Beyond Brand Names of Psychotherapy: Identifying Empirically Supported Change Processes

Home , Focusing (psychotherapy), Psychotherapy

BEYOND BRAND NAMES OF PSYCHOTHERAPY: IDENTIFYING EMPIRICALLY SUPPORTED CHANGE PROCESSES

J. STUART ABLON AND TAI KATZENSTEIN RAYMOND A. LEVY University of California at Berkeley Massachusetts General Hospital and Harvard Medical School There is considerable debate about characterized the treatments: Adher- which empirical research methods best ence to cognitive–behavioral process advance clinical outcomes in psycho- was most characteristic, adherence to therapy. The prevailing tendency has interpersonal and psychodynamic pro- been to test treatment packages using cess, however, was most predictive of randomized, controlled clinical trials. positive outcome. Specific process pre- Recently, focus has shifted to consider- dictors of outcome were identified using ing how studying the process of change the Psychotherapy Process Q-Set. in naturalistic treatments can be a use- These findings demonstrate how process ful complement to controlled trials. Cli- research can be used to empirically nicians self-identifying as psychody- validate change processes in naturalis- namic treated 17 panic disorder tic treatments as opposed to treatment patients in naturalistic psychotherapy packages in controlled trials. for an average of 21 sessions. Patients achieved statistically significant reduc- Keywords: panic disorder, psychother- tions in symptoms across all domains. apy process and outcome, psychody- Rates of remission and clinically signif- namic psychotherapy, change processes icant change as well as effect sizes were commensurate with those of empirically supported therapies for panic Much attention has been paid recently to the conflict between those who favor the empirically disorder. Treatment gains were main- supported treatment (EST) movement and those tained at 6-month follow-up. Intensive who are skeptical of this model. It is beyond the analysis of the process of the treat- scope and not the intention of this paper to review ments revealed that integrative elements the specifics of this complicated argument. This study, however, represents an attempt to bridge the gap between these warring factions. As has J. Stuart Ablon, Raymond A. Levy, Psychotherapy Re- been suggested (Ablon & Jones, 2002; Ablon & search Program, Massachusetts General Hospital and Harvard Marci, 2004; Westen, Novotny, & Thompson- Medical School; and Tai Katzenstein, University of California Brenner, 2004) highlighting the significant limi- at Berkeley. tations of controlled clinical trials does not mean We thank the clinicians from the MGH Psychotherapy that empirical research has nothing to offer the Research Program for their willingness to participate in the practice of psychotherapy. Likewise, assuming study and the members of the Berkeley Psychotherapy Re- that empirical methods can contribute to advanc- search Project for their invaluable help with process ratings. Correspondence regarding this article should be addressed ing clinical outcomes does not mean that experi- to J. Stuart Ablon, PhD, Department of Psychiatry, MA Gen- enced practitioners do not know how to practice eral Hospital, 313 Washington Street, Suite 402, Newton, MA effectively. On the contrary, many researchers 02458. E-mail: [email protected] have argued that a focus on empirically validat-

216 Beyond Brand Names of Psychotherapy ing change processes in naturalistic treatment Among these psychosocial interventions, cogni- would be a fruitful complement to controlled tive–behavioral treatments (CBT) have most trials (Garfield, 1998; Goldfried & Wolfe, 1996; consistently demonstrated efficacy using the con- Howard et al., 1996). This study represents one trolled clinical trial paradigm. The success of attempt to shift the focus on prescriptive treat- cognitive–behavioral protocols, such as Panic- ment packages to studying intensively what a Control Therapy (Barlow et al., 1989), in allevi- group of experienced clinicians do when treating ating panic patients’ symptoms is reflected in the patients with shared diagnoses and presenting delineation of CBT interventions as “standard problems. As such, the treatments we studied treatment” by the NIH Consensus Development might be best described as representing “treat- Conference on Panic Disorder. It is important to ment as usual” in the community. Assuming that note, however, that most ESTs such as Panic experienced clinicians might help their patients Control Therapy have not been tested systemati- achieve symptomatic improvement, we used em- cally against legitimate alternative psychosocial pirical methods to identify the change processes treatments or treatment as usual in the commu- present in a naturalistic treatment so that we nity. Rather, in controlled trials in the laboratory, could learn how and why patients improved. If treatments are usually tested against medication, empirically validated change processes could be wait list controls, psychoeducation, or some ver- identified, we would have an empirical basis from sion of purely supportive intervention. Therefore, which to develop or amend clinically relevant clinicians are often unsure as to the true efficacy treatments. of such approaches relative to other forms of Our previous research (see Ablon & Jones, treatment. In addition, ESTs for diverse clients 1998, 2002) has demonstrated the dangers of with panic disorder have yet to be identified. drawing conclusions about why a treatment is Why do clinicians continue to practice, and effective without studying process correlates of why should we explore the potential utility of outcome. Even under tightly controlled condi- exploratory (as opposed to prescriptive) psycho- tions, treatments often share significant elements therapy for panic disorder given the high efficacy of process borrowed from other theoretical ap- rates of cognitive–behavioral treatment (CBT)? proaches. Interestingly, these borrowed elements Though CBT interventions have proven highly can be the ingredients that predict positive outcomes for patients. In this study, we asked clini- effective for many individuals, this approach is cians who self-identified as psychodynamic in not effective for all patients (Craske & Barlow, their primary theoretical orientation to treat a 2001; Milrod et al., 2001). In closely controlled group of patients with panic disorder as they trials, investigators have reported as many as normally would in their clinical practices while 38% of patients remain symptomatic (Milrod et we studied the process and outcome of the treat- al., 2001) or relapse subsequent to treatment dis- ments. We chose psychodynamic therapists and continuation (Milrod & Busch, 1996). Exposure- panic disorder patients because psychodynami- based interventions that work well for some pa- cally oriented clinicians frequently treat patients tients can be prohibitively overwhelming for with panic disorder despite the fact that psy- others. Some patients also fail to or refuse to chodynamic psychotherapy is not an EST for comply with the directive approaches and out-of- panic disorder and ESTs do exist for this popu- session work constituting the cornerstone of lation (e.g., Panic Control Therapy, see Barlow et many cognitive–behavioral treatments (Milrod et al., 1989) that have been well tested in the labo- al., 2001). For other patients, the idea of a treat- ratory. We briefly describe the background of ment that is not based on the exploration of psychotherapeutic treatment of panic disorder be- the personal meaning of symptoms is intellectu- low before detailing the methods used. ally unsatisfying to the point that they reject the Because of the tremendous and incontrovert- treatment. Clearly, for certain panic patients symp- ible psychological, emotional, and social costs of toms persist despite treatment with cognitive– living with panic disorder (Markowitz, Weiss- behavioral interventions. Nonprescriptive psy- man, & Ouellette, 1989; Swenson, Cox, & chotherapeutic approaches are needed to meet the Woszezy, 1992) much effort has been dedicated treatment demands of the subset of panic patients to developing and implementing effective psy- who do not respond to prescriptive approaches or chotherapeutic interventions for panic disorder. elect to pursue exploratory treatments.

217 Ablon, Levy, and Katzenstein

Problematically, to date in the literature there clinicians practicing psychotherapies that have has been little systematic investigation of nonpre- not been thoroughly evaluated empirically. Fur- scriptive treatments. There is support for the ef- thermore, empirical studies examining change fectiveness of brief psychodynamic treatments in processes in nonprescriptive treatments for panic the form of qualitative case studies and theoreti- disorder do not exist to our knowledge. It is cal papers (Abend, 1989; Gabbard, 1990; Milrod likely, however, that process predictors common & Busch, 1996; Milrod & Shear, 1991; Sifneos, to many therapies, such as the degree of thera- 1972). Studies examining the effectiveness of peutic alliance fostered, are likely predictors of supportive, nondirective, and nonbehavioral outcome in psychotherapy of panic disorder as treatments for panic disorder (D. F. Klein, Zitrin, well (e.g., Crits-Christoph and Gibbons, 2003). Woerner, & Ross, 1983; Shear, Pilkonis, Cloitre, The aims of this study are to (1) examine the & Leon, 1994) offer further empirical evidence degree of change associated with a naturalistic for the potential utility of exploratory treatments. psychotherapy for panic disorder in a within- However, only 2 studies have systematically ex- subject sample; (2) identify which prototypical amined the effectiveness of psychodynamically treatment processes best characterize the treat- oriented psychotherapies for panic disorder (Mil- ments; (3) identify which prototypical processes rod et al., 2001; Wiborg & Dahl, 1996). Wiborg are most predictive of positive outcome; (4) iden- and Dahl (1996) compared the effectiveness of tify the most and least characteristic elements of 15 sessions of brief dynamic psychotherapy in the process of the treatments at a specific, atheo- conjunction with nine months of pharmacother- retical level; and (5) identify which specific pro- apy to pharmacotherapy alone. The investigators cess variables predict positive outcome. Specifi- concluded that brief dynamic psychotherapy in cally, we hypothesized that (1) naturalistic conjunction with pharmacotherapy was a more psychotherapy for panic disorder would be a effective form of treatment than clomipramine highly effective treatment with gains commensu- alone. Milrod et al. (2001) have conducted the rate with those achieved by prescriptive treat- only quantitative investigation to date in the lit- ments; (2) that the treatments would be charac- erature examining the effectiveness of brief psy- terized by a high degree of psychodynamic chodynamic psychotherapy as a primary inter- process and significantly less by elements of in- vention for panic disorder. By the end of terpersonal and cognitive–behavioral process; (3) treatment, the majority of study entrants (16/21) that positive outcome would be predicted by the and completers (16/17) met criteria for remission degree to which psychodynamic (rather than in- as specified by the multicenter panic study (Bar- terpersonal and cognitive–behavioral) process low, Gorman, & Shear, 1997). Statistically sig- was fostered; (4) that the treatments would be nificant and clinically meaningful changes across characterized by elements typical of psychody- a broad range of outcome domains including pri- namic therapy including attention to the thera- mary panic symptoms, phobic sensitivity, and peutic alliance and relationship, interpretation of overall quality of life were reported. These treat- defense mechanisms, identification of uncon- ment gains were maintained at 6-month follow- scious feelings and wishes deemed dangerous, up. The effect sizes reported in this study are and the linking of current symptoms, behaviors, consistent with those reported in studies of ESTs and feelings to past experiences; and (5) that for panic disorder (Milrod et al., 2001). This these specific variables along with a focus on study provides evidence that nonprescriptive facilitating emotional expression would be most treatment, particularly psychodynamic psycho- predictive of positive outcome. therapy, may hold promise as a stand-alone treatment for panic disorder that could be offered as an alternative to compare to ESTs such as Panic Method Control Therapy in controlled trials. Given the Participants paucity of confirmatory studies, however, a rea- sonable conclusion at present is not that nonpre- Participants were 17 patients between the ages scriptive approaches are ineffective. Rather, they of 24 and 55 meeting Structured Clinical Inter- have yet to be tested empirically in a thorough view for DSM–IV (SCID-IV) criteria for diagno- fashion. This is concerning because it is likely sis of panic disorder at the Massachusetts General that there are many psychodynamically trained Hospital (MGH) outpatient psychiatry service in

218 Beyond Brand Names of Psychotherapy

Boston. Several different recruitment mecha- asked not to make changes in medicine until nisms were utilized: Advertisements in the scien- termination of the study so that changes during tific study section of a city newspaper, posters in the study period would not be confounded by general and psychiatric waiting rooms throughout medication changes. During the course of the the hospital, posters in major local universities’ study, however, 1 participant changed medication mental health clinics, description of the study in dosage under psychiatric supervision. After con- hospital-wide emails, and letters to psychiatrists, sultation with the psychiatrist, this participant psychologists, and social workers affiliated with was retained because the small change was not the MGH Department of Psychiatry. Prior to believed to represent a serious confound. While gathering any assessment information, the Re- receiving treatment, 2 participants also ceased search Coordinator explained the study, including taking benzodiazapenes on an as needed basis any potential risks and benefits, to potential pa- because of improvement in their functioning dur- tients and then obtained informed consent with ing the study period. oversight from the Human Subjects Committee of Partners Health Care System and the Massa- Therapists chusetts General Hospital. Exclusion criteria for this study included current drug or alcohol abuse, The 7 participating clinicians were all affiliated bipolar disorder, psychosis, suicidality, concur- with the Outpatient Department of Psychiatry at rent psychotherapy or counseling, and any antic- Massachusetts General Hospital. The group in- ipated or actual changes to medication (dosage or cluded 1 psychiatrist, 1 psychiatric resident, 2 psy- type) less than 8 weeks prior to study entry. chologists, and 3 psychology interns/postdoctoral Of the 17 patients entering treatment, 88.2% fellows. Clinicians averaged 12 years of clinical were female, 11.8% were male. Within the pool experience. Five of the participating clinicians of participants, 77.8% identified themselves as were male, 2 were female, all were Caucasian. Caucasian and 22.4% described themselves as The number of patients seen by the same clini- Haitian, Hispanic, or Asian-Indian. The average cian ranged from 1 to 4. All clinicians identified age of participants enrolled in this study was 35. their primary theoretical orientation as psychody- All subjects met diagnostic criteria for current namic. Therapists were not asked to describe panic disorder. In terms of comorbid disorders, their orientation with any greater specificity be- approximately 6% of participants in this sample cause a primary aim of the study was in fact to met diagnostic criteria for current major depres- identify empirically what type of treatment pro- sion, 50.0% met criteria for past major depres- cesses they fostered in actual practice using con- sion, 66.7% met criteria for panic disorder with sensus based definitions from different theoreti- agoraphobia, 38.9% met criteria for current gen- cal orientations. Although not an EST for panic eralized anxiety, 11.1% met criteria for current disorder, psychodynamically oriented clinicians social phobia disorder, and 5.6% met criteria for often treat panic disorder patients using nonpre- obsessive–compulsive disorder. Among study scriptive therapies. participants, 61.1% reported having previously taken some form of psychotropic medication. Treatment Seventy-six percent of the sample reported previously pursuing psychotherapy. In order to replicate nonprescriptive psycho- Since this was a naturalistic treatment, patients therapy conducted in the community, clinicians were allowed to continue taking psychotropic were asked to conduct a nonprescriptive therapy medication during the study as long as no as they normally would in their clinical practice. changes were made 2 months prior to study en- As described previously, nonprescriptive thera- rollment and the patient still met criteria for panic pies are frequently used to treat panic disorder disorder at baseline. While enrolled in the study, patients who cannot tolerate or do not respond to 52.9% (n ϭ 9) of the participants were concur- medicine or ESTs such as Panic Control Therapy. rently on medication (5 patients taking benzodi- In addition, many patients seek nonprescriptive azapines, 2 taking a benzodiazapine and antide- alternatives to medicine or ESTs. There were no pressant, 1 patient taking only an antidepressant, restrictions on the kind of therapy offered other and 1 patient taking medications belonging to than the length of treatment. Therefore, the char- multiple pharmacological classes). Patients were acteristics of the treatments provided were not

219 Ablon, Levy, and Katzenstein known by the investigators until after the study pertain to domains such as attack frequency and was completed. Thus, speciﬁc descriptions of the level of distress experienced. This questionnaire components of the treatments are presented in the has demonstrated high interrater reliability with results section of this study. Prior to the initial intraclass correlations of .88 and individual item session, patients were told that treatment entailed reliability ranging from .73 to .87 (Shear et al., one 50-minute session per week for 6 months for 1997). Shear et al. (1997) have demonstrated that a total of approximately 24 sessions. The mean individual items on the PDSS correlate strongly number of sessions in this study was 21. On with other measures assessing similar constructs. average, patients were seen for a total of 7 The Symptom Checklist-90–Revised (SCL- months, but treatment length ranged from 6 to 48 90–R) (Derogatis, Lipman, Rickels, Uhlenhuth, weeks. Clinicians were asked to treat the patient & Covi, 1974) and Quality of Enjoyment and for 22–26 sessions. This range was selected so Satisfaction Questionnaire (Q-LES-Q) (Endicott, that termination would not be completely arbi- Nee, Harrison, & Blumenthal, 1993) were used to trary and clinicians would have some freedom to assess general psychological and physical func- decide when to terminate treatment. tioning. The SCL-90–R consists of 9 principle symptom subscales and 3 global indices assess- Outcome Measures ing general psychological distress. It is a 90-item measure. Each item is endorsed on a 1 (not at all) Outcome measures designed to assess patient to 5 (extremely) scale. The SCL-90–R has dem- functioning across a range of domains and from onstrated good reliability, validity, and sensitivity different perspectives (patient, therapist, indepen- to psychotherapeutic change (Derogatis, 1994). dent rater) were administered at monthly inter- The Q-LES-Q contains a total of 93 items. vals. Follow-up questionnaires (patient) were ad- Ninety-one items comprise 8 scales assessing ministered 6 months subsequent to termination. quality of life functioning across a range of do- Independent raters assessed the severity and in- mains including school, work, and physical tensity of patients’ panic at baseline and termina- health. The summary scales of the Q-LES-Q have tion following one-to-one discussions with pa- demonstrated satisfactory test–retest reliability tients about their attacks. Certain outcome ranging from .62 to 89, satisfactory internal con- measures were selected because of their assigna- sistency ranging from .80 to .96, as well as con- tion as standard instruments to be used in empir- current validity (Endicott et al., 1993). ical investigations of panic disorder by the 1994 NIH conference report (Shear & Maser, 1994). Clinician Measures Other measures were chosen by virtue of their relevance to exploratory psychotherapies. The Clinical Global Impression Scale (CGI) (Guy, 1976) was selected to assess subjects’ panic Patient Self-Report Measures symptomotology. The CGI scales assess degree of global improvement and severity of illness ona1to The Anxiety Sensitivity Index (ASI) (Reiss, 7scale.Numberofsymptoms,levelofanxiety, Peterson, Gursky, & McNally, 1986) and Panic avoidance, and level of functioning serve as the Disorder Severity Scale (PDSS) (Shear et al., criterion upon which these ratings are based. The 1992) were selected to assess panic symptomatol- Global Assessment of Functioning Scale (GAF) ogy. The ASI is a self-report questionnaire with (Endicott, Fleiss, & Cohen, 1976) was selected to 16 items geared toward capturing fear of anxiety- gauge general psychological functioning. The GAF related symptoms. Each item is rated on a 5-point assesses psychological, social, and occupational scale. The ASI has demonstrated excellent reli- health on a zero to one-hundred scale. Both the CGI ability across numerous studies ranging from .82 and GAF have demonstrated satisfactory levels of to .91 (Peterson & Reiss, 1993). The measure has reliability and validity. The Defensive Functioning also demonstrated satisfactory test–retest reliabil- Scale (DFS) (DSM–IV)andSocialCognitionand ity ranging from .71 to .75 (Peterson & Reiss, Object Relations Scale (SCORS) (Eudell-Simons, 1993) and has demonstrated both criterion and Stein, DeFife, and Hilsenroth, 2005) were used to construct validity (Peterson & Reiss, 1993; Reiss evaluate subjects’ defensive styles and object rela- et al., 1986). The PDSS is a 7-item scale. All tions. The DFS lists 7 principal levels of defensive items are rated on a 5-point scale. Questions functioning within which current coping styles can

220 Beyond Brand Names of Psychotherapy be evaluated. Clinicians are asked to assign a level Jones, Krupnick, & Kerig, 1987; Jones & Pulos, of defensive functioning at the time of assessment 1993). ranging from normal to psychotic. Two examples The PQS utilizes a fixed distribution to ipsa- drawn from opposite ends of the spectrum include tively describe most and least characteristic ele- high adaptive level (defense mechanisms represent- ments in a psychotherapeutic hour. By enforcing ing optimal adaptation to managing stressors in- a fixed normal distribution, the measure ensures cluding humor, affiliation, and self-assertion) and multiple evaluations of items and attenuates rater action level (defense mechanisms that rely on action biases such as response sets and halo effects. or withdrawal to manage stressors [e.g., acting out Different from other process measures in the field and help-rejecting complaining]). The SCORS con- which typically examine segments of the thera- tains 8 questions assessing such dimensions as un- peutic hour, the PQS uses an entire hour as the derstandings of social causality and identity and unit of analysis thereby facilitating a much more self-coherence. All ratings are made ona1to7 representative view of the hour. The PQS’s pan- scale. The DFS and SCORS have been less widely theoretical orientation enables comparisons be- applied in research settings and are therefore less tween different treatment orientations. well established psychometrically. Prototypes of ideal psychotherapeutic process have been developed using the PQS for a range of theoretical perspectives. Previous research has Independent Rater Measure demonstrated how correlating process ratings with the prototypes can provide an empirical The Multi-Center Panic Anxiety Scale (MC- measure of the degree to which a treatment ad- PAS) (Barlow et al., 1997) was used to assess the heres to the theoretical principles of a given ori- frequency and severity of participants’ panic entation (see Ablon & Jones, 1998, 2002 for symptoms at baseline and termination. This in- more information on the development of the pro- strument is identical in content to the previously totypes). Prototypes of interpersonal, psychody- described patient PDSS with the same psycho- namic, and cognitive–behavioral process were metric properties, only completed by an indepen- used in this study. dent rater as opposed to the patient. A pool of 8 research-oriented psychologists and master’s level graduate students in clinical Process Measure psychology, trained in use of the PQS, completed ratings of audiotapes from this study. For the The Psychotherapy Process Q-Set (PQS). Thera- purposes of this study, Session 12 (the midpoint peutic process was examined using the PQS of most treatments) was selected as a representa- (Jones, 2000). The PQS is an instrument consist- tive hour to be Q-sorted. For each session, inde- ing of 100 items describing actions, behaviors, pendent ratings were completed by at least 2 and thoughts of both therapist and patient in judges. A composite Q-rating for both raters was individual as well as dyadic terms. Several char- then calculated for each item. If reliability be- acteristics of the PQS speak to its strengths as a tween raters was below r ϭ .50, a 3rd rater was measure. It has demonstrated reliability and va- added. To counter rater drift, periodic calibration lidity across a variety of different treatment sam- meetings were conducted. The average alpha co- ples including archived treatments of psychody- efficient reliability for raters completing Q-sorts namic, cognitive–behavioral, client-centered, of therapy sessions in this sample (n ϭ 17) was gestalt, rational-emotive, and interpersonal ther- .85. This far surpasses the generally acceptable apies (Ablon & Jones, 1999, 2002; Jones, Cum- criterion (.70) used to determine acceptable reli- ming, & Horowitz, 1988; Jones, Hall, & Parke, ability in therapy process and outcome research 1991; Jones & Pulos, 1993). The interrater reli- (Orlinsky & Howard, 1986). ability across all 100 PQS items has consistently Several methods were used to analyze the yielded alpha coefficients between .83-.89 per PQS process data. The process ratings were rater pair. Reliability analyses for individual correlated with the prototypes of ideal treat- Q-items have also yielded acceptable values (be- ment process to determine the degree to which tween .50 and .95) across samples. The measure’s the treatments adhered to the prescriptions of construct and discriminant validity has been dem- each theoretical orientation. Adherence to the onstrated across studies (Jones et al., 1988, 1991; prototypes was also correlated with outcome.

221 Ablon, Levy, and Katzenstein

The most and least characteristic items of the and raters reported decreases in the severity of treatments were listed to describe the overall patients’ panic attacks as well as improvement process. Finally, individual process items were in general functioning (Clinical Global Impres- correlated with outcome. sion, Multicenter Panic Anxiety Scale & Global Assessment of Functioning). From cli- Results nicians’ perspectives, patients demonstrated no statistically significant change in defensive Outcome: Is Naturalistic Psychotherapy an functioning from pre- to posttreatment. Clini- Effective Treatment for Panic Disorder? cians, however, did report statistically signifi- Treatment gains across time points were cal- cant change (p Ͻ .05) in aspects of object culated using multiple statistical methods relations including patients’ emotional invest- (Lambert & Hill, 1994). Table 1 reports pre ment in values and moral standards (p ϭ .02) and posttreatment means on questionnaires ad- and changes in self-esteem (p ϭ .02). ministered to patients, clinicians, and indepen- Effect sizes were calculated as another index dent raters. Statistical significance and effect of change by subtracting posttreatment means size represent the change between these time- from pretreatment means and dividing by pre- points. Paired-sample t tests were used to de- treatment standard deviation. The effect sizes tect statistically significant mean differences suggest substantial improvement in outcome between pre and posttreatment. From pre to from baseline to endpoint. Fifty-three percent posttreatment, patients reported statistically of patients achieved remission according to a significant decreases in both the anticipation criterion used in several other studies in the and experience of anxiety (Anxiety Sensitivity literature (Milrod et al., 2001; Otto, Pollack, Index & Panic Disorder Severity Scale) as well Penava, & Zucker, 1999). Clinically significant as significant increases in overall functioning change was also calculated using a stringent (Symptom Checklist & Quality of Life Enjoy- method suggested by Jacobson and Truax ment and Satisfaction Questionnaire). Consistent with the patients’ perspectives, (1991). This method involves using mean clinicians and independent raters reported a scores for the “normal” population (i.e., adults statistically significant decrease in panic and with no Axis I psychiatric disorder) as well as anxiety from baseline to endpoint. Clinicians the “dysfunctional” population (i.e., in this case adult patients diagnosed with panic disorder). Patients were said to achieve clinically TABLE 1. Outcome of Naturalistic Psychotherapy for Panic significant change only if their posttreatment Disorder: Baseline to Endpoint means were closer to the normal mean than the Statistical Effect dysfunctional mean. This calculation was com- Measures Pre-mean Post-mean significance sizes pleted for the 2 measures for which population Patient means were available. Sixty-four percent and SCL-90–R 0.87 0.54 .01* .74 70% achieved clinically significant change on ASI 30.30 17.60 .00* 1.30 the Symptom Checklist and Anxiety Sensitivity PDSS 10.40 5.88 .00* 1.10 Index, respectively. Outcome analyses were Therapist CGI 4.20 2.60 .01* 2.40 also calculated after stratifying by medication GAF 59.8 71.60 .00* 1.80 status. Patients not concurrently taking medi- Rater cation achieved equivalent or better outcomes MC-PAS 5.03 3.30 .00* 1.10 at termination across all measures. Analyses of Note. Significant pre/post means were determined using the 6-month follow-up data revealed no statis- within-sample t-tests. Effect sizes were calculated using tically significant changes from endpoint to the formula: pretreatment mean-posttreatment mean/ pretreatment standard deviation. SCL-90–R ϭ Symptom follow-up. In other words, patients maintained Checklist-90–Revised; ASI ϭ Anxiety Sensitivity Index; treatment gains across all outcome measures 6 PDSS ϭ Panic Disorder Severity Scale; CGI ϭ Clinical months after termination. Thus, our first hy- Global Impression Scale; GAF ϭ Global Assessment pothesis, that naturalistic psychotherapy would of Functioning Scale; MC-PAS ϭ Multi-center Panic Anxiety Scale. N ϭ 17. be highly effective for treating panic disorder, *p Յ .01. was confirmed.

222 Beyond Brand Names of Psychotherapy

Adherence to Prototypes: Were the Treatments (t ϭ .70, df ϭ 16, p ϭ .496). Our second hypoth- Characterized Mostly by Psychodynamic esis, that the treatment would be best character- Process? ized by prototypical psychodynamic process, was not confirmed. To measure adherence to prototypes of ideal therapeutic process as stipulated by expert panels, prototypes were correlated with the Q-sort ratings Prototype—Outcome Correlations: Did the of the actual treatment sessions. For every pa- Degree of Psychodynamic Process Fostered tient, the composite Q-rating of each PQS item Best Predict Positive Outcome? was correlated with each item’s factor score from the corresponding prototype (refer to Ablon & Table 2 contains partial correlation coefficients Jones, 1998, 2002 for a more in-depth discussion representing the degree to which adherence to of prototype methodology). Pearson correlations each prototype predicted outcome on the main were transformed to z-scores using Fisher r to z outcome measures (Symptom Checklist, Anxiety transformations. Figure 1 displays the correla- Sensitivity Index, & Panic Disorder Severity tions with the 3 prototypes. The correlations with Scale). Partial correlations were used to control the CBT prototype (z score M ϭ .50, SD ϭ .14) for pretreatment level of severity. Positive corre- were the strongest followed by the psychody- lations reflect outcome in the desired direction. namic (z score M ϭ .35, SD ϭ .16) and interper- Adherence to the psychodynamic prototype was sonal prototypes (z score M ϭ .32, SD ϭ .09), significantly associated with positive outcome on respectively. There was a statistically significant 1 (Symptom Checklist) of the 3 outcome mea- difference in adherence to the cognitive– sures. Adherence to the interpersonal prototype behavioral versus psychodynamic and interper- was significantly associated with positive out- sonal prototypes (t ϭϪ2.4, df ϭ 16, p Ͻ .05; t ϭ come on two (Symptom Checklist, Anxiety Sen- 6.2, df ϭ 16, p Ͻ .001). No statistically signifi- sitivity Index) of the three outcome measures. cant difference in adherence to the psychody- Adherence to the cognitive–behavioral prototype namic versus interpersonal prototypes emerged was not associated with positive outcome. Thus,

FIGURE 1. Adherence to prototypes of ideal treatment process.

223 Ablon, Levy, and Katzenstein

TABLE 2. Prototypes Correlated with Outcome in a iors, raters agreed that clients generally felt un- Naturalistic Psychotherapy for Panic Disorder derstood by their therapists (Q 14), accepted their Cognitive therapist’s comments and observations (Q 42, r), Outcome behavioral Interpersonal Psychodynamic were trusting and secure (Q 44, r), as well as measures prototype prototype prototype compliant and deferential (Q 20 r). In addition, patients were viewed as active in the therapy (Q SCL-90–R Ϫ.18 .62** .50* ASI Ϫ.03 .64** .22 15, r), having little difficulty beginning the hour PDSS Ϫ.39 .22 .03 (Q 25, r), and as understanding therapists’ comments (Q 5, r). Note. Positive correlations reflect favorable associa- tions with outcome. All Pearson correlations are partial From the perspective of observers completing correlations controlling for pretreatment scores. N ϭ 17. SCL-90-R ϭ Symptom Checklist-90–Revised; ASI ϭ Anxiety Sensitivity Index; PDSS ϭ Panic Disorder Se- TABLE 3. Rank Ordering of PQS Item Means in verity Scale. Naturalistic Psychotherapy for Panic Disorder *p Ͻ .05. **p Ͻ .01. PQS # Psychotherapy Process Items Mean our third hypothesis that stated that the degree of 10 Most Characteristic Items prototypical psychodynamic process fostered 69 Patient’s current or recent life situation is would be the best predictor of positive outcome emphasized in discussion. 8.06 was only partially confirmed. 31 Therapist asks for more information or elaboration. 8.03 6 Therapist is sensitive to the patient’s Most and Least Characteristic Elements of feelings attuned to the patient; empathic. 8.00 Process: Were the Treatments Best 65 Therapist clarifies, restates, or rephrases Characterized by Elements Typical of patient’s communications. 7.97 23 Dialogue has a specific focus. 7.94 Psychodynamic Therapy? 62 Therapist identifies a recurrent theme in the patient’s experience or conduct. 7.56 The most and least characteristic aspects of the 18 Therapist conveys a sense of non- therapeutic process were calculated using Q-item judgmental acceptance. 7.50 means. As depicted in Table 3, means ranged 63 Patient’s interpersonal relationships are a from a high of 8.06 to a low of 1.47 on a scale major theme. 7.41 ranging from 1 to 9. In order to identify the 10 81 Therapist emphasizes patient feelings in order to help him or her experience them most and least characteristic items of treatment more deeply. 7.35 for the purposes of highlighting the most descrip- 92 Patient’s feelings and perceptions are linked tive processes, Q-items were rank ordered ac- to the past. 7.03 cording to their means. The Q-item numbers in 10 Least Characteristic Items the text correspond to the items in the table. R (reversed) indicates items where content was in- 9 Therapist is distant, aloof (vs. responsive verted to maintain narrative consistency. and affectively involved). 1.47 77 Therapist is tactless. 1.59 According to observer ratings, certain thematic 51 Therapist condescends to, or patronizes the similarities emerged across this group of treat- patient. 1.65 ments. Patients’ current or recent life situations 14 Patient does not feel understood by the (Q 69) as well as their interpersonal relationships therapist. 1.76 (Q 63) emerged as frequent topics of discussion. 5 Patient has difficulty understanding the therapist’s comments. 2.44 Generally speaking, dialogue between patients 42 Patient rejects (vs. accepts) therapist’s and therapists tended to have a specific focus comments. 2.59 (23). As judged by raters, the stance most char- 15 Patient does not initiate topics; is passive. 2.65 acteristic of clinicians was one in which clini- 44 Patient feels wary or suspicious (vs. trusting and secure). 2.79 cians were empathic and sensitive to patients’ 25 Patient has difficulty beginning the hour. 3.12 feelings (Q 6), responsive and affectively in- 20 Patient is provocative, tests limits of the volved (Q 9, r), tactful (Q 77 r), nonjudgmental therapy relationship. 3.18 (Q 18), and not condescending or patronizing (Q Note. PQS ϭ Psychotherapy Process Q-set. Ratings 51). range from 9 (most characteristic) to 1 (least character- Possibly in response to these therapist behav- istic). N ϭ 17.

224 Beyond Brand Names of Psychotherapy ratings, therapist interventions constituted salient TABLE 4. Individual Item Process Correlates of Outcome aspects of the therapeutic process. Therapists Effect asked for more information or elaboration (Q 31) PQS # Psychotherapy Process Items Size as well as clarified, restated, or rephrased patient’s communications (Q 65). They identified 81 Therapist emphasizes patient’s feelings to .70 recurrent themes in the patient’s experience or deepen them 74 Humor is used .52 conduct (Q 62). Therapists also tended to empha- 1 Patient verbalizes negative feelings toward .50 size patients’ feelings in order to help them ex- therapist perience these feelings more deeply (Q 81) as 97 Patient is introspective, explores inner .49 well as link patients’ feelings or perceptions to thoughts/feelings 73 Patient is committed to the work of therapy .49 the past (Q 92). 8 Patient is concerned/conflicted about .49 Our fourth hypothesis was that the treatments dependence on the therapist would be characterized by elements typical of 72 Patient understands the nature of therapy, .47 psychodynamic therapy. This was not strongly what is expected 75 Termination of therapy discussed .47 supported when examining the most and least 50 Therapist draws attention to feelings patient .43 characteristic items of the treatments. regards unacceptable 28 Therapist accurately perceives therapeutic .42 process 11 Sexual feelings and experiences are .40 Specific Process Correlates of Outcome: Were discussed the Elements Typical of Psychodynamic 96 Discussion of scheduling or fees .38 Therapy Predictive of Positive Outcome? 32 Patient achieves a new understanding or .37 insight In order to ascertain the specific aspects of the 71 Patient is self-accusatory expresses shame, .37 therapeutic process that were strongly associated guilt with positive outcome, partial correlations (con- 22 Therapist focuses on patient’s feelings of .34 guilt trolling for pretreatment) between Q-sort items 6 Therapist is sensitive to the patient’s .34 and patients’ outcome scores on the Symptom feelings, attuned, empathic Checklist were calculated. Given the relatively 12 Silences occur during the hour .32 small sample size, effect size rather than signifi- 92 Patient’s feelings/perceptions are linked to Ϫ.45 the past cance level was used to identify process corre- 38 Discussion of activities/tasks to do outside Ϫ.47 lates of outcome. A moderate correlation of r Ͼ session .3 was chosen as a cutoff point to identify robust 25 Patient has difficulty beginning the hour Ϫ.37 process correlates of outcome. The importance of 30 Discussion centers on cognitive themes, Ϫ.36 ideas, beliefs decreasing the probability of Type II error at the 76 Therapist suggests patient accept Ϫ.34 risk of increasing the probability of Type I error responsibility for problems influenced the selection of this value based on 45 Therapist adopts supportive stance Ϫ.33 Cohen’s estimates (Cohen, 1988). 37 Therapist behaves in a teacher-like Ϫ.33 Table 4 lists the 28 PQS items emerging as (didactic) manner 48 Therapist encourages independence of Ϫ.32 process correlates of outcome on the Symptom action/opinion Checklist. Of these, roughly the same number of 52 Patient relies upon therapist to solve his/her Ϫ.54 items described patient (N ϭ 11) and therapist problems (N ϭ 10) within session characteristics, experi- 35 Patient’s self-image is focus of discussion Ϫ.62 85 Therapist encourages patient to try new Ϫ.67 ences, and qualities. Several items described the ways of behaving with others nature of the interaction between the two (N ϭ 7). Some of the process-correlate items appeared to Note. Positive correlations reflect favorable associa- tions with outcome on the SCL-90–R. PQS ϭ Psycho- be thematically related. therapy Process Q-set; N ϭ 17. Several items reflecting a focus on feelings and negative emotion by patients and therapists were associated with positive outcome: The patient feelings regarded as unacceptable by the patient being self-accusatory/expressing shame and guilt (Q 50), and emphasizing patients’ feelings in an (Q 71), verbalizing negative feelings toward the effort to deepen them (Q 81). therapist (Q 1), the therapist focusing on patient’s Another group of items describing common feelings of guilt (Q 22), drawing attention to factors contributing to a strong therapeutic alli-

225 Ablon, Levy, and Katzenstein ance emerged as robust predictors of positive The active ingredients also included a supportive patient outcome on the Symptom Checklist: Hu- working alliance and the absence of a prescrip- mor being used (Q 74), the patient being intro- tive stance taken by the therapist. spective (Q 97), committed to the therapeutic work (Q 73), and understanding the nature and the expectations of the therapy (Q 72), the ther- Discussion apist accurately perceiving the therapeutic pro- Outcome of Naturalistic Psychotherapy for cess (Q 28), adopting a supportive stance (Q 45), Panic Disorder and being sensitive to the patient’s feelings, attuned and empathic (Q 6). As predicted, patients receiving a nonmanual- Patients feeling conflicted about dependence ized, exploratory, naturalistic psychotherapy for on the therapist (Q 8) was an additional item panic disorder demonstrated statistically signifi- correlating with positive outcome. Two other cant improvements in their anticipation and ex- items generally characteristic of a psychody- perience of anxiety as well as their social and namic viewpoint emerged as strongly associated relational functioning from baseline to endpoint with positive outcome: Sexual feelings are dis- and from baseline to follow-up. Improvements cussed (Q 11), and termination is discussed (Q also extended to the realms of physical health, 75). The following items were also correlates of quality of feelings, quality of social relations, positive outcome: Silences occur during the hour quality of general activities, and overall life sat- (Q 12), discussion of scheduling or fees occurs isfaction. Independent ratings of patients’ sever- (Q 96), and patient has difficulty beginning the ity and frequency of panic attacks were reduced hour (Q 25). on average by half from baseline to endpoint. Many of the process correlates associated with Effect sizes for patient, clinician, and observer negative outcome also shared thematic similari- completed measures all tended toward the same ties. Several of these items reflected hallmark direction. However, consistent with the literature, aspects of psychodynamic and cognitive– clinician effects sizes were approximately double behavioral approaches to treatment: Discussion those for patients and independent raters. The of activities/tasks to do outside session (Q 38), magnitude of the effect sizes in this pilot study discussion centers on cognitive themes (Q 30), ranged from slightly less to roughly equal those and patient’s feelings/perceptions are linked to reported in studies employing similar outcome past (Q 92). Other items associated with negative measures (Milrod et al., 2001; Otto et al., 1999). outcome reflected therapists’ actions: Therapist As might be expected, the remission rates in this suggests that patient accept responsibility for study were higher than those reported after group problems (Q 76), therapist behaves in a teacher- cognitive–behavioral therapy with “inadequately like (didactic) manner (Q 37), therapist encour- medicated” patients and lower than those re- ages independence of action/opinion (Q 48), ther- ported with “adequately treated” patients (Otto et apist encourages patient to try new ways of al., 1999). At both endpoint and follow-up, pa- behaving with others (Q 85). The following items tients’ levels of symptomatology and sensitivity were also correlated with negative outcome: pa- toward anxiety were closer to the mean of the tient relies upon therapist to solve his/her prob- normal population than the mean of adults diag- lems (Q 52), and patient’s self-image is a focus of nosed with panic disorder. The fact that this non- discussion (Q 35). manualized, naturalistic, exploratory treatment Our fifth and final hypothesis was that ele- was associated with statistically and clinically ments of psychodynamic process along with a significant change from baseline to endpoint war- focus on facilitating emotional expression would rants further exploration of similar naturalistic be most predictive of positive outcome. This was treatments for panic disorder. only partially confirmed in that the active ingredients did include a focus on helping patients to Adherence to Theoretical Models in recognize, experience and express negative or Naturalistic Psychotherapy disavowed emotions, sexual desires, and fears of dependence, but did not include linking patient’s Contrary to what one might expect, the treat- feelings and perceptions to the past or significant ments were characterized most highly by the focus on defense or transference interpretations. presence of cognitive–behavioral process despite

226 Beyond Brand Names of Psychotherapy the fact the clinicians all self-identified as psy- Prototype Correlates of Outcome chodynamic in orientation. Correlations with the prototypes did indicate moderate adherence to In terms of predicting outcome, adherence to prototypes of psychodynamic and interpersonal the interpersonal and to a lesser degree psychody- process as well but to a significantly lesser degree namic process predicted positive outcome. The than adherence to the CBT prototype. Several fact that the process characterizing these treat- factors may account for these curious findings. ments as a whole (cognitive–behavioral) was not One explanation may center on the divide sepa- reflected in the aspects of process predicting pos- rating theory from practice. As practitioners in itive outcome (interpersonal and psychodynamic) the field frequently note, the reality of “what has a precedent in the empirical literature. Results works” with a given patient often requires a mix- from a programmatic line of research indicate ing of technique, intervention, and therapeutic that the most prevalent aspects of therapeutic stance. In naturalistic settings where patients process are not necessarily the components pre- present with comorbidities and multiply deter- dicting outcome (Ablon & Jones, 1998, 2002). mined problems, clinicians frequently report Taken together, one implication of these findings drawing on different kinds of interventions from may be that the conceptualization of therapeutic various schools of thought as the specifics of the interventions as “purely” one orientation or an- case dictate. Previous research using the same other may in the end be more of a conceptual than process measure has indicated that, for instance, clinical reality. Another important implication psychodynamic clinicians tend to employ a di- might be that focusing on the predominant as- verse range of interventions when conducting pects of process alone can be misleading when it brief psychotherapy, fostering as much of a comes to the active ingredients of a treatment. cognitive–behavioral process as a psychody- Given the above findings, both the theoretical namic process (Ablon & Jones, 1998). Also, the orientation of the therapists and the most charac- nature of panic disorder itself, which is heavily teristic elements of the process did not offer characterized by somatic and physiological much useful information about what helped pa- symptoms, may have been more consistent with tients get better and, it could be argued, were interpersonal and cognitive–behavioral models misleading in the absence of a fine-grained anal- of intervention. These two approaches have tra- ysis of process correlates of outcome. Fortu- ditionally been more symptom-focused than psy- nately, we also identified process correlates of chodynamic approaches. In addition, the level of outcome irrespective of theoretical orientations activity of the therapists may have been due to and brand names. Positive outcome was most patients’ inability to put their emotions into robustly predicted by a focus on identifying and words, resulting in therapists’ appropriate active expressing (particularly negative) emotion and exploration and questioning in order to assist feelings. The link between focusing on emotion with this specific task. Finally, the fixed duration and positive outcome that emerged in this study of the study combined with clinicians’ awareness resonates to varying degrees with different theo- of the study objective to gauge pre/post treatment retical perspectives. Shear’s Emotion Focused change, may have influenced clinicians to behave Treatment (EFT) (1995) for panic disorder rests in a more structuring and active fashion than they on the notion that poorly articulated emotionality would have otherwise in a non time-limited dy- fuels the sense of helplessness underlying panic namic treatment. Nonetheless, it is significant disorder. In EFT, clinicians and patients success- that a group of experienced psychodynamic ther- fully explore unexplained emotional reactions by apists, conducting therapy without constraints identifying response patterns (shame, guilt, fear imposed by a research protocol on their approach, and anger) to situations eliciting panic attacks. provided a highly effective treatment by fostering Clarifying emotional reactions and identifying a process most characterized as cognitive– possible negative emotional triggers is consid- behavioral. Given that the process findings sug- ered therapeutic because it decreases patients’ gest that the treatments could not be character- sense of helplessness and increases their sense of ized accurately as psychodynamic treatment of self-efficacy and mastery. panic disorder, it might be more accurate to de- The important role of emotion/feeling in panic scribe them as “treatment as usual” in the disorder is also present in Milrod’s delineation of community. unidentified anger as a primary dynamic under-

227 Ablon, Levy, and Katzenstein lying the disorder. According to Milrod’s concep- predicated on the identification and exploration tualization, patients’ fear of their angry feelings of emotional reactions specifically. The link to and the conscious and unconscious fantasies this negative outcome could also be explained if pa- spawns can trigger attacks (Milrod, Busch, Coo- tients used cognitive themes as a way of intellec- per, & Shapiro, 1997). Milrod emphasizes that tualizing and defending against troubling the expression of anger in many patients with thoughts and feelings. Support for this hypothesis panic disorder fosters feelings of abandonment, exists at a descriptive level as evidenced by the the internal experience of losing the love of the fact that clinicians assigned over half of the pa- object or loss of the capacity to evoke the positive tients in the sample to the level of defensive image of the object at whom anger was directed. functioning characterized by compromise forma- Additionally, some patients can become quite tion (including intellectualization and isolation of guilt-ridden and experience retribution anxiety affect). Items reflecting the suggestive influence reactive to the expression of anger at important of the therapist also tended to be negatively cor- objects or those representing important objects. related with outcome. Patients feeling conflicted about dependence on The results of this study illustrate the value of the therapist and discussing sexual feelings and being able to examine therapeutic process on termination were also all associated with positive both a macro (mean adherence to prototypes outcome in our analysis. Milrod has hypothesized across the sample) as well as a micro (item by that core dynamics for panic disorder patients item analyses of individual components contrib- often involve fears of separation and fear of ex- uting to process) level. To make the point meta- pressing and pursuing sexual desires. phorically, in order to better identify better the The role of emotion/feelings in panic disorder salient characteristics of the specimen on the has also been considered from cognitive– slide (i.e., what kind of therapy is actually being behavioral perspectives. Barlow (2000) suggests employed), the resolving power on the micro- that panic patients often constrict the range of scope must be turned higher. The ultimate goal their emotions in an attempt to evade the experi- underlying process-outcome research is to iden- ence of threatening somatic symptoms. The em- tify aspects of the psychotherapeutic process con- pirical connection between alexithymia (charac- tributing to positive and negative outcomes. The terized by difficulty recognizing and verbalizing results of this study suggest that effective treat- feeling, a paucity of fantasy life, concrete speech ment for panic disorder focuses on helping pa- and thought closely tied to external events) and tients to recognize, experience and express neg- panic disorder supports this assertion. ative emotions, sexual desires, and fears of As would be expected, items reflecting com- dependency and separation all in the context of a mon factors contributing to positive therapeutic supportive working alliance in the absence of a alliance (e.g., patient being introspective, patient prescriptive stance from the therapist. committed to work, patient understanding expectations of therapy, therapist being supportive, Limitations therapist accurately perceiving process, and therapist being sensitive to patients feelings) were The results of this study must be considered in associated with positive outcome. This trend is light of the methodological limitations that are consistent with literature highlighting the impor- involved when conducting naturalistic research. tant connection between therapeutic alliance and This pilot study was an open design without positive outcome across a range of different psy- random assignment to a comparison group. chotherapies (Frieswyck et al., 1986; Krupnick, Causal statements about the relationship between Sotsky, Elkin, Watkins, & Pilkonis, 1996). the intervention and outcome, therefore, cannot That focusing on cognitive themes emerged as be made because this study utilized a correla- a negative process correlate of outcome deserves tional design. In addition, future studies of natu- consideration. Shear’s emotion-centered frame- ralistic psychotherapy for panic disorder could work offers a potential explanation for this trend. include a more even distribution of the number of Cognitive themes may be too far removed from clinicians per patient, an increased number of the more proximate role of emotions to be asso- patients in the sample, and a greater diversity of ciated with positive outcome. In Shear’s emotion patients both in terms of ethnicity and gender. focused conceptualization, therapeutic gains are The generalizability of the findings is certainly

228 Beyond Brand Names of Psychotherapy limited by the sample size and the demographic in clinical practice, we studied what experienced characteristics of the sample. Our sample con- clinicians did when left to their own devices to sisted of 89% women and 78% Caucasian pa- treat patients as they normally do outside re- tients. Clearly, these methods could be expanded search protocol. We learned that clinicians were to assess such treatment with a more heteroge- highly effective but that their self-identified pri- nous sample. Also, although consistent with the mary orientation was not necessarily reflective of literature, the fact that the effect sizes of clini- the predominant therapeutic processes they fos- cians’ ratings of improvement far surpassed those tered with their patients. We also learned that the of patients or independent raters deserves further most predominant process was not necessarily exploration. Finally, some critics of the prototype the most important one. In other words, without method using the PQS have doubted the ability of careful examination of process at multiple levels, the 100 Q-set items to comprehensively assess we could have erroneously labeled the treatment process from certain theoretical perspectives as psychodynamic (due to the clinician’s orienta- (e.g., Markowitz, 2003), despite the fact that tions) or cognitive–behavioral (due to the most measure has continued to successfully identify prominent features of process). Furthermore, nei- significant predictors of outcome across a range ther label would have contributed much to our of different treatment modalities. understanding of how to help patients improve. Much like diagnoses, which in many instances do Implications and Future Directions not guide treatment, brand names of orientations and therapies leave much to be desired when it The treatments described in this study appear comes to understanding what specifically pro- to represent an effective form of what might best motes therapeutic change and likely often con- be described as “treatment as usual” in the com- tribute to faulty assumptions about why patients munity. Although the term “treatment as usual” got better. itself seems to connote a degraded version of In this study, by focusing on an atheoretical more effective treatment offered somewhere else analysis of process, we were able to identify than in the community, our results suggest that empirically derived change processes. These when clinicians are left to their own devices to change processes are ones that could be used to treat complicated patients, they can produce ex- build effective and clinically relevant treatments cellent results! A plausible next step would be to specifically for panic disorder patients. Another use the design of a randomized, controlled clini- unique advantage afforded by reversing the typ- cal trial to compare ESTs for panic disorder (such ical steps involved in testing treatments is that as Panic Control Therapy) to viable alternative treatments constructed around empirically vali- treatments (such as Milrod’s psychodynamic psy- dated change processes practiced by experienced chotherapy) and treatment as usual (such as the clinicians would be embraced by practitioners to treatments provided in this study). Such a com- a greater degree than manualized treatments de- prehensive study would provide more clarity as veloped by researchers in laboratories. Thus, the to the true relative efficacy of various forms of sizable obstacle of generalization to practitioners’ treatment for panic disorder. offices would not be the issue it is for manualized Although the standard procedure for evaluat- treatments. This study is an example of a research ing treatments is to test them as packages of paradigm that is a useful complement to random- intervention under controlled conditions before ized, controlled trials of ESTs. It exemplifies how “exporting” them to clinical practice outside the process researchers and experienced clinicians laboratory, the results of this study also suggest can learn from each other by empirically validat- the intriguing possibility that the reverse may ing change processes in naturalistic treatments. make more sense. Our results demonstrate the utility of studying what clinicians are actually References doing in clinical practice that is associated with patient change before building treatments around ABEND, S. (1989). Psychoanalytic psychotherapy. In C. these specific components and testing them under Lindemann (Ed.), Handbook of phobia therapy: Rapid symptom relief in anxiety disorders (pp. 393–403). controlled conditions. Having assumed that clini- Northvale, NJ: Jason Aronson. cians might learn how to conduct effective treat- ABLON, J. S., & JONES, E. E. (1998). How expert clini- ments through experience with hypothesis testing cians’ prototypes of an ideal treatment correlate with

229 Ablon, Levy, and Katzenstein

outcome in psychodynamic and cognitive-behavioral GABBARD, G. O. (1990). Psychodynamic psychiatry in therapy. Psychotherapy Research, 8, 71–83. clinical practice. Washington, DC: APA Press. ABLON, J. S., & JONES, E. E. (1999). Psychotherapy pro- GARFIELD, S. L. (1998). Some comments on empirically cess in the NIMH Collaborative Research Program. supported treatments. Journal of Consulting and Clin- Journal of Consulting and Clinical Psychology, 67, 64– ical Psychology, 66, 121–125. 75. GOLDFRIED, M. R., & WOLFE, B. E. (1996). Psychother- ABLON, J. S., & JONES, E. E. (2002). Validity of con- apy practice and research: Repairing a strained alliance. trolled clinical trials of psychotherapy: Findings from American Psychologist, 51, 1007–1016. the NIMH treatment of depression collaborative re- GUY, W. (1976). Assessment manual for psychopharma- search program. American Journal of Psychiatry, 159, cology. Washington, DC: U.S. Government Printing 775–783. Office. ABLON, J. S., & MARCI, C. S. (2004). Psychotherapy HOWARD, K. I., MORAS, K., BRILL, P. B., MARTINOVICH, process: The missing link: Comment on Westen, No- Z., & LUTZ, W. (1996). Evaluation of psychotherapy: votny, and Thompson-Brenner. Psychological Bulletin, Efficacy, effectiveness, and patient progress. American 130, 664–668. Psychologist, 51, 1059–1064. BARLOW, R.H. (2000.) Unraveling the mysteries of anx- JACOBSON, N. S., & TRUAX, P. (1991). Clinical signifi- iety and its disorders from the perspective of emotion cance: A statistical approach to defining meaninfgul theory. American Psychologist, 55, 1247–1263. change in psychotherapy research. Journal of Consult- BARLOW, D. H., CRASKE, N. G., CERNY, J. A., & KLOSKO, ing and Clinical Psychology, 59, 12–19. J. S. (1989). Behavioral treatment of panic disorder. JONES, E. E. (2000). Therapeutic action. Northvale, NJ: Behavior Therapy, 20, 261–282. Jason Aronson, Inc. BARLOW, D. H., GORMAN, J. M., & SHEAR, M. K. (1997). JONES, E. E., CUMMING, J. D., & HOROWITZ, M. J. (1988). Results from the multi-center comparative treatment Another look at the nonspecific hypothesis of therapeu- study of panic disorder: Acute and maintenance out- tic effectiveness. Journal of Consulting and Clinical come. Canterbury, England: British Association of Be- Psychology, 56, 48–55. havioral and Cognitive Psychotherapy. JONES, E. E., HALL, S., & PARKE, L. A. (1991). The COHEN, J. (1988). Statistical power analysis for the behav- process of change: The Berkeley Psychotherapy Re- ioral sciences. Hillsdale, NJ: Erlbaum. search Group. In L. Beutler & M. Crago (Eds.), Psy- CRASKE, M. G., & BARLOW, D. H. (2001). Panic Disorder chotherapy research: An international review of pro- and Agoraphobia. In D. H. Barlow (Ed.), Clinical grammatic studies. Washington, DC: American Handbook of Psychological Disorders: A Step-by-Step Psychological Association. Treatment Manual (pp. 1–60). New York: The Guilford JONES, E. E., KRUPNICK, J. H., & KERIG, P. K. (1987). Press. Some gender effects in a brief psychotherapy. Psycho- CRITS-CHRISTOPH, P., & CONNOLLY GIBBONS, M. B. therapy, 24, 336–352. (2003). Research developments on the alliance in psy- JONES, E. E., & PULOS, S. M. (1993). Comparing the chodynamic psychotherapy. Psychoanalytic Inquiry, 23, process in psychodynamic and cognitive-behavioral 332–349. therapies. Journal of Consulting and Clinical Psychol- DEROGATIS,L.R.(1994).SymptomChecklist-90-R ogy, 61, 306–316. (SCL-90-R): Administration, scoring, and proce- KLEIN, D. F., ZITRIN, C. M., WOERNER, M. G., & ROSS, dures manual. Minneapolis, MN: National Computer D. C. (1983). Treatment of phobias: II. Behavior ther- Systems. apy and supportive psychotherapy: Are there any spe- DEROGATIS, L. R., LIPMAN, R. S., RICKELS, K., UHLEN- cific ingredients? Archives of General Psychiatry, 40, HUTH, E. H., & COVI, L. (1974). The Hopkins symptom 139–145. checklist: A self-report symptom inventory. Behavioral KRUPNICK, J. H., SOTSKY, S. M., ELKIN, I., WATKINS, J., & Science, 19, 1–15. PILKONIS, P. A. (1996). The role of therapeutic alliance ENDICOTT, J., NEE, J., HARRISON, W., & BLUMENTHAL, R. in psychotherapy and pharmacotherapy outcome: Find- (1993). Quality of life enjoyment and satisfaction ques- ings in the National Institute of Mental Health Treat- tionnaire: A new measure. Psychopharmacology Bulle- ment of Depression Collaborative Research Program. tin, 29, 321–326. Journal of Consulting and Clinical Psychology, 64, 532– ENDICOTT, J., SPITZER, R. L., FLEISS, J. L., & COHEN, J. 539. (1976). The Global Assessment Scale: A procedure for LAMBERT, M. J., & HILL, C. E. (1994). Assessing psycho- measuring overall severity of psychiatric disturbance. therapy outcomes and processes. In A. E. Bergin & Archives of General Psychiatry, 33, 761–771. S. L. Garfield (Eds.), Handbook of psychotherapy and EUDELL-SIMMONS, E. N., STEIN, M. B., DEFIFE, J. A., & behavior change (pp. 72–113). New York: Wiley. HILSENROTH, M. J. (2005). Reliability and validity of MARKOWITZ, J. C. (2003). Letter to the Ed.: Controlled the Social Cognition and Object Relations Scale trials of psychotherapy. American Journal of Psychia- (SCORS) in the assessment of dream narratives. Jour- try, 160, 186–187. nal of Personality Assessment, 85, 325–333. MARKOWITZ, J. S., WEISSMAN, M. M., & OUELLETTE, R. FRIESWYCK, S. H., ALLEN, J. G., COLSON, D. B., COYNE, (1989). Quality of life in panic disorder. Archives of L., GABBARD, G. O., HOROWITZ, L., ET AL. (1986). General Psychiatry, 46, 984–992. Therapeutic alliance: Its place as a process and out- MILROD, B., & BUSCH, F. (1996). Long-term outcome of come variable in dynamic psychotherapy research. panic disorder treatment: A review of the literature. Journal of Consulting and Clinical Psychology, 54, 32– Journal of Nervous and Mental Diseases, 184, 723–730. 38. MILROD, B., BUSCH, F., COOPER, A., & SHAPIRO, T.

230 Beyond Brand Names of Psychotherapy

(1997). Manual of Panic-Focused Psychodynamic Psy- Multicenter collaborative panic disorder severity scale. chotherapy. Washington, DC: American Psychiatric American Journal of Psychiatry, 154, 1571–1575. Press, Inc. SHEAR, M. K., CLOITRE, M., & HECKELMAN, L. (1995). MILROD, B., BUSCH, F., LEON, A. C., ARONSON, A., Emotion-focused treatment for panic disorder: A brief, ROIPHE, J., RUDDEN, M., ET AL. (2001). A pilot open dynamically informed therapy. In J. P. Barber & P. trial of brief psychodynamic psychotherapy for panic Crits-Christoph (Eds.), Dynamic therapies for psychiat- disorder. Journal of Psychotherapy Practice Research, ric disorders (p. 460). New York: Basic Books, Inc. 10, 239–244. SHEAR, M. K., & MASER, J. D. (1994). Standardized MILROD, B., & SHEAR, M. K. (1991). Dynamic treatment assessment for panic disorder research: A conference of panic disorder: Three case histories. Hospital and report. Archives of General Psychiatry, 51, 346–354. Community Psychiatry, 42, 311–312. SHEAR, M. K., PILKONIS, P. A., CLOITRE, M., & LEON, A. ORLINSKY, D. E., & HOWARD, K. I. (1986). Process and C. (1994). Cognitive behavioral treatment compared outcome in psychotherapy. In A. E. Garﬁeld & Bergin with nonprescriptive treatment of panic disorder. Ar- (Ed.), Handbook of psychotherapy and behavior chives of General Psychiatry, 51, 395–401. change, 3rd ed. New York: Wiley. SHEAR, M. K., SHOLOMSKAS, D. E., BARLOW, D. H., GORMAN, J. M., WOODS, S. M., & CLOITRE, M. (1992). OTTO, M. W., POLLACK, M. H., PENAVA, S. J., & ZUCKER, Panic Disorder Severity Scale (PDSS). Pittsburgh, PA: B. G. (1999). Group cognitive-behavior therapy for Department of Psychiatry, University of Pittsburgh patients failing to respond to pharmacotherapy for School of Medicine. panic disorder: A clinical case series. Behavior Research SIFNEOS, P. E. (1972). Short-term psychotherapy and emo- and Therapy, 37, 763–770. tional crisis. Cambridge, MA: Harvard University PERSONS, J. B. (1991). Psychotherapy outcome studies do Press. not accurately represent current models of psychother- SWENSON, R. P., COX, B. J., & WOSZEZY, C. B. (1992). apy: A proposed remedy. American Psychologist, 46, Use of medical services and treatment for panic disor- 99–106. der with agoraphobia and for social phobia. Canadian PETERSON, R. A., & REISS, S. (1993). Anxiety Sensitivity Medical Journal, 147, 878–883. Index Revised test manual. Worthington, OH: IDS Pub- WESTEN, D., NOVOTNY, C. M., & THOMPSON-BRENNER, lishing Corporation. H. (2004). The empirical status of empirically sup- REISS, S., PETERSON, R. A., GURSKY, D. M., & MCNALLY, ported psychotherapies: Assumptions, ﬁndings, and re- R. J. (1986). Anxiety sensitivity, anxiety frequency and porting in controlled clinical trials. Psychological Bul- the prediction of fearfulness. Behavior Research and letin, 130, 631–663. Therapy, 24, 1–8. WIBORG, I. M., & DAHL, A. A. (1996). Does brief dy- SHEAR, M. K., BROWN, T. A., BARLOW, D. H., MONEY, namic psychotherapy reduce the relapse rate of panic R., SHOLOMSKAS, D. E., WOODS, S. W., et al. (1997). disorder? Archives of General Psychiatry, 53, 689–694.

231