Quality of Life, Mood, and Sexual Function: a Path Analytic Model of Treatment Effects in Men with Erectile Dysfunction and Depressive Symptoms

Home , Erectile dysfunction

Quality of life, mood, and sexual function: a path analytic model of treatment effects in men with erectile dysfunction and depressive symptoms

RC Rosen1*, SN Seidman2, MA Menza1, R Shabsigh2, SP Roose3, LJ Tseng4, J Orazem4 and RL Siegel4

1University of Medicine and Dentistry, Robert Wood Johnson Medical School, Piscataway, New Jersey, USA; 2Columbia Presbyterian Medical Center, New York, New York, USA; 3New York State Psychiatric Institute, New York, New York, USA; and 4Pfizer Inc., New York, New York, USA

Erectile dysfunction (ED) is commonly associated with depressed mood and diminished quality of life (QoL), but few studies have investigated the causal associations involved. Therefore, we evaluated the correlation between several measures of mood, QoL, and sexual function in a retrospective analysis of a sample of depressed men (n ¼ 152), with ED enrolled in a clinical trial of sildenafil citrate (VIAGRAs). Strong correlations were observed at baseline among measures of erectile function (EF), mood, and overall QoL. Significant treatment effects were observed on all three domains, with significant interactions between changes in mood and QoL. Based on multiple regression and path analysis, a model was developed in which EF changes were associated with improved mood and quality of sexual life, which resulted in improved partner satisfaction, family life, and overall life satisfaction. These data suggest that QoL changes associated with ED therapy may be mediated by changes in sexual function, mood, and family relationships. International Journal of Impotence Research (2004) 16, 334–340. doi:10.1038/sj.ijir.3901197 Publication online 12 February 2004

Keywords: pharmacologic studies in sexual function; oral vasoactive agents

Introduction tion.6,7 Some data also suggest that ED worsens psychological adaptation to comorbid medical conditions; in a study of 1460 diabetic men, those with In the United States, 10 to 20 million men are comorbid ED had lower QoL, significantly higher estimated to have some degree of erectile dysfunction levels of diabetes-specific health distress, worse (ED), and it is well established that ED is associated psychological adaptation to and acceptance of dia- with a wide range of psychosocial problems, such as betes, and a less satisfactory sexual life.8 Moreover, decreased quality of life (QoL) and self-esteem and an these men were more easily frustrated and discour- increased incidence of depression and interpersonal aged by their diabetes, which could lead to worse 1–4 relationship problems. Numerous clinical and epi- metabolic control. demiological studies have demonstrated that ED has General health and QoL can be negatively affected a strong negative impact on QoL; for example, Jønler by the breakdown of a significant personal relation- 5 et al have shown that patients with loss of erectile ship, which is one of the most stressful life events. A function (EF) within the past year had significantly UK-based survey demonstrated that men with ED lower QoL than men without ED. Similarly, in other and multiple risk factors had poorer QoL than men studies, men with a complaint of ED had poorer QoL with ED and no risk factors; in addition, up to 28% than age-matched men from the general popula- of these men believed that their sexual dysfunction was directly responsible for the termination of their last relationship.7 One limitation of these two studies7,8 is the fact that, although a correlation *Correspondence: RC Rosen, PhD, University of Medicine between QoL and ED was demonstrated, the effects and Dentistry, Robert Wood Johnson Medical School, Department of Psychiatry, 675 Hoes Lane, Piscataway, NJ of ED treatment on QoL were not examined. 08554-5653, USA. Indeed, whereas most studies have demonstrated E-mail: [email protected] a clear relation between ED and QoL, the strength of Received 4 December 2002; revised 20 February 2003; association has varied significantly from one study accepted 30 March 2003 to another, partially because of the broad range of Quality of life, mood, and sexual function RC Rosen et al 335 QoL measures used. QoL has not been measured in criteria), and 58% were treatment nonresponsive. a consistent way, studies have not controlled for Of those men defined as responders, 73% received potential confounding effects of mood or partner sildenafil compared with 14% receiving placebo. relationships, and the nature of the causal relation Hamilton Depression Rating Scale (HDRS) and Beck has not been systematically examined in any of Depression Index (BDI) scores were significantly these studies. This is the first study to investigate improved among treatment-responsive subjects, ir- the specific mechanisms by which ED treatment respective of whether they received sildenafil or leads to improved QoL. placebo. These effects were comparable with the Epidemiological studies have demonstrated a highly beneficial mood effects usually observed in trials of significant association between ED and depression, antidepressant therapy.14 For purposes of the pre- irrespective of age, marital status, or comorbid- sent study, we evaluated the relation among sexual ities.2,3,9–11 For example, in the Massachusetts Male function, depression, and QoL before and after Aging Study,9 patients with depression had a 1.8- treatment in these same patients using a variety of times higher chance of concomitant ED than patients statistical procedures. The overall goal was to without depression, whereas in another study evalu- develop a conceptual model of the causal relation ating men with ED or benign prostatic hyperplasia, among these variables and to test the strength of this men with ED were 2.6 times more likely to report model by means of path analytical procedures. depressive symptoms.2 Similarly, a history of depression was significantly associated with the prevalence of ED in a Brazilian study of sexual behavior.11 Material and methods However, the coexistence between ED and depression is complex, and the causal relation remains unclear. The psychosocial distress that often accom- Study design panies ED may cause depressive illness in suscep- tible individuals, or ED can be one of the symptoms This was a retrospective analysis of a multicenter, 12- of a major depressive disorder, which is associated week, randomized, double-blind, placebo-controlled, with decreased libido, diminished EF, and decreased 12 12 flexible-dose study. The study was conducted at 20 sexual activity. In addition, the use of antidepres- urology clinics in the United States, each of which sants has been linked to problems with ejaculation, 13–15 had a dual specialty team consisting of a urologist orgasm, desire, and erections. Furthermore, a and a psychiatrist. Men with a primary complaint of subgroup of men with major depressive disorder ED completed the Center for Epidemiologic Studies developed a reversible loss of nocturnal penile 16,17 Depression (CES-D) Scale and were offered an tumescence, suggesting that depression can di- appointment with the study psychiatrist if their rectly affect the erectile process. Finally, hormonal CES-D Scale score was Z16, the standard screening changes in the aging male, such as a reduction of threshold for depressive illness. Institutional review circulating testosterone, or cardiovascular disease, boards at each center reviewed and approved the may cause both conditions, or alternatively, both protocol, and all men gave written informed consent. conditions can coexist but be etiologically unrelated. Men with ED and mild-to-moderate depression were Both major depressive disorder and milder de- randomly assigned to receive sildenafil 50 mg or pressive syndromes are characterized by a pervasive matching placebo. This dose could be adjusted to 100 low mood and/or loss of interest in daily activities, or 25 mg, based on efficacy and tolerability. Drug was and little is known about the possible relation 18 to be taken approximately 1 h before sexual activity between mood disturbances and QoL. Few studies but not more than once daily. At baseline (week 0) on the relation between ED and QoL have controlled and week 12, subjects were assessed by a psychiatrist for the potential mediating effects of mood and other using the HDRS. In addition, a number of self-report variables. Thus, it is possible that ED may impact instruments were completed. QoL indirectly through the effects of mood or other psychosocial mediators, such as partner or family relationships. Study subjects Numerous clinical trials have demonstrated that ED can be successfully treated in men with comorbid depression, whether or not they are taking The inclusion criteria for this study were as follows: concomitant antidepressant medications, resulting men aged 18 y or older in a stable relationship with in a subsequent improvement in QoL.19–23 For a female partner for Z6 months; a documented example, a recent study by our group showed strong diagnosisofEDforZ6 months; a diagnosis of effects of ED therapy (drug or placebo) on concomi- depressive disorder not otherwise specified, based tant mood and QoL changes.23 Of 152 men on an abbreviated Structured Clinical Interview from diagnosed with ED and mild depression and the Diagnostic and Statistical Manual of Mental randomized to sildenafil or placebo, 42% were Disorders, Fourth Edition;24 and an HDRS25 score classified as treatment responsive (ie, met defined Z12 at the first and second screening interview.

International Journal of Impotence Research Quality of life, mood, and sexual function RC Rosen et al 336 The exclusion criteria were as follows: presence (GEQs), with age, smoking status, duration of ED, of another current axis I psychiatric disorder, etiology of ED, and baseline score as covariates. including substance abuse or dependence; current Least-squares mean scores were adjusted for the use of nitrates or antidepressant medication; abnor- value of covariates. mal serum hormone levels (prolactin, testosterone, A series of multiple regression analyses that thyroid); and a history of major hematologic, renal, or controlled for age, duration of ED, and treatment hepatic abnormalities, uncontrolled diabetes, retini- group were performed on the change scores to tis pigmentosa, or serious cardiovascular disease. determine the direct and mediational linkages among those variables that constitute the path analysis model. Based on results of previous studies, Study measures it was hypothesized that the mood variable was mediating the relation between EF and QoL. Finally, a causal model was tested and developed Beck Depression Inventory:26 The BDI is a 21-item using path analysis.30 Causal modeling, using either questionnaire rated on a scale of 0 (minimal) to 3 path analysis or structural equation modeling, is a (severe). conceptual and statistical approach for examining International Index of Erectile Function (IIEF):27 the relation between key predictor variables (eg, age, Individual questions from the IIEF were scored on a duration of illness) and the mediating or intervening scale of 1 (worst response) to 5 (best response), with effect on either variables (eg, relationship satisfac- 0 indicating no sexual activity. The 15 questions are tion, depression) or more distal and remote out- divided into five different domains: EF (Q1–5, 15), comes (overall life satisfaction). Our analysis was intercourse satisfaction (Q6–8), orgasmic function performed using observable (manifest) variables (Q9–10), sexual desire (Q11–12), and overall satis- only, that is, without the inclusion of latent (non- faction (Q13, 14). observable) variables in the model. The relation (or Global Efficacy Questions: ‘Did treatment improve path) between changes in QoL Q1–4 was examined your erections?’ and ‘Did treatment improve your first, and then the paths were expanded by adding ability to have sexual intercourse?’ A priori criteria the remaining variables one by one until the whole for identifying treatment-responsive individuals path model was completed. In the regression were a response of ‘yes’ to both GEQs and a score analysis, the dependent variable was the criterion of Z22 on the EF domain of the IIEF at the end of variable led by all the independent variables (pre- treatment, which indicates no or minimal ED.28 dictors or mediators). Each path relation was based Life Satisfaction Checklist (LSC):29 The eight-item on the beta coefficient of the corresponding inde- LSC is a measure of life satisfaction that includes pendent variable in the regression equation. A questions on life as a whole (Q1), sexual life (Q2) mediational relation occurred when a predictor led and partner relationships (Q3), social relationships to the mediator and the mediator led to the criterion with family (Q4) and friends (Q5), and satisfaction variable in a significant manner, as tested using with leisure life (Q6), vocation (Q7), and finances Sobel’s t-ratio (Figure 1).31,32 The model developed (Q8). Categorical responses range from 1 (‘very dissatisfying’) to 6 (‘very satisfying’).

Statistical analyses

Statistical analyses were carried out in a series of stages, beginning with the construction of a correla- tional matrix for the relation between variables at baseline and end of treatment. Initially, separate Pearson’s correlation coefficients were calculated to examine the associations among QoL (as determined using questions from the LSC), the EF domain, and mood measurements (as determined using the BDI) at baseline and end of treatment (week 12), and for change from baseline. As a result, six highly associated variables (change from baseline) were selected for further analyses: QoL Q1–4, the EF domain score, and the BDI (mood) total score. Subsequently, the effects of sildenafil treatment or placebo on EF (IIEF, GEQ), mood (BDI), and QoL (LSC) measures were examined using analysis of covariance (IIEF, BDI, LSC) or logistic regression Figure 1 Mediational model.

International Journal of Impotence Research Quality of life, mood, and sexual function RC Rosen et al 337 here is intended to provide a conceptual framework Treatment effects on correlations of ED, mood, and for further studies evaluating the effects of ED and life satisfaction ED treatments on QoL.

Marked treatment effects were observed after 12 weeks on all three parameters, with significant Results interactions between change scores for mood and life satisfaction (Pearson’s coefficient r ¼ 0.68), Baseline correlations among ED, mood, and QoL mood and EF (r ¼ 0.48), and EF and life satisfaction (r ¼ 0.37; Po0.0001 for all 3; Table 2). Additional correlations of note following treatment A previous report established the safety and efficacy were those between partner relationship and of treating ED in men with concomitant mild- sexual life satisfaction (r ¼ 0.43; Po0.0001), to-moderate depression.12 A significant improve- and partner relationship and mood (r ¼ 0.48; ment in ED was associated with significant Po0.0001). improvement in depression, regardless of treatment (ie, sildenafil or placebo). Similarly, significant effects were noted in treatment responders com- Change score correlations and multivariate analyses pared with nonresponders in overall life satisfaction (LSC Q1), sexual life (LSC Q2), and partner relationship (LSC Q4; Table 1). There were significant Path analysis of change scores following treatment correlations at baseline between measures of mood demonstrated that EF changes were associated with and overall satisfaction (Pearson’s coefficient improved mood (Po0.0001). and quality of sexual r ¼ 0.62; Po0.0001) and between mood and EF life (Q2; Po0.0001), which led to improved partner (r ¼ 0.22; Po0.0001), and there was a trend for a satisfaction (Q3; Po0.0001), family life (Q4; correlation between EF and overall life satisfaction P ¼ 0.0014), and overall life satisfaction (Q1; (r ¼ 0.16; P ¼ 0.055; Table 2). Ratings of the partner Po0.0001; Figure 2). relationship were highly correlated with family life satisfaction at baseline (r ¼ 0.57; Po0.0001), whereas a nonsignificant trend toward a positive Final path model for change from baseline variables association between partner relationship and sexual life satisfaction were observed at baseline (r ¼ 0.12; P ¼ 0.15). Additional correlations of note at baseline Change in overall life satisfaction (Q1) was the are correlations between mood and sexual QoL criterion variable because it was highly correlated (r ¼ 0.28; P ¼ 0.0004) and mood and partner relation- to other QoL questions. For example, changes in ship (r ¼ 0.34; Po0.0001). Q2 (sexual life), Q3 (partner satisfaction), and Q4

Table 1 Least-squares mean scores (s.e.) at baseline and after 12 weeks of placebo or sildenafil treatment

Baseline Placebo (n ¼ 75) Sildenafil (n ¼ 64) P-value

EF domain 9.32 12.4 (1.2) 23.4 (1.5) o0.0001a OS domain 3.67 4.5 (0.4) 7.8 (0.5) o0.001a GEQ1 (%) — 10.5 91.0 o0.0001a GEQ2 (%) — 11.9 89.5 o0.0001a

Baseline Treatment responders (n ¼ 58) Treatment nonresponders (n ¼ 83) P-value

Q1: Life as a whole 3.61 4.7 (0.1) 3.9 (0.2) 0.0001b Q2: Sexual life 1.77 4.6 (0.2) 2.1 (0.2) 0.0001b Q3: Partnership 4.19 5.0 (0.2) 4.2 (0.2) 0.0001b Q4: Family life 4.19 4.7 (0.2) 4.4 (0.2) 0.0709b BDI (mood) 15.6 6.4 (0.9) 13.7 (1.0) o0.0001b HDRS 16.7 6.4 14.2 o0.001b

EF ¼ erectile function; OS ¼ overall satisfaction; GEQ ¼ global efficacy question; HDRS ¼ Hamilton Depression Rating Scale; BDI ¼ Beck Depression Inventory.26 a For sildenafil vs placebo. b For treatment responders vs nonresponders, based on an analysis of covariance adjusted for baseline values, except for GEQ1/2 (logistic regression).12 EF domain ¼ Q1–5 and 15 from the IIEF, score range 1–30; OS domain ¼ Q13 and Q14 from the IIEF, score range 2–10; Q1– Q4 are from the Life Satisfaction Checklist.29

International Journal of Impotence Research Quality of life, mood, and sexual function RC Rosen et al 338 Table 2 Pearson’s correlation coefficients at baseline and after 12 weeks

Q1 Q2 Q3 Q4 EF BDI OS HDRS

Q1 Baseline 1 0.286 0.489 0.670 0.157 0.623 0.311 0.384 Overall quality of life 12 weeks 1 0.485 0.615 0.635 0.367 0.682 0.480 0.594

Q2 Baseline 0.286 1 0.117 0.126 0.480 0.283 0.644 0.122 Sexual quality of life 12 weeks 0.485 1 0.435 0.263 0.780 0.574 0.871 0.702

Q3 Baseline 0.489 0.117 1 0.570 À0.010 0.342 0.284 0.136 Partnership relation 12 weeks 0.615 0.435 1 0.702 0.320 0.484 0.508 0.460

Q4 Baseline 0.670 0.126 0.570 1 0.006 0.482 0.163 0.310 Family life 12 weeks 0.635 0.263 0.702 1 0.141 0.455 0.334 0.362

EF Baseline 0.157 0.480 À0.010 0.006 1 0.214 0.386 0.082 Erectile function 12 weeks 0.367 0.780 0.320 0.141 1 0.482 0.813 0.587

BDI Baseline 0.623 0.283 0.342 0.482 0.214 1 0.336 0.450 Mood 12 weeks 0.682 0.574 0.484 0.455 0.482 1 0.528 0.675

OS Baseline 0.311 0.644 0.284 0.163 0.386 0.336 1 0.171 Overall satisfaction 12 weeks 0.480 0.871 0.508 0.334 0.813 0.528 1 0.680

HDRS Baseline 0.384 0.122 0.136 0.310 0.082 0.450 0.171 1 Depression rating 12 weeks 0.594 0.702 0.460 0.362 0.587 0.675 0.680 1

EF ¼ erectile function; OS ¼ overall satisfaction; GEQ ¼ global efficacy question; HDRS ¼ Hamilton Depression Rating Scale; BDI ¼ Beck Depression Inventory. All Pearson’s correlation coefficients shown in bold are significant at Po0.05 EF domain ¼ Q1–5 and 15 from the IIEF, score range 1–30; OS domain ¼ Q13 and Q14 from the IIEF, score range 2–10 Q1–Q4 are from the Life Satisfaction Checklist.

Figure 2 Final path model for change from baseline variables

(family life) were significantly linked to change in satisfaction with family life were related directly Q1 (overall satisfaction) when regressed separately. to changes in overall satisfaction and also mediated In addition, changes in sexual life were linked by partner satisfaction (t ¼ 2.22, P ¼ 0.013). Changes significantly to changes in partner satisfaction. in mood directly precipitated both changes in sexual Sobel’s t-ratio further indicated that changes in life and family life satisfaction, and furthermore partner satisfaction mediated the relation between mediated the pathway from change in EF domain overall satisfaction and satisfaction with sexual score to change in sexual life satisfaction (t ¼ 3.74, life (t ¼ 2.10, P ¼ 0.018). Similarly, changes in Po0.0001; Figure 2).

International Journal of Impotence Research Quality of life, mood, and sexual function RC Rosen et al 339 Discussion patients with little or no mood disturbance. Alter- natively, because depression is a strong correlate of ED overall,2,4,18 the model presented here may be Increasing evidence from multiple sources has applicable to a significant proportion of the overall shown a deleterious effect of ED on life satisfaction population of patients with ED. and overall QoL.4,5,29 Despite the lack of standardi- We developed our model based solely on a zation and adequate disease-specific, health-related retrospective analysis of data from a previously QoL measures, in addition to design and methodo- published study.23 The robustness or predictive logical weaknesses in most studies, patients with ED validity of the model remains to be demonstrated of broad-spectrum etiology consistently show re- in further prospective studies. Our model predicts duced life satisfaction compared with age-matched that patients who show less improvement in mood, controls.20,22 These effects have also been observed or whose improvements in sexual function are not in patients with ED and spinal cord injury,33 associated with positive changes in family or depression,23 and multiple sclerosis.34 In the cur- partner relationships, would be less likely to show rent study, we were able to demonstrate a strong overall life satisfaction changes following treatment. negative association at baseline between EF and Conversely, our model implies that ED treatments overall life satisfaction in patients with ED and that combine couples psychotherapy with pharma- mild-to-moderate or subsyndromal depression. ED cological interventions might lead to even greater was also strongly correlated with BDI score. Addi- improvements in life satisfaction as a result of tional baseline correlations showed that satisfaction improved interpersonal relationships. These hy- in the partner relationship was strongly related to potheses remain to be evaluated in further prospec- family life satisfaction and mood, and that sexual tive studies. life satisfaction and mood were strongly correlated. Use of this model could help identify and explain Conversely, beneficial effects of ED therapies on psychosocial outcomes of ED therapy. By consider- health-related QoL could be explained by several ing the mechanisms involved, clinicians can opti- mechanisms. For example, changes in life satisfac- mize the overall clinical impact of ED treatment. tion might be a consequence of improved partner relationships; positive changes in sexual confidence, self-concept, or mood; or other nonspecific effects of treatment. Using multiple regression and path References analytic procedures in the current study, we tested the validity of a new theoretical model to account for 1 Litwin MS, Nied RJ, Dhanani N. Health-related quality of life this relation in patients with ED and mild-to- in men with erectile dysfunction. J Gen Intern Med 1998; 13: moderate depression. Based on this model, we 159–166. hypothesized that patients whose mood and overall 2 Shabsigh R et al. Increased incidence of depressive symptoms in men with erectile dysfunction. Urology 1998; 52: 848–852. sexual functioning are improved following ED 3 Laumann EO, Paik A, Rosen RC. Sexual dysfunction in the treatment are likely to experience greater satisfaction United States: prevalence and predictors. JAMA 1999; 281: in their partner relationship and family life, in 537–544. general, and that these changes will be associated 4 Althof SE. Quality of life and erectile dysfunction. Urology 2002; 59: 803–810. with positive changes in overall life satisfaction. We 5 Jønler M et al. The effect of age, ethnicity and geographical have shown that both mood changes and changes location on impotence and quality of life. Br J Urol 1995; 75: in partner relationship and family life satisfaction 651–655. functioning are predictive of improvements in over- 6 Litwin MS. Health related quality of life in older men without all life satisfaction. The model also implies that prostate cancer. J Urol 1999; 161: 1180–1184. 7 Guest JF, Das Gupta R. Health-related quality of life in a UK- overall life satisfaction can be conceptualized as the based population of men with erectile dysfunction. Pharma- end point or distal consequence of a cascade of coeconomics 2002; 20: 109–117. intervening psychological or interpersonal benefits 8 De Berardis G et al. Erectile dysfunction and quality of life in of ED treatment, such as improved overall sexual type 2 diabetic patients: a serious problem too often over- looked. Diabetes Care 2002; 25: 284–291. satisfaction, partner, and other family relationships. 9 Feldman HA et al. Impotence and its medical and psychoso- Some limitations of the current study warrant cial correlates: results of the Massachusetts Male Aging Study. consideration. First, the inclusion criteria required J Urol 1994; 151: 54–61. that all patients had to have mild-to-moderate 10 Araujo AB et al. The relationship between depressive depression in addition to ED. Given the presence of symptoms and male erectile dysfunction: cross-sectional results from the Massachusetts Male Aging Study. Psychosom disturbed mood in these patients before treatment, Med 1998; 60: 458–465. and the significant effects of treatment on depression 11 Moreira ED et al. Prevalence and correlates of erectile ratings in this study, it is not surprising that mood dysfunction: results of the Brazilian study of sexual behavior. was a significant mediating variable.23 Theoretically, Urology 2001; 58: 583–588. 12 Seidman SN et al. Treatment of erectile dysfunction in men this relation might be predicted to be even stronger with depressive symptoms: results of a placebo-controlled in patients with higher levels of mood disturbance trial with sildenafil citrate. Am J Psychiatry 2001; 158: (eg, major depression) and possibly weaker in 1623–1630.

International Journal of Impotence Research Quality of life, mood, and sexual function RC Rosen et al 340 13 Rosen RC, Lane RM, Menza M. Effects of SSRIs on sexual Edition, (SCID-I/P, Version 2:0). Biometrics Research Depart- function: a critical review. J Clin Psychopharmacol 1999; 19: ment, New York State Psychiatric Institute: New York, 1995. 67–85. 25 Hamilton MA. A rating scale for depression. J Neurol 14 Seidman SN, Roose SP. The relationship between depression Neurosurg Psychiatry 1960; 23: 56–62. and erectile dysfunction. Curr Psychiatry Rep 2000; 2: 26 Beck AT, Epstein N, Brown G, Steer RA. Psychometric 201–205. properties of the Beck Depression Inventory: twenty-five years 15 Nurnberg HG et al. Efficacy of sildenafil citrate for the later. Clin Psychol Rev 1988; 8: 77–100. treatment of erectile dysfunction in men taking serotonin 27 Rosen RC et al. The International Index of Erectile Function reuptake inhibitors. Am J Psychiatry 2001; 158: 1926–1928. (IIEF): a multidimensional scale for assessment of erectile 16 Roose SP, Glassman AH, Walsh BT, Cullen K. Reversible loss dysfunction. Urology 1997; 49: 822–830. of nocturnal penile tumescence during depression: a pre- 28 Cappelleri JC et al. Diagnostic evaluation of the erectile liminary report. Neuropsychobiology 1982; 8: 284–288. function domain of the International Index of Erectile 17 Nofzinger EA et al. Sexual function in depressed men. Function. Urology 1999; 54: 346–351. Assessment by self-report, behavioral, and nocturnal penile 29 Fugl-Meyer AR, Lodnert G, Branholm IB, Fugl-Meyer KS. On tumescence measures before and after treatment with cogni- life satisfaction in male erectile dysfunction. Int J Impot Res tive behavior therapy. Arch Gen Psychiatry 1993; 50: 24–30. 1997; 9: 141–148. 18 Seidman SN. Exploring the relationship between depression 30 Retherford R, Choe MK. Path analysis. In: Statistical Methods and erectile dysfunction in aging men. J Clin Psychiatry 2002; for Causal Analysis. John Wiley & Sons: New York, 1993, pp 63(Suppl 5): 5–12, discussion 23–25. 93–99. 19 Willke RJ et al. Quality of life effects of alprostadil therapy for 31 Sobel M. Asymptotic confidence intervals for indirect effects erectile dysfunction. J Urol 1997; 157: 2124–2128. in structural equation models. In: Leinhard S (ed). Socio- 20 Williams G et al. The effect of transurethral alprostadil on the logical Methodology. Jossey-Bass: San Francisco, 1982, pp quality of life of men with erectile dysfunction, and their 290–312. partners. MUSE Study Group. Br J Urol 1998; 82: 847–854. 32 Sobel M. Some results on indirect effects and their standard 21 Muller MJ, Benkert O. Lower self-reported depression in errors in covariance structure models. In: Tuma NB (ed). patients with erectile dysfunction after treatment with Sociological Methodology. American Sociological Society: sildenafil. J Affect Disord 2001; 66: 255–261. Washington, DC, 1986, pp 158–186. 22 Paige NM et al. Improvement in emotional well-being and 33 Hultling C et al. Quality of life in patients with spinal relationships of users of sildenafil. J Urol 2001; 166: 1774– cord injury receiving Viagra (sildenafil citrate) for the 1778. treatment of erectile dysfunction. Spinal Cord 2000; 38: 23 Seidman SN, Roose RP. Sexual dysfunction and depression. 363–370. Curr Psychiatry Rep 2001; 3: 202–208. 34 Miller J, Fowler C, Sharief M. Effect of sildenafil citrate 24 First MB, Spitzer RL, Gibbon M, Williams JBW. Structured (Viagra) on quality of life in men with erectile dysfunction and Clinical Interview for DSM IV Axis I Disorders—Patient multiple sclerosis. Ann Neurol 1999; 46: 496.

International Journal of Impotence Research