Health Survey and Incidence Study (NEMESIS). British Journal of Psychiatry, 181, 208-213. Svartberg, M. & Stiles, T.C. (1991). Comparative effects of short-term psychodynamic psychotherapy: a meta-analysis. Journal of Consulting and Clinical Psychology, 59: 704-714. Thase, M.E., Greenhouse, J.B., Frank, E., Reynolds III, C.F., Pilkonis, P.A., Hurley, K., Grochocinsky, V., Kupfer, D.J. (1997). Treatment of major depression with psychotherapy or psychotherapy-pharmacotherapy combinations. Archives of General Psychiatry, 54, 1009-1015. Westen, D., Morrison, K. (2001). A multidimensional meta-analysis of treatments for depression, panic, and generalized anxiety disorder: an empirical examination of the status of empirically supported therapies. Journal of Consulting and Clinical Psychology, 69: 875-899. Winnicott, D.W. (1965). The maturational processes and the facilitating environment. Hogahrth, London, 1976.

218 List of Abbreviations

CBT cognitive-Behavioural Therapy CGI-i clinical Global Impression – Improvement CGI-S clinical Global Impression – Severity DSM Diagnostic and Statistical Manual of Mental Disorders EBM evidence Based Medicine es effect Size HDrs hamilton Depression Rating Scale IPR intraPersonal Relationship, IPT interpersonal PsychoTherapy LPSP long-term Psychoanalytic Supportive Psychotherapy LPT long-term PsychoTherapy LPaT long-term PsychoAnalytic Therapy QLDS Quality of Life Depression Scale rct randomized Controlled Trial SCL-90 90 item Symptom Check List SCL-A SCL-90 Anxiety subscale SCL-D scl-90 Depression subscale SCL-S scl-90 Somatic complaints subscale snri selective Noradrenaline Reuptake Inhibitor spsp short-term Psychoanalytic (or Psychodynamic) Supportive Psycho- therapy SSRI selective Serotonine Reuptake Inhibitor stpp short-Term Psychodynamic Psychotherapy WHO World Health Organization

219

Summary

‘On the effectiveness of psychoanalytic therapy, Short if possible, long if necessary?’

Psychoanalytic therapies have been, from the very beginning, contentious: applauded by some and reviled by others. The situation today is not essentially different. The popularity among clinicians and patients alike is underscored by studies of Wiener (1994) and Luborsky et al. (1993). Yet, criticism too persists. Since we live in the era of evidence based medicine, its main argument is that psychoanalytic therapies are not ‘scientifically proven’. contrary to what many people think, psychoanalytic therapies have been investigated intensively. Short forms of psychoanalytic therapies have been investigated by means of Randomized Controlled Trials (RCTs). A large number of reviews and meta-analyses provide the integrated results (Crits-Christoph, 1992, Leichsenring, 2005, Leichsenring, 2001, Anderson and Lambert, 1995, Leichsenring, 2003, Doidge, 1997), pointing to positive evidence for the efficacy of these therapies. Because of serious practical and ethical problems, long-term psychoanalytic therapy has been investigated mainly in cohort studies. Several reviews have been carried out to summarize the evidence for long-term psychoanalytic therapy (Bachrach et al., 1991: Doidge, 1997: Fonagy, 1999; Leichsen- ring, 2005), showing that there is evidence for the effectiveness of long-term psychoanalytic therapy. however, despite the evidence, a striking discrepancy between the enduring ‘belief’ in the effectiveness of psychoanalytic therapies within the psychoanalytic community and the tenacious ‘disbelief’ outside remains. This discrepancy is what motivated the studies in this thesis, researching available existing evidence as well as participating in a project digging for more evidence in this controversial field of psychoanalytical therapy.

The aim of this thesis is: 1. to investigate the efficacy of short-term psychotherapy in the treatment of depression, relative to that of pharmacotherapy and combined therapy. 2. to present the characteristics of Short-term Psychoanalytic Supportive Psy- chotherapy (SPSP) and one of its main therapeutic agents: support. 3. to investigate the efficacy of SPSP in the treatment of depression, relative to that of pharmacotherapy and combined therapy. 4. to investigate the effectiveness of Long-term PsychoAnalytic Therapy (LPaT).

The aim of this thesis translates into eleven specific research questions, that can be formulated and (tentatively) answered as follows.

221 1. What is the efficacy of short-term psychotherapy compared to that of pharmacotherapy in ambulatory psychiatric outpatients presenting a major depressive disorder?

Chapter 2 presents a meta-analysis, based on RCTs published between 1980 and 2005, comparing the relative efficacy of time-limited (max. 6 months) psychotherapy and pharmacotherapy in the treatment of ambulatory, psychiatric outpatients diagnosed with unipolar, non-psychotic major depressive disorder. To enhance the clinical homogeneity among the studies in this meta-analysis, strict inclusion criteria regarding patient samples, diagnoses and treatment settings were applied. Ten studies were included, considering 1.233 patients (640 treated with pharmacotherapy and 593 treated with psychotherapy), suffering, on average, from moderate depression. One study considered IPT; the remaining studies applied Cognitive-Behavioural Therapy (CBT). The dropout rate in pharmacotherapy (28.4%) was significantly higher than in psychotherapy (23.6%). The remission rates (HDRS measured) of psychotherapy (38%) and pharmacotherapy (35%) at treatment termination did not differ. This latter conclusion held for both chronic and non-chronic, mild or moderate depression. Both treatments performed better in mild (HDRS<20) than in moderate (HDRS>20) depression. At follow-up, relapse in pharmacotherapy (56.6%) was larger than in psychotherapy (26.5%).

2. What is the efficacy of short-term psychotherapy compared to that of short-term psychotherapy plus pharmacotherapy in ambulatory, psychiatric outpatients presenting a major depressive disorder?

Chapter 3 presents a meta-analysis, based on RCTs published between 1980 and 2005, comparing the relative efficacy of time-limited (max. 6 months) psychotherapy and combined therapy (psychotherapy and antidepressants) in the treatment of ambulatory, psychiatric outpatients diagnosed with unipolar, non- psychotic major depressive disorder. Again, it aimed at high clinical homogeneity. Seven studies were included, considering 903 patients (459 treated with psychotherapy and 444 treated with combined therapy), suffering, on average, from moderate depression. One study considered psychoanalytic therapy; the remaining studies applied CBT. Dropout rates for combined therapy (25%) and for psychotherapy (24%) did not differ statistically. The remission rates (HDRS measured) showed that combined therapy (46%) did better than psychotherapy (34%). However, remission rates depended on severity and chronicity. In mild, non-chronic depression combined therapy (42%) and psychotherapy (37%) performed statistically equal. No data on mild, chronic depression were found. In moderate, non-chronic depression remission rates did not differ either (combined therapy 44% and psychotherapy 39%). Only in moderate, chronic depression combined therapy (48%) was superior to psychotherapy (32%).

222 3. What is Short-term Psychoanalytic Supportive Psychotherapy (SPSP)?

Chapter 4 describes the theoretical and practical characteristics of SPSP. SPSP is a face-to-face, six-month, individual psychotherapy consisting of sixteen sessions (starting with eight weekly sessions, followed by eight fortnightly sessions). SPSP’s primary goal is to cure depression. The secondary goal is to reduce patient’s vulnerability to depression. The latter is conceived as the shaping or altering of internal relationships, especially the relationship the patient has with himself (IntraPersonal Relationship, IPR). Taking into account SPSP’s restricted number of sessions, it is all too obvious that personality change will be limited. Setting, frame and contract are ingredients of the supportive approach. spsp unfolds as a discourse in which we distinguish nine levels: 1. physical and psychological complaints and symptoms, 2. life circumstances, which somehow influence or have influenced the depression, 3. problems with external, interpersonal relationships, 4. relational patterns in the patient’s life, 5. the patient’s attitude in life, 6. the persistence of past relationships in the patient’s actual life, 7. the relationship the patient maintains with herself as the consequence of identification with internal-interpersonal relationships, 8. the manifestation of problems discussed at previous levels in the relationship with the therapist, 9. transference neurosis. This last level is not attainable in the short-course of SPSP.

4. How may adequate, psychoanalytically defined support contribute to personality change?

Chapter 5 aims to propose a psychoanalytic definition of support and to elaborate on its role as an agent of change in psychoanalytic therapies (be they short or long). In the classic, psychoanalytical view, interpretation leads to insight, which in turn can result in personality change. This view is not challenged but it is contended that a second, largely neglected process also plays an important part: support may lead to personality change, which in turn may result in insight. essential to this proposition is the notion that internal relationships represent vital personality aspects. Their alteration is crucial to personality change. The longstanding interpretation versus relationship controversy is reformu- lated as an interpretation versus support debate. The proposed psychoanalytic definition of support is: the proper gratification of unmet developmental needs, as they appear in the archaic aspects of the therapeutic relationship. Then, the

223 controversy comes down to divergent views on the relative roles of interpretation and gratification. the importance of differentiating between adequate and inadequate support is stressed. Some ideas about the putative mode of action of support are presented, suggesting that it may reside in its power to evoke, in the patient, the experience of a ‘dissonance’ between two incongruous aspects of the therapeutic relationship: a ‘malignant’ an a ‘benign’ one. Both are simultaneously felt in the present, the benign one being determined by the present, the malignant one by the past. If the benign aspect of the relationship prevails over the malignant one and if it becomes internalised, internal relationships, especially ‘how I relate to Myself’ (the IntraPersonal Relationship, IPR), may change for the better.

5-7. In ambulatory psychiatric patients presenting a major depressive disorder, what is the efficacy of SPSP compared to pharmacotherapy, the efficacy of SPSP compared to SPSP plus pharmacotherapy, and the efficacy of SPSP plus pharmacotherapy compared to pharmacotherapy?

Chapter 6 assesses, by means of a mega-analysis, the relative efficacy of SPSP, pharmacotherapy and their combination in the treatment of ambulatory, psychiatric outpatients diagnosed with mild to moderate depression. The study was based on the data of three consecutive RCTs, carried out and published by the Depression Research Team of JellinekMentrum in Amsterdam. Based on the results of the original trials, three hypotheses were formulated: Combined therapy is more efficacious than pharmacotherapy and SPSP respectively, and SPSP is more efficacious than pharmacotherapy. the first hypothesis was confirmed, the other two were rejected. Accord- ing to the independent observers (HDRS), the main criterion in this study, there was no statistically significant difference in remission rates between SPSP (31%) and pharmacotherapy (24%), nor between combined therapy (40%) and SPSP (31%). However, combined therapy (40%) outperformed pharmacotherapy (24%). secondary criteria showed that, according to the therapists (CGI-S), there was no significant difference between SPSP and combined therapy, and both were superior to pharmacotherapy. According to patients’ opinions on symptom improvement (SCL-D), combined therapy was superior to SPSP, which in turn outperformed pharmacotherapy. According to patients’ evaluations of quality of life (QLDS), there was no significant difference between SPSP and combined therapy, nor between SPSP and pharmacotherapy, but combined therapy was better than pharmacotherapy. Figure 1. presents the results graphically. The symbol > indicates a statistically significant difference at the p < .05 level. NS means no statistically significant difference.

224 > (HDRS) > (CGI-S) > (SCL-D) > (QLDS) Pharmaco CT Therapy

NS (HDRS) NS (HDRS) NS (CGI-S) > (CGI-S) SPSP > (SCL-D) > (SCL-D) NS (QLDS) NS (QLDS)

Figure 1. Comparative efficacies of combined therapy (CT), Short Psychoanalytic Sup- portive Psychotherapy (SPSP) and Pharmacotherapy.

8. What are the answers to the questions 5-7 at the level of two HDRS factors and three SCL-90 subscales of: Depression, Anxiety and Somatic complaints?

Chapter 7 presents a second mega-analysis, using a similar design as the one in Chapter 6. However, the relative efficacy of SPSP, pharmacotherapy and their combination was now assessed examining HDRS factor levels and three SCL subscale levels. In this study, a factor analysis of the mega-analysis sample yielded two main factors, which were labelled ‘Mental’ and ‘Somatic’. A mega- analysis was performed at these HDRS levels and at those of three SCL subscales: Depression (SCL-D), Anxiety (SCL-A) and Somatic complaints (SCL-S). Three hypotheses were formulated: 1. spsp outperforms pharmacotherapy on the ‘Mental’ factor of the HDRS (1.a), the SCL-Depression (1.b) and the SCL-Anxiety (1.c); 2. Pharmacotherapy is better than SPSP on the HDRS ‘Somatic’ factor (2.a) and the SCL-Somatic complaints (2.b); 3. Combined therapy outperforms both monotherapies on all sub-dimensions (3.a-j). At the level of the HDRS ‘Mental’ factor, we found there were no differences between SPSP and pharmacotherapy (1.a rejected), nor between SPSP and combined therapy (3.a rejected), but the latter did outperform pharmacotherapy (3.b confirmed). At the level of the HDRS ‘Somatic’ factor, no differences were found between SPSP and pharmacotherapy (2.a rejected), nor between pharmacotherapy and combined therapy (3.c rejected), but the latter did outperform SPSP (3.d confirmed). At the level of the SCL-D, no differences were found between SPSP and pharmacotherapy (1.b rejected), and combined therapy outperformed both (3.e-f confirmed). At the level of the SCL-A, SPSP equalled combined therapy (3.g rejected) and both were superior to pharmacotherapy

225 NS (HDRS Somatic) > (HDRS Mental) > (SCL-Depression) > (SCL-Anxiety) NS (SCL-Somatic) Pharmaco CT Therapy

> (HDRS Somatic) NS (HDRS Somatic) NS (HDRS Mental) NS (HDRS Mental) > (SCL-Depression) SPSP NS (SCL-Depression) NS (SCL-Anxiety) > (SCL-Anxiety) NS (SCL-Somatic) NS (SCL-Somatic)

Figure 2. Comparative efficacies of Combined Therapy (CT), Short Psychoanalytic Supportive Psychotherapy (SPSP) and Pharmacotherapy at HDRS factor level and SCL subscale level.

(1.c and 3.h confirmed). At the level of the SCL-S, no differences were found between the two treatments (2.b and 3.i-j rejected). summarizing, it could be stated that the first hypothesis is rejected (two out of three comparisons rejected), as well as the second (two out of two comparisons rejected); the third hypothesis seems undecided so far (five of ten comparisons rejected). figure 2. presents the results graphically. The symbol > indicates a statistically significant difference at the p < .05 level. NS means no statistically significant difference.

9. What study type provides the best available evidence regarding the effectiveness of long-term psychotherapy?

Chapter 8 compares the characteristics of Randomized Clinical Trials (RCTs) and cohort studies. Evidence-based medicine ranks RCTs higher than cohort studies, and does so for good (methodological) reasons. However, the acceptability, feasibility and decisive power of RCTs in Long-Term Psycho- therapy (LPT) may be questioned. Randomization nearly always presents practical and ethical problems for LPT research. It could be postulated that the RCT design in itself can also be divided into a kind of ‘hierarchy’. The most informative RCTs (i.e. with highly conclusive power regarding the effectiveness of a treatment) compare a given treatment with no treatment, assignment to a waiting list or a placebo treatment. However, the control conditions that are most informative are so unacceptable to patients that desicive RCTs are, in most cases, unfeasible. At a lesser informative level, comparisons can be made with Treatment As Usual (TAU), treatment in a low dose, another treatment with

226 strong efficacy evidence and another treatment that is assumed (but without strong evidence) to be effective. The ‘lower’ RCT levels are more feasible but less informative. The advantage of cohort studies is their greater capability to reflect day-to-day clinical practice, their limitation concerns the internal validity. Their decisive power is determined by their methodological quality and knowledge of the natural course of the investigated disorders. Such knowledge of disorders suitable for LPT treatment is limited but existent.

10. What is the effectiveness of Long-term Psychoanalytic Therapy (LPaT) in clinical terms?

Chapter 9 presents a systematic review of studies, published between 1970 and 2005, assessing the effects of LPaT on clinical variables. In order to enhance clinical homogeneity, it was restricted to individual, ambulatory treatment in adult patients with mostly ‘regular’ indications for long-term psychoanalytic therapy. In addition it distinguished between two LPT types: long-term psychotherapy and psychoanalysis. Before inclusion in the review, the quality of potential studies was systematically assessed, using an explicit quality criterion. Nineteen out of the 27 studies found, met the quality criterion, all but one cohort studies. Patients suffered from a wide range of disorders, with high rates of co-morbidity of personality and mood disorders. Effectiveness data were pooled, both for symptom reduction and personality change. Psychotherapy yielded large mean ESs (0.78 at termination, 0.94 at follow-up) and high mean overall success rates (67% at termination, 55% at follow-up) in patients presenting moderate pathology. The mean ES was larger for symptom reduction (1.05) than for personality change (0.57). In severe pathology, the results were similar, but it was only possible to calculate the mean ES for personality change (1.09). Psychoanalysis too achieved large mean ESs (0.96 at termination, 1.18 at follow-up) and high mean overall success rates (70% at termination, 54% at follow-up) in patients presenting moderate pathology. Again, the mean ES for symptom reduction was larger (1.23) than for personality change (0.83). Insufficient data regarding severe pathology was available. Success assessments by therapists and patients were in broad agreement. They thought there was more symptom reduction than personality change. There were no differences found between the results of studies meeting the quality criterion and of those of lower quality studies.

11. What is the effectiveness of Long-term Psychoanalytic Therapy (LPaT) in cost and benefit terms?

Chapter 10 presents a systematic review of studies, published between 1970 and 2005, assessing the effects of LPaT on health care use and work impairment. In addition, it presented a financial evaluation of the costs of LPT in these two areas, compared to its benefits. Changes in health care use and work impair-

227 ment were translated into data regarding numbers of hospital days, medical visits, percentages of patients taking medication and numbers of sick leave days. They were compared with the costs of treatment. Before inclusion in the review, the quality of potential studies was systematically assessed, using an explicit quality criterion. Seven studies (N= 861) met the inclusion criteria. The results show that the mean cost of LPaT per patient is € 20.900. Average yearly cost reduction for days in hospital is 85% at treatment termination and 59% at follow-up (mean 2.9 years). For medical consultations costs it is 54% at termination and 56% at follow-up, for medication costs 70% and 19% respectively and for sick leave costs 61% and 67% respectively. Average yearly health care use and sick leave costs fall by € 5.584 at treatment termination (a 66% reduction). At follow-up (mean 2.9 years) these reductions still amount to € 5.371 (64%). The break-even point for benefits and treatment costs is approximately three years after treatment termination. The reduction in work impairment appears to be the main factor (65%-75%) in these positive results.

Chapter 11 discusses the main findings, the methodological strengths and weaknesses and merits and limitations of the studies. This thesis investigated the efficacy of short-term psychotherapy (among which Short-term Psychoanalytic Supportive Psychotherapy), pharmacotherapy and their combination in the treatment of psychiatric outpatients suffering from mild to moderate depression. The results show that these treatments do not differ much from each other, yield modest results and probably have short-lived effects. A stepped care approach therefore seems mandatory. Existing guidelines propose various short-term options in case of disappointing first choice treatments. Long- term psychotherapy could be a serious alternative in a stepped care approach. Long-term psychotherapy aims at personality change: reducing vulnerability and increasing strength. This thesis presents data regarding the effectiveness of long-term psychoanalytic therapy (LPaT). It appears that it yields long- standing results in a variety of disorders, among them depression. It also shows that many patients have sought treatment before, apparently to no avail. It seems that the added value of LPaT lies in the changes it may bring about in personality aspects. As mentioned before, LPaT is not a panacea. It can alleviate but not eradicate the damaging influence of an unresolved past on the present, not even in patients most suitable for this type of treatment. a sensible stepped care approach in the psychological treatment of depression seems to start with short-term treatments (psychotherapy, pharmacotherapy or their combination) primarily targeting symptoms and complaints and to leave it at that if patient and clinician consider it will do. If the first steps appear insufficient, long-term psychotherapy aiming at personality characteristics is recommendable. This adage ‘short-term if possible, long if necessary’ may be applicable in a large range of so-called stress related disorders.

228 Samenvatting ‘Over de effectiviteit van psychoanalytische therapie, kort als het kan, lang als het moet?’

De psychoanalytische therapie was van meet af aan controversieel: geprezen door de een, gehoond door de ander. Tot op de dag van vandaag is de situatie niet wezenlijk anders. De populariteit onder zowel clinici als patiënten wordt onderschreven door studies van Wiener (1994) en Luborsky et al (1993). De kritiek houdt echter ook aan. Aangezien wij leven in de tijd van ‘evidence based medicine’, richt de kritiek zich op het argument dat psychoanalytische thera- pieën niet “wetenschappelijk bewezen” zouden zijn. in tegenstelling tot wat veel mensen denken, is er intensief onderzoek gedaan naar psychoanalytische therapieën. Kortdurende psychoanalytische thera- pieën zijn onderzocht d.m.v. Randomized Controlled Trials (RCTs), waarvan de geïntegreerde resultaten zijn weergegeven in vele reviews en meta-analyses (Crits-Christoph, 1992, Leichsenring, 2005, Leichsenring, 2001, Anderson en Lambert, 1995, Leichsenring, 2003, Doidge, 1997). Vanwege aanzienlijke praktische en ethische problemen zijn de langdurende psychoanalytische therapieën voornamelijk onderzocht in cohort studies. Verschillende studies geven een overzicht van de bewijzen voor de effectiviteit van langdurende psychoanalytische therapie (Bachrach et al., 1991: Doidge, 1997, Fonagy, 1999; Leichsenring, 2003). Ondanks deze onderbouwing, blijft er een opvallende discrepantie bestaan tussen het vaste “geloof “ in de effectiviteit van psychoanalytische therapie in de psychoanalytische wereld en het hardnekkige “ongeloof” daarbuiten. Deze discrepantie vormt de motivatie voor de studies in dit proefschrift, waarin bestaand onderzoek wordt bestudeerd alsmede een bijdrage wordt geleverd aan een project waarin onderzoek wordt verricht naar verder bewijs op dit controversiële vlak.

Het doel van dit proefschrift is: 1. het verrichten van onderzoek naar de relatieve werkzaamheid van kortdurende psychotherapie, farmacotherapie en gecombineerde therapie in de behandeling van depressie. 2. het beschrijven van de kenmerken van Kortdurende Psychoanalytische Steungevende Psychotherapie (KPSP) en één van de belangrijkste werkzame componenten: steun. 3. het verrichten van onderzoek naar de relatieve werkzaamheid van KPSP, farmacotherapie en gecombineerde therapie in de behandeling van depressie. 4. het verrichten van onderzoek naar de effectiviteit van Langdurende Psycho- Analytische Therapie (LPaT).

229 De vier hoofddoelstellingen van dit proefschrift laten zich vertalen naar elf specifieke onderzoeksvragen, welke tentatief als volgt beantwoord kunnen worden.

Vraag 1 en 2. Wat is, bij ambulante psychiatrische patiënten met een depressieve stoornis, de werkzaamheid van kortdurende psychotherapie vergeleken met (1) farmacotherapie en (2) kortdurende psychotherapie plus farmacotherapie (‘gecombineerde therapie’)?

Hoofdstuk 2 en 3 presenteren twee meta-analyses die bovenstaande vragen trachten te beantwoorden. Zij zijn gebaseerd op RCTs die gepubliceerd zijn tussen 1980 and 2005. De eerste meta-analyse (psychotherapie versus pharmacotherapie) omvat tien onderzoeken, die samen 1.233 patiënten betreffen (640 behandeld met pharmacotherapie, 593 met psychotherapie). Bij het afsluiten van de behandeling was er geen significant verschil tussen het (met de HDRS gemeten) herstelpercentage van psychotherapie (38%) en dat van farmacotherapie (35%). Dit gold zowel voor chronische als voor niet-chronische depressies, en zowel voor lichte als voor matige depressieve stoornissen. De twee behandelingsmethoden boekten betere resultaten bij lichtere (HDRS<20) dan bij ernstigere (HDRS>20) depressies. Bij follow-up daarentegen was het recidiefpercentage van farmacotherapie (56.56%) significant groter dan dat van psychotherapie (26.51%). De tweede meta-analyse (psychotherapie versus gecombineerde therapie) omvat zeven onderzoeken, die samen 903 patiënten betreffen (459 behandeld met psychotherapie en 444 behandeld met de gecombineerde therapie). Bij het afsluiten van de behandeling was het (met de HDRS gemeten) herstelpercentage van gecombineerde therapie (46%) significant hoger dan dat van psychotherapie (34%). Dit resultaat bleek echter afhankelijk te zijn van de ernst en de duur van de depressie. Bij lichte en bij matige, niet-chronische depressie was gecombineerde therapie (resp. 42% en 44%) niet significant werkzamer dan psychotherapie (resp. 37% en 39%).We vonden geen gegevens over lichte, chronische depressie. Alleen bij matige, chronische depressie bleek gecombineerde therapie (48%) significant effectiever dan psychotherapie (32%).

Vraag 3. Wat is Kortdurende Psychoanalytische Steungevende Psy- chotherapie (KPSP)?

Hoofdstuk 4 beschrijft de theoretische en praktische kenmerken van KPSP. KPSP is een vis-à-vis, individuele psychotherapie van zes maanden, bestaande uit 16 sessies (te beginnen met acht wekelijkse, gevolgd door acht 2-wekelijkse sessies). Het primaire doel van KPSP is het genezen van depressie. De secundaire doelstelling is de patiënt minder kwetsbaar maken voor depressie. Dat laatste wordt gezien als het vormen of veranderen van interne relaties, vooral de

230 relatie die de patiënt met zichzelf heeft (IntraPersoonlijke Relatie, IPR). Gelet op het beperkte aantal sessies van KPSP zal het duidelijk zijn dat persoonlijkheidsveranderingen slecht beperkt zijn. Setting, kader, therapeutische technie- ken en afspraken zijn ingrediënten van de ondersteunende aanpak. Binnen de KPSP behandeling zijn negen gespreksniveaus te onderscheiden: 1. fysieke en psychologische klachten en symptomen, 2. levensomstandigheden die op een of andere manier de depressie beïnvloe- den of beïnvloed hebben, 3. problemen met externe, interpersoonlijke relaties, 4. relationele patronen in het leven van de patiënt, 5. de levenshouding van de patiënt, 6. het voortbestaan van relaties uit het verleden in het huidige leven van de patiënt, 7. de relatie die de patiënt met zichzelf onderhoudt als resultaat van identifi- catie met interne interpersoonlijke relaties, 8. de manifestatie van problemen die op voorgaande niveaus zijn besproken in de relatie met de therapeut, 9. de overdrachtsneurose. Dit laatste niveau is in het korte tijdsbestek van KPSP niet haalbaar.

Vraag 4. Wat is psychoanalytisch gedefiniëerde steun en hoe kan steun een therapeutisch effect hebben?

Hoofdstuk 5 omvat een voorstel voor een psychoanalytische definitie van steun en bespreekt de rol ervan als veranderingsinstrument in psychoanalytische the- rapieën (zowel kort- als langdurende). Volgens de klassieke psychoanalytische opvatting leidt interpretatie tot inzicht, wat vervolgens, kan leiden tot persoonlijkheidsverandering. Deze opvatting wordt niet weersproken, maar gesteld wordt dat een tweede, grotendeels verwaarloosd proces ook een belangrijke rol speelt: steun kan leiden tot persoonlijkheidsverandering, wat vervolgens kan leiden tot inzicht. essentieel voor deze stelling is de notie dat interne relaties wezenlijke persoonlijkheids-aspecten zijn. Persoonlijkheidsverandering betekent dus het wij- zigen van deze interne relationele patronen. De tot op dit moment geldende ‘interpretatie versus relatie’ controverse wordt hergedifiniëerd naar een ‘interpretatie versus steun’ debat. De voorgestelde psychoanalytische definitie voor steun luidt: de adequate gratificatie van onvervulde ontwikkelingsbehoeften, zoals deze zich voordoen in de primaire aspecten van de therapeutische relatie. Zo komt de controverse neer op een verschil in opvatting over de relatieve rol van interpretatie en gratificatie. Met nadruk wordt gesteld hoe belangrijk het is onderscheid te maken tussen adequate en inadequate steun. Er wordt een aantal ideeën gepresenteerd over de veronderstelde werkwijze van steun. Dit zou kunnen liggen in het vermogen om een patiënt een ”dissonantie” te doen ervaren tussen twee aspecten van

231 de therapeutische relatie: een “goed-” en een “kwaadaardig” aspect. Beiden worden tegelijkertijd ervaren in het heden, het goedaardige bepaald door het heden, het kwaadaardige door het verleden. Indien het goedaardige aspect van de relatie de overhand heeft over het kwaadaardige en indien het geïnternali- seerd wordt, dan kunnen er interne relaties, met name “de relatie met mijzelf” (de IntraPersoonlijke Relatie, IPR), ten gunste veranderen.

Vraag 5, 6 en 7. Wat is, bij ambulante psychiatrische patiënten met een depressieve stoornis, de werkzaamheid van KPSP vergeleken met farmacotherapie (vraag 5) en met gecombineerde therapie (KPSP plus farmacotherapie) (vraag 6). Wat is de werkzaamheid van gecombineerde therapie vergeleken met farmacotherapie (vraag 7)?

Hoofdstuk 6 beschrijft een mega-analyse waarin de bovenstaande vragen beantwoord worden. Deze is gebaseerd op drie achtereenvolgende RCTs, die samen 313 patiënten betreffen. De RCTs zijn uitgevoerd en gepubliceerd door het Depressie Onderzoeksteam van JellinekMentrum, een GGZ instelling te Amsterdam. Bij het afsluiten van de behandeling was er geen significant verschil tussen het (met de HDRS gemeten) herstelpercentage van KPSP (31%) en dat van farmacotherapie (24%) (vraag 5). Er werd geen significant verschil tussen gecombineerde therapie (40%) en KPSP (31%) gevonden (vraag 6). Gecombi- neerde therapie (40%) bleek significant effectiever dan farmacotherapie (24%) (vraag 7). Deze resultaten worden genuanceerd door uitslagen op grond van secundaire criteria. De therapeuten (CGI-S) zagen geen significant verschil in werkzaamheid tussen KPSP en gecombineerde therapie. Zij vonden wél dat deze twee behandelingsmethoden effectiever waren dan farmacotherapie. De patiënten (SCL-D) vonden dat gecombineerde therapie werkzamer was dan farmacotherapie en KPSP, dat op zijn beurt werkzamer was dan farmacotherapie. Wat betreft de kwaliteit van hun leven (QLDS) vonden de patiënten geen significant verschil tussen KPSP en gecombineerde therapie, noch tussen KPSP en farmacotherapie. Zij vonden wél dat gecombineerde therapie in dit opzicht effectiever was dan farmacotherapie. De resultaten worden hierna schematisch weergegeven (figuur 1). Het teken > verwijst naar een statistisch significant verschil (p < .05). NS betekent statistisch Niet Significant.

Vraag 8. Wat zijn de antwoorden op de vragen 5-7 op het niveau van twee HDRS factoren en van drie SCL-90 subschalen?

Hoofdstuk 7 bevat een tweede mega-analyse die deze vragen beantwoordt. De onderzoeksopzet is vergelijkbaar met die van hoofdstuk 6. De relatieve werkzaamheid van KPSP, farmacotherapie en gecombineerde therapie werd nu echter onderzocht op het niveau van twee HDRS factoren, de factor ‘Mental’ (HDRS-M) en de factor ‘Somatic’ (HDRS-S), en op dat van drie SCL subschalen, Depressie (SCL-D), Angst (SCL-A) en Somatische klachten (SCL-S).

232 > (HDRS) > (CGI-S) > (SCL-D) > (QLDS) Pharmaco CT Therapy

NS (HDRS) NS (HDRS) NS (CGI-S) > (CGI-S) SPSP > (SCL-D) > (SCL-D) NS (QLDS) NS (QLDS)

Figuur 1: Relatieve effectiviteit van gecombineerde therapie (CT), kortdurende Psy- choanalystische steungevende psychotherapie (SPSP) en farmacvotherapie.

De HDRS-M factor bracht geen significant verschil aan het licht tussen KPSP en farmacotherapie, noch tussen KPSP en gecombineerde therapie. Gecombi- neerde therapie was wel werkzamer dan farmacotherapie. De HDRS-S factor bracht geen significant verschil aan het licht tussen KPSP and farmacotherapie, noch tussen farmacotherapie en gecombineerde therapie. Gecombineerde therapie was wel werkzamer dan KPSP. De SCL-D vond geen significant verschil tussen KPSP en farmacotherapie maar toonde aan dat gecombineerde therapie werkzamer was dan de twee andere behandelingsmethoden. De SCL-A vond geen significant verschil tussen KPSP en gecombineerde therapie maar toonde aan dat deze twee behandelingsmethoden allebei werkzamer waren dan farmacotherapie. De SCL-S bracht geen significante verschillen aan het licht tussen de drie behandelingsmethoden. De resultaten worden hierna schematisch weergegeven (figuur 2). Het teken > verwijst naar een statistisch significant verschil (p < .05). NS betekent statistisch Niet Significant.

Vraag 9. Welk type onderzoek is het meest geschikt voor empirisch onderzoek naar de werkzaamheid van langdurige psychotherapie?

Hoofdstuk 8 vergelijkt de kenmerken van Randomized Clinical Trials (RCTs) en cohort studies. In de hiërarchie van ‘evidence based medicine’ staan de RCTs hoger dan cohort studies en daar zijn goede redenen voor. Echter, bij Langdu- rende Psychotherapie (LPT) is het de vraag in hoeverre RCTs acceptabel en uitvoerbaar zijn. Randomisatie levert bij LPT-onderzoek vrijwel altijd praktische en ethische problemen op. Men zou kunnen stellen dat ook binnen RCTs sprake is van een hiërarchie. De meest informatieve RCTs (d.w.z. met de groot- ste zeggingskracht betreffende de effectiviteit van een behandeling) vergelijken

233 NS (HDRS Somatic) > (HDRS Mental) > (SCL-Depression) > (SCL-Anxiety) NS (SCL-Somatic) Pharmaco CT Therapy

> (HDRS Somatic) NS (HDRS Somatic) NS (HDRS Mental) NS (HDRS Mental) > (SCL-Depression) SPSP NS (SCL-Depression) NS (SCL-Anxiety) > (SCL-Anxiety) NS (SCL-Somatic) NS (SCL-Somatic)

Figuur 2: Relatieve effectiviteit van gecombineerde therapie (CT), kortdurende Psy- choanalystische steungevende psychotherapie (SPSP) en farmacvotherapie. de behandeling met het niet geven van een behandeling, het plaatsen op een wachtlijst of het geven van een placebobehandeling. Na dit niveau, kan men vergelijkingen maken met Treatment as Usual (TAU), behandeling met een lage dosering, een andere behandeling met een duidelijk aangetoonde werkzaamheid en een andere behandeling die wordt geacht effectief te zijn (zonder duidelijk bewijs daarvoor). De meest informatieve controlecondities zijn zo onacceptabel voor patiënten dat beslissende RCTs in de meeste gevallen onuitvoerbaar zijn. “Lagere” niveaus van RCTs zijn beter te realiseren, maar zijn minder informatief. Het voordeel van cohort studies is hun grotere vermogen om de dagelijkse, klinische praktijk weer te geven; hun beperking ligt in de interne validiteit. Hun zeggingskracht wordt bepaald door de methodologische kwaliteit en de kennis van het natuurlijke beloop van de onderzochte stoornissen. Kortom, bij LPT- onderzoek kennen de meest informatieve RCTs onoverkomelijke uitvoerbaar- heidsproblemen, die inherent aan fer methode zijn. Derhalve wordt het best beschikbare bewijs geleverd door cohort studies.

Vraag 10. Wat is de werkzaamheid van Langdurige PsychoAnalyti- sche Therapie (LPaT) in klinische termen?

Hoofdstuk 9 omvat een systematisch overzicht van studies, gepubliceerd tussen 1970 en 2005, die de effectiviteit van LPaT beoordelen. Om de klinische homo- geniteit te bevorderen, is alleen gekeken naar individuele, ambulante behandelingen van volwassen patiënten met hoofdzakelijk “reguliere” indicaties voor langdurige psychoanalytische therapie. Bovendien is er onderscheid gemaakt tussen twee LPaT-types: langdurende psychotherapie en psychoanalyse. De kwaliteit van potentiële studies is systematisch beoordeeld met een expliciet kwaliteitscriterium. Van de 27 gevonden studies voldeden er 19 aan het kwa-

234 liteitscriterium, op één na allemaal cohort studies. De patiënten vertoonden een breed scala aan stoornissen, met hoge comorbiditeit van persoonlijkheidsstoornissen en toestandsbeelden. Psychotherapie leverde hoge gemiddelde ‘effect sizes’ (ESs) op (0,78 bij beëindiging, 0,94 bij follow-up) en hoge, gemiddelde, algemene succespercentages (67% bij beëindiging, 55% bij follow-up), bij patiënten met matige pathologie. De gemiddelde ES lag voor symptoomreductie (1,05) hoger dan voor persoonlijkheidsverandering (0,57). Ernstige pathologie leverde vergelijkbare resultaten op, maar het was alleen mogelijk de gemiddelde ES voor persoonlijkheidsverandering (1,09) uit te rekenen. Ook psychoanalyse haalde hoge gemiddelde ESs (0,96 bij beëindiging, 1,18 bij follow-up) en hoge, gemiddelde, algemene succespercentages (70% bij beëindiging, 54% bij follow-up) bij patiënten met matige pathologie. Wederom lag de gemiddelde ES voor symptoomreductie (1,23) hoger dan die voor persoonlijkheidsverandering (0,83). Voor ernstige pathologie waren er onvoldoende gegevens beschik- baar. Succesbeoordelingen door therapeuten en patiënten kwamen grotendeels overeen. Men vond dat er meer symptoomreductie was dan persoonlijkheidsverandering. Er zijn geen verschillen gevonden tussen de resultaten van de studies die aan het kwaliteitscriterium voldeden en die van studies met een lagere kwaliteit.

Vraag 11. Wat is de werkzaamheid van Langdurige PsychoAnalyti- sche Therapy (LPaT) in termen van kosten en baten?

Hoofdstuk 10 bevat een systematisch literatuuroverzicht van de tussen 1970 en 2005 gepubliceerde onderzoeken naar het effect van LPaT op kosten van gezondheidszorg en arbeidsongeschiktheid. Verandering in gezondheidszorg- kosten en ziekteverzuim werd gedefinieerd als aantal ziekenhuisdagen, medische consulten, percentage patiënten dat medicatie gebruikt en het aantal ziek- teverlofdagen. Zeven studies (N=861) voldeden aan de toelatingscriteria. Uit de resultaten blijkt dat LPaT per patiënt gemiddeld €20.900 kostte. De gemiddelde, jaarlijkse kostenvermindering in aantal ziekenhuisdagen was 85% bij beëindiging van de behandeling en 59% bij de follow-up (gemiddeld 2,9 jaar). Voor medische consulten lag dit op 54% bij beëindiging en 56% bij de follow-up, voor medicatiekosten op respectievelijk 70% en 19% en voor ziekteverlofkosten respectievelijk op 61% en 67%. Het gemiddelde zorgverbruik per jaar liep bij beëindiging terug met €5.584 (een vermindering van 66%). Bij de follow-up (gemiddeld 2,9 jaar) bedroeg deze vermindering nog altijd €5.371 (64%). Het ‘break-even-point’ van de behandelingskosten en de besparingen lag op ongeveer 3 jaar na beëindiging van de behandeling. Kortom, de gegevens duiden erop dat LPaT het zorgverbruik en ziekteverlof aanzienlijk terugdringt. De voor- delen lijken tot jaren na beëindiging te beklijven. Vanaf 3 jaar na beëindiging zijn de baten groter dan de kosten.

Tot slot worden in hoofdstuk 11 de resultaten, de methodologische sterkten en zwakten en de implicaties van het onderzoek besproken. Het hoofdstuk is

235 ingedeeld in twee delen: kortdurende (psychoanalytische) psychotherapie en langdurende psychoanalytische therapie.

Kortdurende (psychoanalytische) psychotherapie De twee meta-analyses betroffen kortdurende psychotherapie (ongeveer 6 maanden) bij depressie. Zij bleken achteraf in hoofdzaak over Cognitieve GedragsTherapie (CGT) te gaan. De resultaten kunnen dan ook gevoeglijk aan CGT worden toegeschreven. Over de relatieve werkzaamheid van CGT en farmacotherapie kan worden gezegd dat er bij afsluiting van de behandeling geen significant verschil is gevonden, maar dat na het staken van de behandeling (bij follow-up) de met CGT behandelde patiënten minder vaak dan de met farmacotherapie behandelde patiënten een recidief vertonen. Over de relatieve werkzaamheid van CGT en gecombineerde therapie kan worden gezegd dat ook hier geen significant verschil is gevonden, behalve waar het om matige, chronische depressies gaat. In dat laatste geval is gecombineerde therapie werkzamer. Al met al lijken de verschillen, zo ze al bestaan, klein. Over de relatieve werkzaamheid van gecombineerde therapie en farmacotherapie is in het kader van dit proefschrift geen onderzoek gedaan. Echter, andere onderzoekers, onder wie Pampallona et al. (2005) melden op grond van meta-analyses dat gecombineerde therapie werkzamer is dan farmacotherapie. De twee mega-analyses leiden tot de voorzichtige conclusie dat, in de behandeling van ambulante patiënten met een lichte tot matige depressieve stoornis (‘major depressive disorder’), de werkzaamheid van KPSP vergelijkbaar is met die van farmacotherapie en met die van gecombineerde therapie. Een tweede conclusie is dat gecombineerde therapie werkzamer is dan farmacotherapie. De resultaten verschillen wanneer rekening wordt gehouden met HDRS factoren en met SCL subschalen, maar in geen van de vergelijkingen blijkt dat farmacotherapie het beter doet dan KPSP of gecombineerde therapie. farmacotherapie, CGT en hun combinatie zijn algemeen aanvaarde behandelingen van depressie. KPSP voegt hier een alternatief aan toe met totaal verschillende eigenschappen. Een rechtstreekse vergelijking van KPSP en CGT is nog niet gepubliceerd. Voorlopig moet men het dus stellen met een ‘historische’ vergelijking. Deze leert dat de herstelpercentages niet wezenlijk verschillen (KPSP 31%, CGT 34%-38%). De twee methoden zijn even effectief als farmacotherapie en als gecombineerde therapie behalve bij matige, chronische depressie. In combinatie met farmacotherapie zijn ze beide werkzamer dan farmacotherapie. Op grond hiervan valt tentatief te concluderen dat de werkzaamheid van KPSP en die van CGT vergelijkbaar is wat betreft symptoomreductie op korte termijn. Deze opvatting wordt gesteund door de resultaten van de meta- analyse van Leichsenring (2001). het hiervoor gaande moge waar zijn, maar is er wel behoefte aan weer een nieuwe kortdurende behandelmethode voor depressie? Hierbij dient men te overwegen dat “patiënten met een ernstige depressieve stoornis” nog altijd een behoorlijk heterogene categorie vormen. Daarom ligt het niet voor de hand dat slechts enkele methodes volstaan om alle patiënten adequaat te behandelen.

236 Integendeel, het lijkt plausibel dat sommigen gebaat zijn bij farmacotherapie, anderen bij CGT en weer anderen bij KPSP (of combinaties ervan). Aan kortdurende behandelmethoden voor depressie is geen gebrek maar het kan niet worden ontkend dat hun werkzaamheid op korte termijn beperkt is. De herstelpercentages blijven doorgaans ruim onder de vijftig procent. Daar komt bij dat een eenmaal bereikt herstel lang niet altijd aanhoudt na het staken van de behandeling. Hoewel dit onderwerp in dit proefschrift niet uitgebreid is onderzocht, zijn terugvalpercentages van 27%-57% gevonden in de meta-analyse van dit proefschrift waarin psychotherapie met farmacotherapie wordt vergeleken. In hun meta-analyse van de behandeling van depressie vinden Westen en Morrison (2001) dat, van de herstelde patiënten, slechts 37% dat nog steeds was bij de follow-up. Kortom, er is alle reden om te vrezen dat het succes van kortdurende therapie voor velen kortstondig is. Voor hen lijkt een kortdurende therapie gewoon niet genoeg.

Langdurige psychoanalytische therapie Op grond van dit onderzoek mag worden geconcludeerd dat er voldoende empi- rische evidentie bestaat om te stellen dat Langdurige PsychoAnalytische The- rapie (LPaT) een effectieve behandeling is voor een breed scala aan matige tot ernstige psychiatrische pathologieën. Behandelingen zijn intensief en duren lang maar het lijkt erop dat de resultaten tot jaren na beëindiging van de behandeling stabiel blijven. De behandelkosten zijn hoog maar na een relatief kort tijdsbestek worden die kosten gecompenseerd door lagere zorgen arbeidson- geschiktheidskosten. is er een verschil in werkzaamheid tussen de twee vormen van LPaT: psychoanalyse en langdurige psychoanalytische psychotherapie? Uit ons onderzoek komen geen grote verschillen naar voren. Echter, een dergelijke vergelijking is, strikt genomen, een methodologische ‘zonde’: het vergelijken van onverge- lijkbare groepen. Aangezien de groepen niet gerandomiseerd zijn, is er geen enkele garantie dat de behandeling de enige factor is die het resultaat bepaalt. Mogelijke verklaringen voor het ontbreken van verschillen zijn: (1) er zijn geen (grote) verschillen, (2) de relevante aspecten, met name de persoonlijkheidskenmerken, zijn niet onderzocht, (3) de beoordelingsinstrumenten waren niet adequaat, of (4) de verschillen worden verdoezeld door onbekende baselinever- schillen in de patiëntenpopulaties, wederom persoonlijkheidskenmerken in het bijzonder. Een RCT zou hierover uitsluitsel kunnen geven, een moeilijke maar geen onmogelijke opgave. is LPaT werkzamer dan kortdurende (psychoanalytische) therapie? Het lijkt niet zo. De literatuur vermeldt voor kortdurende psychoanalytische therapieën uiteenlopende effectgrootten: 0.16 (Svartberg & Stiles, 1991), 0.71 (Anderson & Lambert, 1995), 1.10 (Crits-Christoph, 1992) en 1.46 (Leichsenring & Leibing, 2003). Als gemiddelde effectgrootten van kortdurende therapieën in het algemeen vermelden Smith and Glass (1977) 0.68 en Smith, Glass & Miller 0.85. LPaT lijkt deze resultaten niet te overtreffen. De twee behandelmethoden zijn echter niet goed met elkaar te vergelijken. Men kan ervan uitgaan dat patiënten die geselecteerd worden voor een kortdu-

237 rende behandeling in belangrijke opzichten verschillen van patiënten die een langdurige behandeling nodig hebben. Bovendien zijn de resultaten grotendeels gebaseerd op symptoomgerelateerde beoordelingen. Wie zich toch aan een ‘historische’ (niet een rechtstreekse) vergelijking waagt, zal concluderen dat kortdurende psychotherapieën qua symptoombestrijding net zo effectief zijn als langdurende psychotherapie. De toegevoegde waarde, zo die er is, moet men zoeken in persoonlijkheidsverandering. Dat LPaT op dit gebied wat te bieden heeft is aannemelijk. Rudolf et al. (1994 en 2004) melden dat, waar het gaat om structurele persoonlijkheidsveranderingen, langdurende psychoanalytische therapie beter presteert dan psychoanalytische therapie van middellange duur (M= 60 sessies). Kopta et al. (1994) stellen dat verbeteringen in de persoonlijk- heidsstructuur meer tijd vergen dan symptoombestrijding. Perry et al. (1999) schatten dat 50% van de patiënten met een persoonlijkheidsstoornis zouden herstellen, d.w.z. niet meer voldoen aan de DSM-criteria voor persoonlijkheidsstoornissen, na een behandeling van 1,3 jaar (ofwel 92 sessies) en 75% na 2,2 jaar (ofwel 216 sessies). Let wel dat de studies in ons verslag psychoanalytische en niet DSM-criteria voor succes hanteren en dat patiënten die niet langer aan de DSM-criteria voor persoonlijkheidsstoornissen voldoen nog altijd aanzienlijk gestoord kunnen zijn vanuit psychoanalytisch oogpunt. psychoanalytische therapie, ook LPaT, is minder succesvol in het veranderen van persoonlijkheidskenmerken dan in het bestrijden van symptomen. Dat zal niemand verbazen. Symptomen zijn soms hardnekkig, persoonlijkheidskenmerken zijn het altijd. Daar staat tegenover dat op de lange termijn een beperkte persoonlijkheidsverandering klinisch belangrijker kan zijn dan een forse symptoomreductie. Persoonlijkheidskenmerken bepalen immers in hoge mate de kwaliteit van leven, het sociaal functioneren en de kwetsbaarheid voor recidief.

Eindopmerkingen Dit proefschrift heeft de effectiviteit onderzocht van kortdurende psychotherapie (waaronder Kortdurende Psychoanalytische Steungevende Psychotherapie), farmacotherapie en de combinatie van deze twee bij psychiatrische patiënten met een depressieve stoornis. De resultaten laten zien dat de werkzaamheid van deze drie behandelmethoden beperkt is, elkaar niet veel ontloopt en waarschijnlijk van korte duur is. Een stapsgewijze benadering lijkt daarom noodza- kelijk. Richtlijnen beschrijven verschillende mogelijkheden in het geval dat een eerste behandeling niet aanslaat. Voorbeelden zijn overstappen van psychotherapie naar farmacotherapie of vice versa en veranderen van monotherapie naar gecombineerde therapie. Doorgaans leveren deze stappen verdere successen op, maar die zijn beperkt in aantal. Bovendien is er weinig bewijs waar de clinicus op kan varen bij de keuze tussen de verschillende, tweedekeus alternatieven. Er lijkt wel enige onderbouwing dat gecombineerde therapie werkzamer is bij ernstiger, chronische depressie. In dat geval is het echter waarschijnlijk een eerste keus behandeling. temidden van al deze onzekerheden staat het onbetwistbare en treurige feit vast dat de werkzaamheid van elke volgende kortdurende stap hard terugloopt.

238 Depressie is een lastig te bedwingen stoornis, zeker in de tweedelijns zorg, waar patiënten veelal aan ernstigere en chronische depressies lijden. Is er nog een alternatief? Wij menen dat langdurende psychotherapie een serieuze kandidaat is. een veelvoud aan factoren bepaalt het ontstaan, voortduren, herstel en terugval van een depressie. Daaronder bevinden zich persoonlijkheidsfactoren en hun tegenhangers, de levensomstandigheden, welke een belangrijke rol spe- len. De stelling dat depressie een stressgerelateerde stoornis is, betekent niets anders dan stellen dat depressie het gevolg is van een verstoorde balans tussen de weerbaarheid (sterke en zwakke punten) en de levensomstandigheden (steunende en belastende aspecten) van een persoon. Geldt dit alleen voor depressie? Waarschijnlijk niet. Een grotere verscheidenheid aan zogeheten stressgerelateerde stoornissen, zoals angst- en somatoforme stoornissen, zou wel eens het resultaat kunnen zijn van dezelfde scheve balans. Langdurende psychotherapie is gericht op persoonlijkheidsverandering: mensen sterker maken en minder kwetsbaar. Levensomstandigheden worden niet direct aangepakt. Men vertrouwt erop dat na een succesvolle therapie de persoon in kwestie beter in staat is ermee om te gaan of ze te veranderen. langdurende Psychoanalytische Therapie (LPaT) blijkt bij verscheidene stoornissen, waaronder depressie, resultaten op te leveren die lang standhou- den, zelfs bij patiënten die eerder, klaarblijkelijk zonder succes, zich hebben laten behandelen. Het lijkt erop dat de toegevoegde waarde van LPaT ligt in de mogelijke veranderingen die het kan bewerkstelligen in persoonlijkheidskenmerken. LPaT is echter geen panacee. Het kan de schadelijke invloed van een onverwerkt verleden op het heden verminderen, maar niet uitwissen, zelfs niet bij de meest geschikte patiënten voor dit type behandeling. In lijn daar- mee herinneren wij eraan dat LPaT op zich niet leidt tot geluk. Een succesvolle LPaT verhoogt iemands kansen zich onder de juiste omstandigheden gelukkig te voelen en het verhoogt de kans dat de er sprake is van dergelijke levensomstandigheden. Kortom, een verstandige, stapsgewijze aanpak bij de behandeling van depressie lijkt aan te vangen met een kortdurende behandeling (psychotherapie, farmacotherapie of de combinatie), voornamelijk gericht op klachten en symptomen, en laat het daarbij indien patiënt en behandelaar dat afdoende vinden. Blijken de eerste kortdurende stappen ontoereikend te zijn, dan is langdurende psychotherapie gericht op verandering van persoonlijkheidskenmerken aan te bevelen. Het adagium ‘Kort als het kan, lang als het moet’ is wellicht ook bij andere stress gerelateerde stoornissen van toepassing.

239

Curriculum Vitae

Saskia de Maat werd geboren op 2 maart 1963 in Arnhem. In 1981 behaalde zij het diploma Atheneum-A aan het Stedelijk Lyceum te Zutphen. Zij studeerde van 1981 tot 1988 Nederlandse Taal- en Letterkunde aan de Universiteit van Utrecht, met als afstudeerrichting Communicatie (cum laude). van 1989 tot 2000 werkte zij bij Akzo Nobel in verschillende functies. Zij begon op de afdeling Corporate Communications in Arnhem, waar zij verant- woordelijk was voor de arbeidsmarktcommunicatie richting academische doel- groepen. Na de opleidingen Nima A, B en C vervolgde zij haar loopbaan in mar- keting bij de business unit Decorative Coatings. Zij was hier achtereenvolgens junior product manager, product manager en senior brand manager. van 1999 tot 2003 studeerde zij Psychologie, eerst aan de Universiteit van Leiden, later aan de Vrije Universiteit Amsterdam, met als afstudeerrichting Klinische Psychologie (cum laude). in 2002 begon zij bij JellinekMentrum als stagiaire in het kader van de studie Psychologie en als onderzoeksassistent voor het Depressieonderzoek in Amsterdam Noord. Van 2005 tot 2007 volgde zij, eveneens bij JellinekMen- trum, de opleiding tot GZ-psycholoog en in september 2007 begon zij aan de opleiding tot Psychotherapeut. In 2004 startte zij haar promotieonderzoek, onder begeleiding van Jack Dekker, Robert Schoevers en Frans de Jonghe. Het onderzoek betreft de effectiviteit van psychoanalytische therapieën, waaronder Kortdurende Psychoanalytische Steungevende Psychotherapie in het voor- noemde Depressieonderzoek. saskia is getrouwd met Marcel de Rooij.

241

List of publications

Publications of this thesis

Maat, S., de, Dekker, J., Schoevers, R., Jonghe, F., de. (2006). Relative efficacy of psychotherapy and pharmacotherapy in the treatment of depression: a meta-analysis. Psychotherapy Research, 16(5): 562-572. Maat, S., de, Dekker, J., Schoevers, R., Jonghe, F., de. (2007). The effectiveness of long- term psychotherapy: methodological issues. Psychotherapy Research, 17(1): 59-65. Maat, S., de, Dekker, J., Schoevers, R., Jonghe, F., de. (2007). Relative efficacy of psychotherapy and combined therapy in the treatment of depression: a meta-analysis. European Psychiatry, 22, 108. Maat, S., de, Dekker, J., Schoevers, R., Aalst, A., van, Gijsbers-van Wijk, C., Hendriksen, M., Kool, S., Peen, J., Van, H., Jonghe, F., de. (2007). Short Psychodynamic Sup- portive Psychotherapy, Antidepressants and their Combination in the Treatment of Major Depression: A mega-analysis based on three randomized clinical trials. Depression and Anxiety (in press). Maat, S., de, Philipszoon, F., Dekker, J., Jonghe, F., de. (2007). Costs and benefits of long-term psychoanalytic therapy: changes in health care use and work impairment. Harvard Review of Psychiatry (in press). Maat, S., de, Schoevers, R., Jonghe, F., de, Aalst, A., van, Gijsbers-van Wijk, C., Hen- driksen, M., Kool, S., Peen, J., Van, H., R., Dekker, J. Comparison of Short Psychody- namic Supportive Psychotherapy, Pharmacotherapy and their Combination on sub dimensions of the HDRS and sub scales of the SCL-90: a mega-analysis of three randomized controlled trials regarding depressed patients. Submitted. Maat, S., de, Jonghe, F., de, Schoevers, R., Dekker, J. The effectiveness of long-term psychoanalytic therapy: a systematic review of empirical studies. Submitted. Jonghe, F., de, Maat, S., de, Van, H., Hendriksen, M., Kool, S., Aalst, G., van, Schoevers, R., Dekker, J. Short-term Psychoanalytic Supportive Psychotherapy for depressed patients. Submitted. Jonghe, F., de, Maat, S., de, Van, H., Hendriksen, M., Kool, S., Aalst, G., van, Schoevers, R., Dekker, J. Support and personality change: A psychoanalytic view. Submitted.

Other publications

Jonghe, F., & de, Maat, S., de (2002). De andere waarheid, een interpretatie vanuit psy- chodynamisch perspectief van de roman Gloed van Sandor Marai, Tijdschrift voor Psychotherapie, (28) 4. Maat, S., & de, Jonghe, F., de (2003). De smachtende God, een psychodynamische interpretatie van een gedicht van Willem Kloos, Tijdschrift voor Psychotherapie, (29). Jonghe, F., & de, Maat, S., de (2003). The end of the affair, De Taal van het gevoel, VU Medisch Centrum Amsterdam. Maat, S., de, & Jonghe, F., de (2004). Een gedoemde liefde, een interpretatie van The end of the affair, Tijdschrift voor Pyschiatrie, 46 (6), 379-387. Kool, S., Schoevers, R.A., de Maat, S., Van, H.L., Molenaar, P.J., Vink, A., Dekker, J. (2005). Efficacy of pharmacotherapy in depressed patients with and without personality disorders: a meta-analysis. Journal of Affective Disorders, 88(3): 269- 278.

243

Dankwoord

Zoals de psychoanalyse elke keer zoekt naar het verhaal achter het verhaal, zo schuilen in mijn werk vele lagen van mensen die mij kennis, inspiratie, hulp, steun en plezier hebben gegeven bij het werken aan dit onderzoek.

Als eerste wil ik mijn promotor, Jack Dekker, en mijn copromotoren, Robert Schoevers en Frans de Jonghe, bedanken.

Jack, je hebt mij in 2002 aangenomen als stagiaire psychologie en onderzoeksassistent voor JellinekMentrum Noord en vanaf dat moment heb je mij jouw vertrouwen en steun gegeven. Mede dankzij jouw inzet, kon ik een opleidingsplek bij JellinekMentrum realiseren, waarbij ik deels de GZ-opleiding kon doen en deels kon werken aan een promotieonderzoek. “Wat in een goed vat zit, verzuurt niet”, zijn de gevleugelde woorden waarmee je me altijd opbeurde als er weer eens een artikel was afgewezen. En ik dacht dan wel eens ‘maar wat als het vat niet goed is?’ Jij bleef echter altijd vertrouwen hebben in de uiteindelijke uitkomst (we konden immers altijd nog ‘naar Zaragossa gaan’) en trachtte menigmaal mijn te strenge geweten en eisen te verzach- ten. Ik heb veel geleerd van je snelle analyses van artikelen en wetenschappelijk werk. Tijdens de jaren dat ik je ken, ben je hooggeleerd geworden en heb je de onderzoeksafdeling stevig uitgebouwd tot een ware pijler binnen JellinekMentrum. Dank voor je steun en ik hoop in de toekomst nog veel met je samen te werken.

Robert, ik ben blij en vereerd dat je mijn co-promotor bent. Je bent zeer betrokken en een ware motivator. Je las mijn artikelen snel en grondig en je grondige commentaar bracht mij altijd verder. Je had vaak je complimenten klaar en kon putten uit eigen ervaring rond je promotie om mij te motiveren weer verder te gaan. Ik herinner me nog de vele steunende mailtjes. Het is een genoegen om met je te discussiëren over wetenschappelijke onderwerpen; zo herinner ik mij levendig onze gesprekken over de RCTs versus cohort studies en deze hebben mijn denken daarover aangescherpt. Ook je commentaar op het Engels was immer vruchtbaar, je hebt daar een bijzonder gevoel voor. Ik vond en vind het ook altijd gezellig om met je te praten over andere zaken, zoals werk, wonen en leven in het algemeen. Ik wil je bedanken voor je prettige en zorgvuldige begeleiding en wens je in alle opzichten alle goeds voor de toekomst. Ik hoop dat wij elkaar daarin blijven tegenkomen.

Frans, je hebt een speciale plaats in mijn ‘promotoren-troika’. Je bent de afgelopen jaren niet alleen mijn inspirator geweest op wetenschappelijk terrein, maar je was (en bent) ook een van de belangrijkste opleiders in mijn klinische opleiding tot GZ-psycholoog, en nu tot Psychotherapeut. Jij hebt mij laten ervaren dat wetenschap echt leuk kan zijn. Je bent een begenadigd docent met een

245 verbazingwekkend vermogen om de meest ingewikkelde problemen en vraag- stukken te reduceren tot een van je befaamde, verhelderende twee-bij-twee tabellen. Je kunt de brug slaan tussen wetenschap en de klinische praktijk. Je weet over te brengen dat de ‘N-en’ in een onderzoek patiënten zijn en wat onderzoek betekent voor de klinische praktijk. Je kritische en scherpe blik hebben mij veel geleerd. Jij bent betrokken geweest bij al mijn artikelen en het is met plezier dat ik aan de vele gesprekken erover terugdenk. Ook ben je een taal-virtuoos die onvermoeibaar op zoek gaat naar verbeteringen en mij menig nieuw Engels begrip heeft aangedragen. De afgelopen jaren ben je ook mijn supervisor geweest van mijn klinische opleiding, vooral voor de KPSP-therapieën. Met veel humor en diepgang bespraken we in de groepssupervisies het pad dat ik met mijn patiënten liep en de echo daarvan klinkt zeker na in al mijn behandelingen. Niet in de laatste plaats inspireren onze gesprekken en lezingen over boeken, films en filosofie mij bijzonder en hoop ik dat wij die nog lange tijd zullen voortzetten. Ik ben je veel dank verschuldigd.

De beoordelingcommissie van mijn proefschrift, Aartjan Beekman, Pim Cuijpers, Jan Swinkels, Willem van Tilburg en Thijs de Wolf, wil ik bedanken voor hun zorgvuldige aandacht die ze aan mijn werk hebben besteed. Ik ben vereerd dat jullie dit proefschrift wilden beoordelen en als opponenten met mij wilden discussiëren over zijn merites. Dank daarvoor.

Een volgend verhaal betreft de mensen die bij het JellinekMentrum Depres- sieonderzoek zijn betrokken vanaf het eerste uur. Simone Kool, Gerda van Aalst en Joseff Iping hebben de start en vliegende vaart van het onderzoek mogelijk gemaakt, samen met Frans en Jack, en zijn tot op de dag van vandaag betrokken bij en bevlogen over het onderzoek. Simone was bij mijn weten de eerste die er op promoveerde en haar goede voorbeeld doet vol- gen. Daarnaast zijn er uiteraard vele anderen die meewerkten en meewer- ken aan het onderzoek, zoals de medewerkers van de onderzoeksafdeling, de onderzoeksassistenten en een groeiende groep therapeuten binnen Jellinek- Mentrum die bijdraagt aan het versterken van de ‘evidence based medicine’ binnen JellinekMentrum. Ik wil jullie allen danken voor de inzet voor het depressieonderzoek, zonder welke dit proefschrift niet tot stand had kunnen komen.

Twee andere pijlers zijn Rien Van en Mariëlle Hendriksen. Jullie zijn met hart en ziel betrokken bij het onderzoek en het psychoanalytisch gedachtengoed. Onze avonden met de ‘KPSP-kerngroep’ zijn inspirerend en vooral ook informatief wat betreft de laatste roddels in ons vakgebied. Rien, we konden altijd even op de gang in Noord ervaringen uitwisselen over publicaties en de frustraties van een promotietraject. Fijne steun! Mariëlle, we zijn helaas toch niet samen begonnen aan de NVPP opleiding, maar nu kan ik mooi van jou horen hoe het er daar aan toegaat. Ik wens je veel succes met het schrijven van vele mooie artikelen.

246 Een aparte plaats als ‘strijder van het eerste uur’ verdient Jaap Peen. Jaap, ik kan je met recht de ‘stille kracht’ achter het onderzoek noemen. Je bent vanaf het begin betrokken geweest bij het Depressieonderzoek en bent voor de vele publicaties die er inmiddels uit voortkomen, het statistische brein en geweten. Je hebt ook de statistische analyses verzorgd van de meeste van de artikelen in dit proefschrift. Je was altijd bereid om analyses en weer nieuwe analyses uit te voeren en je bescheidenheid is legendarisch. Dank voor je werk. Ik wens je veel succes en plezier toe in het afronden van je eigen proefschrift.

Angelique Vergeer heeft mijn grote dank voor alle publicaties die zij de afgelopen jaren met engelengeduld voor mij heeft achterhaald in de verschillende bibliotheken. Niet voor niets hangt er in de bibliotheek van JellinekMentrum een foto met het passende opschrift ‘Bibliotheca Angelica’. JellinekMentrum moet zuinig op je zijn! Jurrijn Koelen heeft veel monnikenwerk verricht bij het elektronisch ‘submitten’ van mijn artikelen. Dit is een niet te onderschatten ingewikkelde klus, die ik maar wat graag uit handen gaf. Je deed het met veel geduld en overleg, en dat naast je eigen onderzoekswerk. Ik weet dat je mijn interesse in het psychoanalytische gedachtengoed deelt en ik wens je toe dat je daarin ook helemaal je plek gaat vinden. Wilma van Gelderen heeft eveneens een belangrijk aandeel gehad in het ‘klaarstomen’ van artikelen voor publicatie. Je was heel precies en vooral bij de review van LPaT studies heb je gezorgd dat de eindeloze reeks referenties uiteindelijk, na een rommelige start van mij, helemaal goed terechtkwam. Dank ook voor alle andere werkzaamheden die je voor mijn proefschrift hebt gedaan!

Else van Geldorp, mijn hoofd bij JellinekMentrum Noord, wil ik bedanken voor haar bereidheid om mij een parttime opleidingsplek op ‘haar’ poli te geven. Ik werkte parttime als GZ-psycholoog in opleiding en kon zo ook nog tijd aan onderzoek besteden. Else, we hebben samen eindeloos geworsteld met alle verschillende arbeidscontracten die ik heb gehad. Ik ben begonnen bij je als stagiaire (de eerste, zo heb ik begrepen) en onderzoeksassistent, daarna als werkervarings-psycholoog, als GZ-psycholoog in opleiding (ook nog met verschillende uren) en nu als GZ-psycholoog. Het kostte eindeloos veel tijd en moeite om de bijbehorende contracten in goede banen te leiden, maar je was er altijd heel betrokken en vasthoudend in. Je was, in samenwerking met Jack, ook bereid om mee te werken aan de weken ‘schrijfverlof’ die ik wilde om aan dit proefschrift te schrijven. Ik ben je dankbaar voor je flexibiliteit en de fijne manier waarop je zorgt dat het in Noord zo goed toeven is.

Mijn werkbegeleiders in de GZ-opleiding van de afgelopen jaren, Wies Beens en Barbara Jedding, wil ik bedanken voor het vele dat ik van ze heb geleerd op het psychologisch vakgebied. Jullie zijn heel belangrijk in het verhaal van mijn wording tot psycholoog. Maar ook en vooral wil ik jullie bedanken voor jullie goede zorgen en adviezen dat ik ook ‘goed op mezelf moest passen’. Die had ik zeker nodig met deze combinatie van klinische opleiding en onderzoekswerk. Ik heb veel geluk gehad met jullie.

247 Alle collega’s in Noord wil ik bedanken voor de leuke en inspirerende samenwerking en hun belangstelling voor onderzoek en wat daarbij komt kijken. We hebben daar in Noord echt een fantastische poli en dat maakt de combinatie van onderzoek en klinisch werk helemaal goed. Ik hoop nog lang met jullie te kunnen samenwerken!

Verschillende mensen hebben meegewerkt aan de totstandkoming van dit boek. Dirk de Jonghe heeft veel mooi werk verricht met betrekking tot de engelse vertalingen. John Kanters heeft de lay-out van het binnenwerk verzorgd en ook nog binnen een heel kort tijdsbestek en Annelies Frölke heeft het fraaie omslag ontworpen. Allen dank, ik ben heel blij met het resultaat. Renate Kleinveld wil ik bedanken voor haar bijdrage aan het verzenden van dit boek, een voor mij ingewikkelde klus was voor jou een fluitje van een cent!

Natuurlijk heb ik van vrienden en familie veel steun en belangstelling mogen krijgen tijdens het werk aan dit proefschrift. Het zijn er te veel om op te noemen, maar iedereen die mijn verhalen daarover moest aanhoren, dank!

Mijn ouders, Jopie en Frans, wil ik bedanken voor hun liefde, hun onwan- kelbare vertrouwen in mij en het immer stimlueren van mijn persoonlijke ont- wikkeling. Jullie zijn natuurlijk in alle opzichten ‘de basis’ van mijn verhaal. Ik wil Caroline, mijn zus, en Daniëlle, mijn schoonzus, bedanken dat zij mijn paranimfen wilden zijn. Jullie gingen mij beiden voor in het promoveren, dus wisten precies hoe het allemaal voelde.

Heel speciaal wil ik Marcel, mijn man, bedanken. Marcel, jij hebt het allemaal van heel dichtbij meegemaakt. Elke hobbel, tegenvaller, succes, de dagen achter de computer, de twijfels, de ups en downs in de motivatie; jij was er altijd om me te steunen, op te beuren als het minder ging en je was altijd nog blijer en trotser dan ikzelf als een artikel was geaccepteerd. Zonder jou had ik niet alleen dit proefschrift niet kunnen schrijven, maar zou ik ook niet de ‘switch’ hebben kunnen maken van het bedrijfsleven naar de psychologie, nu alweer enkele jaren geleden. Zonder jou zou mijn leven trouwens in het algemeen aanzienlijk minder leuk zijn! Je verstaat de kunst om mensen te inspireren en te motiveren om te ‘worden wie ze zijn’. Wat anders doet een therapeut? Voor mij, en voor iedereen in jouw omgeving, is jouw ‘human talent’ altijd voelbaar. Dank je daarvoor.

248