Orally administered sildenafil enhances penile erection in

Transcription

1 Cognitive Changes Predict Continued Recovery of Erectile Functioning Versus Relapse After Discontinuation of Sildenafil Treatment for Male Erectile Dysfunction JACQUES J.D.M. VAN LANKVELD, PHD, MARCEL A. VAN DEN HOUT, PHD, MARK G. SPIGT, MSC, AND GOMMERT A. VAN KOEVERINGE, MD, PHD Objective: To examine whether erectile functioning after termination of sildenafil intake in men with psychogenic ED (erectile dysfunction) can be predicted with psychological measures. Method: The subjects in a nonrandomized controlled trial were 65 heterosexual men with acquired psychogenic ED, aged years. Sildenafil medication was taken as required before sexual activity, up to two times per week. Response to a global end point question ( Did the treatment you took during the study improve your physical response during sexual activity in the last 4 weeks? ) was recorded after 6 weeks of sildenafil use and, subsequently, 6 weeks without medication. Other measures of sexual functioning and cognitive predictor measures were also administered. Results: Of the 65 participants who commenced sildenafil treatment, 37% withdrew from the study before follow-up assessment. At posttreatment, 89% of participants reported that treatment had improved or cured their erectile functioning. At follow-up, 66% of participants had maintained posttreatment gains. Response at follow-up could be predicted (p.001) with 96% sensitivity and 50% specificity by entering changes in sexual self-confidence and the participant s rating of his partner s wish to continue treatment in a logistic regression model. Higher odds for recovery of erectile functioning were found in participants reporting increased sexual self-confidence and the estimation that their partner wanted them to continue sildenafil use. High pretreatment sexual desire was found to further increase the odds for positive responding at follow-up. Conclusions: The present results indicate that continued improvement of erectile functioning is possible after discontinuation of sildenafil use in men with psychogenic ED. Maintenance of gains can be predicted from cognitive changes before medication is withdrawn. Key words: erectile dysfunction, sildenafil, cognition. ED erectile dysfunction; EDITS Erectile Dysfunction Inventory of Treatment Satisfaction; EDITS-P Erectile Dysfunction Inventory of Treatment Satisfaction, Partner Version; GRISS Golombok Rust Inventory of Sexual Satisfaction; IIEF International Index of Erectile Function; MANOVA multivariate analysis of variance; VAS visual analog scales. From the Department of Medical, Clinical, and Experimental Psychology (J.J.D.M.VL., M.A.VDH.) and the Department of General Practice (M.G.S.), University Maastricht, Maastricht, The Netherlands; and the Department of Medical, Clinical, and Experimental Psychology (J.J.D.M.VL) and the Department of Urology (G.A.VK.), Academic Hospital Maastricht, Maastricht, The Netherlands. Address reprint requests to: Jacques van Lankveld, PhD, Department of Medical, Clinical and Experimental Psychology, Academic Hospital Maastricht, PO Box 5800, 6202 AZ Maastricht, The Netherlands. jvla@groupwise.azm.nl Received for publication May 10, 2002; revision received September 12, DOI: /01.PSY E INTRODUCTION Orally administered sildenafil enhances penile erection in men with both organic and psychogenic erectile dysfunction (ED) (1, 2), provided that adequate sexual stimulation is generated. Return of normal sexual functioning in ED patients after temporary use of intracavernous self-injections or oral therapy has been reported (3, 4), but prediction of recovery in individual patients has not yet been addressed. Return of good erections on sexual stimulation, unassisted by pharmacological agents, is certainly not experienced by all men with ED after cessation of treatment. Recently, significant relapse into erectile dysfunction was reported of patients with psychogenic and mixed organic/psychogenic ED who were treated for 16 weeks with open-label sildenafil and who subsequently received sildenafil or placebo in a double-blind study phase (5). Compared with 82% of the sildenafil group members, only 26% of the placebo group members indicated improved erections. Other research suggests the development of psychological dependence on medication in men with ED (6). How could restoration of erectile functionality be explained? Psychogenic ED is commonly considered to result from inadequate information processing (7, 8). Most relevant in the present context is that sexually dysfunctional males are held to have low sexual self-confidence (9). They do not expect to be able to perform sexually in partner interactions. They underestimate their actual level of erection, compared with sexually functional men (10). Their low expectancy of sexual efficacy, once present, may serve to maintain the disorder (8). One would thus expect the efficacy of both psychological and pharmacological ED treatment to be mediated by increases in sexual self-confidence. The sildenafilmediated experience of functional sexual performance may help change the patient s cognitions. The present study aims to examine whether improvement of erectile functioning after termination of sildenafil intake vs. relapse into ED can be predicted by employing measures of the psychological processes that are commonly held to maintain psychogenic ED. METHODS Study Sample Men, aged 18 to 70 years, with a heterosexual relationship of at least 3 months duration, with acquired psychogenic ED were eligible for enrollment. Applicants were included if they reported having morning erections, sleeptime erections, erections on masturbation, or during sexual interaction with their partner (albeit not of sufficient duration or rigidity to have satisfactory sexual intercourse) in the last 6 weeks. Written informed consent of both partners was required. Participants were patients of a university outpatient urology clinic or were recruited through media advertisements, in which a study on the prediction of long-term effects of sildenafil for erectile problems was announced, including cost-free drug treatment during 6 weeks. Applicants were medically examined by their urologist or general practitioner before enrollment. Men were excluded who had significant cardiovascular disease within the previous 6 months, used nitrate-containing medications, Psychosomatic Medicine 65: (2003) /03/ Copyright 2003 by the American Psychosomatic Society 709

2 J.J.D.M. VAN LANKVELD et al. had concomitant illnesses, clinically relevant penile deformations, liver or kidney dysfunction, active peptic ulcer, blood clotting disorder, retinitis pigmentosa, high risk for priapism, used medications known to affect sexual function, had received psychological treatment for erectile dysfunction before entering the study, or when sexual activity was otherwise inadvisable. Patients were excluded if they had used sildenafil or intracavernosal injections more than once before entering the study, other than for diagnostic purposes. Study Design The study comprised: 1) a 2-week treatment-free run-in phase to collect baseline sexual functioning data; 2) a 6-week phase of sildenafil treatment. Medication was taken as required before sexual activity, up to two times per week. The first two of these 6 weeks constituted a dose-setting period. All participants started with a 50-mg dose. As required, this dose could be continued, lowered to 25 mg, or increased to 100 mg for the remaining 4-week period. 3) During the next 6 weeks, participants refrained from using medication, both sildenafil and any other medication for ED. After assessment at the end of this period, study participation ended. The study was approved after review by the hospital s medical ethics board. Assessment Outcome Variables Global End Point To allow comparison of the results with previous work in this area, participants answered a global efficacy question ( Did the treatment you took during the study improve your physical response during sexual activity in the last 4 weeks? ; no/yes, improved, or completely cured) at posttreatment and follow-up. Additional data concerning sexual functioning of the patient and his partner, treatment satisfaction, and satisfaction with relevant life domains were collected to validate global end-point measurement. Patient Diary On a daily basis, participants recorded during the 2-week baseline period, the treatment phase, and the withdrawal/follow-up phase, whether or not they had been sexually active, had experienced difficulty with attaining or maintaining an erection, whether or not their erection was hard enough to enable intercourse, and the extent to which they had been enjoying this sexual activity (1 not at all; 2 somewhat enjoyable; 3 enjoyable; 4 very enjoyable; 5 extremely enjoyable). During the treatment phase, the use of the study medication (yes/no) was recorded every day on the diary data sheet. International Index of Erectile Function (IIEF) This is a 15-item self-report questionnaire measuring sexual functioning. Items each have five or six item-specific answering categories. In case of six answering categories, the first-mentioned response is no sexual activity or did not attempt intercourse. It contains five subscales, derived by principal component analysis, measuring erectile function, orgasmic function, sexual desire, intercourse satisfaction, and overall sexual satisfaction, and has been psychometrically evaluated (11). For use in the present study, it has been translated into Dutch by the investigators. High scores indicate better sexual functioning (15 items, scoring range: 0 5). Erectile Dysfunction Inventory of Treatment Satisfaction (EDITS) This is an 11-item self-report questionnaire measuring satisfaction with medical treatments for erectile dysfunction in the male patient. Questions are answered by choosing one of five response categories tailored to each question and placed in ranked order (eg, very satisfied, somewhat satisfied, neither satisfied nor dissatisfied, somewhat dissatisfied, very dissatisfied ). It has been factor-analytically derived and was tested with respect to its test-retest reliability (12). For use in the present study, it has been translated into Dutch by the investigators. High scores indicate negative evaluations (11 items, scoring range: 1 5). Erectile Dysfunction Inventory of Treatment Satisfaction, Partner Version (EDITS-P) This is a 5-item self-report questionnaire measuring the satisfaction of the female partner with the medical treatments for erectile dysfunction in her male partner. Answering categories are analogous to the Patient Version of the EDITS. The EDITS-P has been factor-analytically derived and was tested with respect to its test-retest reliability (12). For use in the present study, it has been translated into Dutch by the investigators. High scores indicate negative evaluations (5 items, scoring range: 1 5). Golombok Rust Inventory of Sexual Satisfaction (GRISS) The Dutch translation of the GRISS (13) is a 28-item questionnaire with separate forms for males and females to evaluate sexual functioning and sexual satisfaction. Completion requires between 5 and 10 minutes. The factor structure, internal consistency, and stability of the Dutch adaptation have been examined (14) and found satisfactory. The construct validity and predictive validity of the GRISS has been investigated in both male and female sexological patients, in gynecological patients and their male partners (15), and in male urological patients (23). For the present study, male subscales of impotence and premature ejaculation and female subscales of sexual dissatisfaction and anorgasmia were selected. High scores indicate larger sexual problems (28 items, scoring range: 1 5). The Fugl-Meyer Life Satisfaction Checklist (16) is an 8-item self-report instrument to assess life satisfaction across eight different domains of life. For each item, a six-graded scale is used, ranging from very satisfied through very dissatisfied. For the present study, items measuring male and female satisfaction with sexual life and partnership relation were selected. High scores indicate high satisfaction (8 items, scoring range: 1 6). Predictor Variables To measure sexual self-confidence the relevant EDITS item ( How confident has this treatment made you feel about your ability to engage in sexual activity? ) was employed. Answering categories were: much more confident, a little bit more confident, no change in my self-confidence, a little bit less confident, and much less confident. Three visual analog scales (scoring range: 0 100) were administered at baseline, posttreatment, and follow-up, assessing, respectively, probability of occurrence of the individually phrased sexual difficulty during sexual activity (VAS1), severity of its expected negative consequences (VAS2; eg, Then my spouse will doubt if I still find her attractive ), and credibility of an individually phrased proposition combining probability of occurrence and rated severity of negative consequences of the erectile problem (VAS3; eg, I run a large risk of losing my erection during intercourse and my wife will then be very disappointed for not having an orgasm ). Statistical Analysis Independent-samples t tests were used for univariate comparisons at posttreatment and follow-up. Results of evaluation of assumptions for multivariate analysis were satisfactory. Multivariate outliers were deleted when their Mahalanobis distance proved greater than the critical value at after multiple linear regression (17). Associations of various measures with response at posttreatment and follow-up were tested with one-way MANOVA. Wilks (lambda) was used to test significance. Predictive validity of cognitive changes was examined using linear regression, logistic regression, and clinical decision analyses. A constant factor was included in every model. Direct and stepwise hierarchical regression models were calculated with forward entry of variables at each step based on the maximumlikelihood statistic (18). All statistics were tested against RESULTS Participants and Dropout Analysis During a span of 9 months, 101 men applied for participation. Eighty-three went to their general practitioner in response to newspaper ads. Eighteen participants were invited to participate by their general practitioner or urologist without 710 Psychosomatic Medicine 65: (2003)

3 ERECTILE FUNCTION AFTER SILDENAFIL TREATMENT having seen the advertisements. Nineteen applicants were excluded: 1 man with nitrate medication for cardiac disease, 1 homosexual man, 11 men who had previously used sildenafil, 2 men whose partner declined participation, and 4 men with organic etiology of ED. Eighty-one applicants completed baseline questionnaires. Table 1 displays demographic characteristics and data about sexual functioning. Baseline IIEFdata revealed that the experienced dysfunction severity was substantial (5, 19). Between screening and start of treatment, 16 participants withdrew. Sixty-five participants started medication use. Between the start of medication and final assessment, 24 men (37%) left the study: 21 participants withdrew before posttreatment assessment and another 3 participants were lost before follow-up was completed. No demographic differences were found between the men who dropped out after initial screening but before medication start and the men who started treatment. Compared with follow-through participants, dropouts in this phase reported higher credibility of the proposition combining chance of occurrence and severity of consequences on the visual analog scale concerned (VAS3: M completers 81 vs. M dropouts 93; t(26) 2.71; p.05). Baseline data of the men who dropped out before posttreatment assessment were compared with the data of men who completed this phase. Dropouts between medication start and posttreatment assessment were younger than completers (M completers 55.6 vs. M dropouts 49.5; t(63) 2.29; p.05). Four patients who dropped out before posttreatment reported a lack of medication effect without occurrence of adverse events. Other reasons were absence of sexual activity during treatment period, severe marital problems, intercurrent major illness, start of antidepressant medication, and scrotal pain. From baseline diaries, it seemed that occasional good erections enabling intercourse were present in five dropout participants. In seven cases no information on reasons for dropout could be retrieved. Sexual Functioning and Treatment Satisfaction Posttreatment Thirty-nine of 44 men (89%) answered at the global endpoint question that treatment had improved or cured their physical response during sexual interaction (responders) whereas 11% answered not improved (nonresponders). Responders and nonresponders did not differ on demographic characteristics (Table 2). Additional outcome variables, assessed at posttreatment, were entered in MANOVA with response at posttreatment assessment as between-subjects factor. A main effect of posttreatment response was found (F(28,8) 3.730, p.026, , power 0.84). Univariate follow-up tests revealed that, among other things, responders reported stronger overall sexual satisfaction (F(1,35) 4.73, p.037); more enjoyment when sexually active (F(1,35) 7.858, p.008); higher satisfaction with treatment (F(1,35) , p.001); and that their erection felt more natural (F(1,35) , p.000) compared with nonresponders (Table 2). In a separate MANOVA, a marginally significant main effect was also found of differences exhibited by the female partners (F(9,26) 2.037, p.076, , power 0.73). Compared with partners of nonresponders, partners of responders (Table 3) reported higher satisfaction at posttreatment assessment with their partnership relation (F(1,34) 5.098, p.030). Furthermore, they viewed their partner s wish to continue treatment as stronger (F(1,34) 7.731, p.009). Thus, differences with regard to additional measures at posttreatment strongly corroborated global end-point results. Follow-up At follow-up, 27 of 41 (66%) of the participants answered at the global end-point question that their physical reaction during sexual activity was improved or cured (responders) vs. 34% reporting no improvement (nonresponders). All respond- TABLE 1. Demographic and Sexual Characteristics of Men With Psychogenic Erectile Dysfunction (N 81) M SD Range Age (years) Length of relationship (years) Children (N) Education (years) Months worked during last year (N) Sexual diary Sexually active during 2-week baseline period (N: days) Problem with erection when sexually active (% of times) Erection sufficient to enable intercourse when sexually active (% of times) International Index of Erectile Functioning Frequency of erection when sexually active Frequency of erection sufficient for intercourse Frequency of sustained erection during intercourse Erectile functioning during last month % Morning erections (% yes) 78 Nightly erections (% yes) 46 Masturbation erections (% yes) 52 Erections with partner (% yes) 75 Other erections (% yes) 32 Psychosomatic Medicine 65: (2003) 711

4 J.J.D.M. VAN LANKVELD et al. TABLE 2. Demographic, Sexual, Marital, and Other Aspects of Functioning and Treatment Satisfaction at Posttreatment Assessment of Men With Erectile Dysfunction Who Responded (N 39) or Did Not Respond (N 5) With Improved or Cured Erectile Functioning After 6 Weeks of Sildenafil Treatment Nonresponders Responders t p Age (years) 57.4 (5.9) 55.4 (9.5).640 NS Length of relationship (months) (180.4) (158.1) NS Education (years) 10.8 (3.4) 12.1 (4.3).026 NS Children (N) 1.4 (0.9) 1.8 (1.3).000 NS Worked during last year (months) 9.6 (5.4) 6.9 (5.8).388 NS IIEF Erectile function 20.2 (8.3) 21.7 (7.6).095 NS Orgasmic function 6.2 (2.5) 7.4 (3.1).080 NS Sexual desire 7.0 (1.6) 6.9 (1.6).056 NS Intercourse satisfaction 8.8 (2.8) 10.0 (3.9).819 NS Overall sexual satisfaction 5.6 (2.7) 7.8 (1.9) GRISS Impotence 13.6 (2.7) 11.5 (3.1) NS Premature ejaculation 10.6 (5.1) 10.0 (3.0).819 NS Fugl-Meyer Scale Sexual life 3.0 (1.2) 3.7 (1.2) NS Partnership relation 4.8 (1.1) 5.2 (1.3) NS Visual Analog Scales Chance of occurrence 58.6 (25.0) 57.0 (31.0).001 NS Severity of consequences 82.2 (14.2) 71.8 (27.7) NS Credibility a 80.0 (6.9) 69.1 (29.2).545 NS Patient diary Sexually active (% of days) 34.3 (23.2) 27.3 (16.3).544 NS Has problem with erection (% b ) 38.0 (37.0) 20.9 (24.1) NS Erection sufficient for coitus (% b ) 72.5 (48.6) 77.9 (25.8).099 NS Enjoyment rating of sexual activity 2.1 (0.6) 3.2 (0.7) EDITS Satisfied with treatment 3.0 (1.2) 1.4 (0.8) Treatment meets expectations 4.0 (1.7) 2.1 (1.1) Wants to continue treatment 4.2 (0.8) 1.8 (1.0) Treatment easy to do 2.2 (1.6) 1.7 (1.0) NS Satisfied with treatment speed 3.4 (1.8) 1.5 (0.7) Satisfied with treatment length 3.0 (1.4) 1.8 (0.7) More confident in sexual ability 2.6 (0.9) 1.6 (0.8) Partner satisfied with results 3.0 (1.2) 1.6 (0.7) Partner wish to continue treatment 3.0 (0.0) 1.9 (0.8) Natural feeling erection 3.2 (1.5) 1.4 (0.7) Natural stiffness of erection 2.8 (1.1) 2.0 (1.1) NS Sildenafil use Times medication taken (N) 8.4 (2.5) 8.4 (3.7) Dose c % % 25 mg NS 50 mg mg a Credibility rating of an individually phrased proposition combining chance of occurrence of erectile dysfunctioning and the severity of its consequences. b Percentage of times when sexually active. c Sildenafil dose during last 4 weeks of treatment period. ers but one reported the same result at posttreatment, indicating maintenance of gains after terminating medication. Of 14 nonresponders, 11 were responders at posttreatment, implicating relapse after previous improvement. Additional outcome variables were entered in MANOVA with response at follow-up as between-subjects factor. A main effect of follow-up response was found (F(27,6) 6.087, p.016, , power 0.90). Univariate follow-up tests revealed that responders reported, among other things, less erectile difficulties (F(1,32) 4.873, p.035); reduction of premature ejaculation (F(1,32) 3.980, p.055); being more satisfied with their sexual lives (F(1,32) 8.792, p.006); experiencing more enjoyment when sexually active (F(1,32) 5.038, p.032); higher satisfaction with treatment (F(1,32) 6.821, p.014); and the experience that their erection felt more natural (F(1,32) , p.000) as well as more natural stiffness of erection (F(1,32) 4.831, p.035) compared with nonresponders (Table 4). In a separate 712 Psychosomatic Medicine 65: (2003)

5 ERECTILE FUNCTION AFTER SILDENAFIL TREATMENT TABLE 3. Sexual, Marital, and Other Aspects of Functioning and Treatment Satisfaction at Posttreatment Assessment of Partners of Men With Erectile Dysfunction, Who Responded (N 39) or Did Not Respond (N 5) With Improved or Cured Erectile Functioning After 6 Weeks of Sildenafil Treatment Partners of nonresponders Partners of responders F p GRISS Sexual dissatisfaction 9.0 (6.9) 8.7 (3.8).006 NS Anorgasmia 7.5 (4.4) 10.6 (4.3) NS Fugl-Meyer Scale Sexual life 3.8 (1.5) 4.0 (1.1).219 NS Partnership relation 4.3 (2.1) 5.3 (0.6) EDITS-P Satisfied with partner s treatment 2.8 (1.3) 1.8 (1.0) NS Does treatment meet your expectations 3.8 (1.9) 2.3 (1.4) NS Feel more sexually attractive for partner 3.0 (0.0) 2.6 (0.7).830 NS Satisfied with partner s erectile ability 2.8 (1.3) 1.9 (1.0) NS Your partner wishes to continue treatment 3.5 (1.0) 1.9 (1.1) MANOVA, a main effect was also found of differences on partner variables (F(9,28) 2.600, p.025, , power 0.86). Compared with partners of nonresponders, partners of responders (Table 5) reported higher satisfaction with their male partner s treatment (F(1,36) , p.002); that treatment more closely met their expectations (F(1,36) , p.000); and that they were more satisfied with their partner s erectile ability (F(1,36) , p.002). Thus, differences on additional measures at follow-up of both patients and partners strongly corroborated global end-point results. The Association Between Erectile Functioning at Follow-up and Cognitive Changes During Treatment The association between response/nonresponse at follow-up (6 weeks after discontinuation of sildenafil treatment) and cognitive variables assessed at posttreatment was investigated. In further analyses, only data were included of participants who were responders at posttreatment and whose follow-up data were available (N 37). Sexual self-confidence and participants estimations of the chance and severity of erectile failure (visual analog scales) were included in this analysis on theoretical grounds, as put forward in the introduction. The analysis further comprised cognitive variables that were found to significantly differentiate between responders and nonresponders at posttreatment. Thus, one-way MANOVA was performed in which changes between baseline and posttreatment in estimated frequency of occurrence of the erectile problem (VAS1), estimated severity of the consequences of erectile failure (VAS2), rated credibility of a proposition combining chance of occurrence and severity of consequences (VAS3), sexual self-confidence at posttreatment (EDITS), overall sexual satisfaction (IIEF), enjoyment rating of sexual activity (diary), satisfaction with treatment (ED- ITS), rating of the extent to which treatment met expectations (EDITS), wish to continue treatment (EDITS), satisfaction with treatment speed (EDITS), satisfaction with treatment duration (EDITS), patient-estimated satisfaction of partner with treatment (EDITS), patient-rated wish of partner to continue treatment Psychosomatic Medicine 65: (2003) (EDITS), and natural feeling of erection (EDITS) were entered as dependent variables and response at follow-up (yes/no) as an independent variable. A main effect was found (F(14,19) 3.074, p.012, n , power 0.93). Univariate follow-up comparisons revealed that increased sexual self-confidence (F(1,32) 8.961, p.005); enjoyment rating of sexual activity (F(1,32) 5.498, p.025); satisfaction with treatment (F(1,32) 6.410, p.016); patient-estimated satisfaction of partner with treatment (F(1,32) 5.556, p.025); and patient-rated wish of partner to continue treatment (F(1,32) 9.907, p.004) were associated with follow-up response. Sexual self-confidence at posttreatment was improved in participants who were responders at follow-up (M , corresponding with much more confident ) compared with nonresponders (M , corresponding with a little bit more confident ). The rating how much they had enjoyed sexual activity and their satisfaction with treatment were all higher than in nonresponders. Furthermore, they rated their partner s satisfaction with treatment as higher, as well as their partner s wish that they continue sildenafil treatment. Nonparametric bivariate correlations between sexual selfconfidence and visual analog scales had small effect sizes (r value 0.18) and were nonsignificant. Next, one-way MANOVA was performed in which partner s posttreatment cognitive variables that were found to significantly differentiate responders and nonresponders at posttreatment (satisfaction with their partnership relation and female partner s estimation of their male partner s wish to continue treatment) were entered as dependent variables and participant s response at follow-up (yes/no) as an independent variable. No effects were found. Prediction of Sexual Functioning at 6 Weeks After Discontinuation of Sildenafil Treatment Prediction of Recovery vs. Relapse (Global End-point Question) At the first step in logistic regression analysis, prediction from sexual self-confidence was investigated, with response to the global end-point question at follow-up as the dependent 713

6 J.J.D.M. VAN LANKVELD et al. TABLE 4. Demographic, Sexual, Marital, and Other Aspects of Functioning and Treatment Satisfaction at Follow-up Assessment of Men With Erectile Dysfunction Who Responded (N 27) or Did Not Respond (N 14) With Improved or Cured Erectile Functioning 6 Weeks After Discontinuation of Sildenafil Treatment Nonresponders Responders t p Age (years) 56.9 (6.6) 54.9 (10.7).296 NS Length of relationship (months) (170.6) (150.7) NS Education (years) 12.1 (4.8) 12.0 (4.0).077 NS Children (N) 1.7 (0.9) 1.7 (1.4).026 NS Worked during last year (months) 6.4 (5.9) 7.0 (5.9).293 NS IIEF Erectile function 11.5 (6.3) 15.8 (9.6) NS Orgasmic function 4.4 (3.1) 5.8 (3.8).912 NS Sexual desire 5.4 (1.9) 6.2 (1.8).447 NS Intercourse satisfaction 6.4 (3.2) 7.4 (4.4).174 NS Overall sexual satisfaction 5.1 (2.2) 6.8 (2.4) NS GRISS Impotence 14.7 (2.1) 11.6 (3.4) Premature ejaculation 11.8 (3.9) 9.6 (3.7) Fugl-Meyer Scale Sexual life 2.9 (1.2) 4.0 (1.0) Partnership relation 4.9 (1.3) 5.3 (0.8).489 NS Visual Analog Scales Chance of occurrence 67.6 (28.6) 63.8 (31.7).092 NS Severity of consequences 78.6 (17.6) 62.6 (36.1) NS Credibility a 82.8 (15.4) 61.6 (31.6) Patient diary Sexually active (% of days) 23.3 (26.6) 18.6 (12.0).742 NS Has problem with erection (% b ) 79.5 (29.8) 50.1 (42.1) Erection sufficient for coitus (% b ) 42.7 (38.1) 61.9 (40.8) NS Enjoyment rating of sexual activity 2.0 (0.9) 2.8 (1.1) EDITS Satisfied with treatment 3.0 (1.1) 1.9 (1.2) Treatment meets expectations 4.6 (0.5) 2.7 (1.3) Wants to continue treatment 3.6 (1.5) 2.3 (1.3) Treatment easy to do 3.6 (1.1) 1.7 (0.9) Satisfied with treatment speed 3.5 (1.1) 1.7 (0.9) Satisfied with treatment length 3.4 (1.1) 1.7 (1.0) More confident in sexual ability 2.8 (0.9) 1.8 (0.7) Partner satisfied with results 2.8 (1.1) 1.6 (0.8) Partner wish to continue treatment 3.0 (1.0) 2.0 (0.9) Natural feeling erection 3.0 (1.4) 1.6 (0.8) Natural stiffness of erection 2.9 (1.0) 2.2 (1.2) a Credibility rating of an individually phrased proposition combining chance of occurrence of erectile dysfunctioning and the severity of its consequences. b Percentage of times when sexually active. variable. Although a significant logistic model was produced (model 2 (1) 7.467, p.006), clinical decision analysis revealed that specificity of prediction (30%) of follow-up response from individual sexual self-confidence scores was insufficient (Table 6, upper panel). Sensitivity of this model was 100% with positive and negative predictive values of 78% and 100%, respectively. Thus, the odds are that a man with psychogenic ED who reports increased sexual self-confidence after 6 weeks of sildenafil use will demonstrate continued improvement of erectile functioning when medication is discontinued. The low specificity of prediction, however, is unsatisfactory. Further stepwise hierarchical logistic regression analyses were conducted in which pretreatment and posttreatment variables were entered that univariately differentiated between responders and nonresponders at follow-up, with sexual selfconfidence entered at the first step. Adding pretreatment participant s satisfaction with the relationship, which was the only pretreatment measure differentiating between responders and nonresponders at follow-up, did not improve prediction. Next, stepwise hierarchical logistic regression and clinical decision analyses were performed with posttreatment cognitive scores which were found to differentiate between responders and nonresponders at follow-up. At the second step, enjoyment rating of sexual activity, satisfaction with treatment, patient-estimated satisfaction of partner with treatment, and patient-rated wish of partner to continue treatment were entered. A significant regression model was produced (model 2 (1) , p.0000) (Table 6, middle panel). The 714 Psychosomatic Medicine 65: (2003)

7 ERECTILE FUNCTION AFTER SILDENAFIL TREATMENT TABLE 5. Sexual, Marital, and Other Aspects of Functioning and Treatment Satisfaction at Follow-up Assessment of Partners of Responders (N 27) and Nonresponders (N 14) at Follow-up Assessment 6 Weeks After Discontinuation of Sildenafil Treatment Partners of responders Partners of nonresponders F p GRISS Sexual dissatisfaction 8.4 (4.0) 10.4 (4.1) NS Anorgasmia 9.5 (4.1) 11.8 (3.9) NS Fugl-Meyer Scale Sexual life 4.2 (1.1) 3.6 (1.2) NS Partnership relation 5.3 (0.6) 4.9 (1.2) NS EDITS-P Satisfied with partner s treatment 1.8 (1.0) 3.1 (1.4) Does treatment meet your expectations 2.7 (1.3) 4.5 (0.8) Feel more sexually attractive for partner 2.6 (0.9) 2.9 (0.3) NS Satisfied with partner s erectile ability 2.2 (1.1) 3.5 (1.2) Your partner wishes to continue treatment 2.5 (1.3) 3.09 (1.4) NS TABLE 6. Predictive Validity of Posttreatment Cognitive Measures for Continued Improvement of Erectile Functioning 6 Weeks after Discontinuation of Sildenafil Treatment Measure Subscale B Wald p R EDITS (posttreatment) More confident in sexual ability Constant Model df 1 p.006 EDITS (posttreatment) More confident in sexual ability Partner wishes to continue treatment Constant Model df 1 p.0000 EDITS (posttreatment) More confident in sexual ability Partner wishes to continue treatment IIEF (pretreatment) Sexual desire Constant Model df 2 p.0000 B logistic regression coefficient; R partial correlation coefficient between predictor variable and group membership. patient s sexual self-confidence and his estimation of his partner s wish to continue treatment correctly predicted 83% of the response at follow-up. Sensitivity and specificity of this model were 96% and 50%, respectively, yielding a 20% gain in specificity of prediction, with positive and negative predictive values of 83% and 83%, respectively. Thus, by using two cognitive predictor variables, more satisfactory prediction proved possible. A further exploratory logistic regression analysis was conducted, adding pretreatment sexual functioning data. At the first step, increase in sexual self-confidence and the patient s estimation of his partner s wish to continue treatment were entered. At the second step, pretreatment variables of erectile function (IIEF), orgasmic function (IIEF), sexual desire (IIEF), impotence (GRISS), and premature ejaculation (GRISS) were entered. A significant regression model was produced (model 2 (2) , p.0000) (Table 6, lower panel). The patient s posttreatment sexual self-confidence and his estimation of his partner s wish to continue treatment and pretreatment sexual desire correctly predicted 89% of the response at follow-up. Sensitivity and specificity of this model were 92% and 80%, respectively, yielding another 30% gain in specificity of prediction, with positive and negative predictive values of 92% and 80%, respectively. Compared with nonresponders, responders at follow-up were more likely to report that their sexual self-confidence had increased during Psychosomatic Medicine 65: (2003) 715

8 J.J.D.M. VAN LANKVELD et al. sildenafil treatment, that they estimated their partner to want them to continue medication, and to have reported pretreatment that their sexual desire was high despite the presence of ED. Additionally, we examined if response at follow-up merely reflected successful unassisted erectile functioning during medication. The number or proportion of occasions in which participants were sexually active without taking medication did not differ univariately between responders and nonresponders. Moreover, no differences were observed regarding the number of times that the participant s erection proved unproblematic when sexually active during the medication period without having taken sildenafil ( unproblematic erection; nonresponders (mean SD): ; responders: ) or the number of times their erections were sufficient for having intercourse without having taken sildenafil ( sufficient erection; nonresponders: ; responders: ). The former logistic regression analysis, including male and female posttreatment measures, was repeated, with unproblematic erection and sufficient erection added as independent variables at the second step. Adding these variables proved not to add to the regression model and its predictive power. Prediction of Level of Sexual Functioning (Continuous Measures) Continuous measures of sexual functioning at follow-up assessment were selected, which were found to differentiate between responders and nonresponders (as defined by the global end-point question). These measures were subjected to principal components analysis with varimax rotation in order to identify underlying common factors. Follow-up measures of impotence (GRISS), premature ejaculation (GRISS), satisfaction with their sexual lives (Fugl-Meyer), enjoyment rating when sexually active (diary), satisfaction with treatment (ED- ITS), whether their erection felt natural (EDITS), and feeling of natural stiffness of erection (EDITS) were entered in this analysis. Two components were extracted with eigenvalues exceeding unity and explaining 63.5% of the variance. Rotated components loadings are shown in Table 7. The first component was readily interpretable. High scores on this component, called PCA erectile dysfunction, corresponded with more dysfunctional erectile performance (GRISS impotence TABLE 7. Rotated Component Matrix of Follow-up Continuous Measures of Sexual Function Component 1 2 Natural feeling erection.849 Satisfaction with sexual life.753 Impotence.691 Natural stiffness of erection.662 Premature ejaculation.838 Satisfaction with treatment.773 Enjoyment of sexual activity subscale), stronger dissatisfaction with sexual life (Fugl- Meyer), and with less natural feeling and the experience of lower rigidity of sildenafil-assisted erections (both EDITS). The second component eluded easy interpretation. High scores on it corresponded with dissatisfaction with treatment outcome (EDITS), high ratings of enjoyment of sexual activity (diary), and low complaints of premature ejaculation (GRISS). We decided not to use this component for further analysis but to employ the original follow-up variables (satisfaction with treatment, enjoyment of sexual activity) instead. Despite the low ratio of cases against number of predictor variables, linear regression was performed with PCA erectile dysfunction as the dependent variable, and posttreatment sexual self-confidence, posttreatment patient rating of his partner s wish to continue treatment (both ED- ITS), and pretreatment sexual desire (IIEF) were entered as predictor variables. These predictor variables were selected because of their predictive value for the global end-point response at follow-up, as reported in previous paragraphs. A significant regression model was calculated (F(2,29) 9.966, p.001). For the two regression coefficients that differed significantly from zero, 95% confidence intervals were calculated. The confidence intervals for posttreatment sexual self-confidence were and and those for pretreatment sexual desire were and Thus, high scores on PCA erectile dysfunction at follow-up assessment were associated with decreased sexual self-confidence at posttreatment and low pretreatment sexual desire. Linear regression analysis, with enjoyment of sexual activity at follow-up and satisfaction with treatment outcome at follow-up as dependent variables and entering the same predictor variables, did not produce significant regression models. DISCUSSION The effects on erectile functioning with 6 weeks of sildenafil treatment (89% response) replicated earlier work (69 93%) (5, 19, 20). At 6 weeks after discontinuation of medication, 66% of participants reported continued improvement. This contrasts strongly with Christiansen et al. s (5) findings of only 26% of placebo participants reporting continued improvement after sildenafil treatment. Global end points in both studies were practically identical. Sampling differences may explain discrepant findings. Organic etiology in part (60%) of Christiansen et al. s sample could, more than in the present psychogenic group, have necessitated continuation of medication. Results were not specified for both etiological groups in Christiansen et al. s sample. Differences in study design may offer another explanation. Whereas participation in the present study ended after follow-up assessment (6 weeks after termination of sildenafil treatment), all participants in the Christiansen et al. study received a 1-year open-label extension of sildenafil treatment. This might have installed expectancies of continued need for pharmacological support in participants. Increased sexual self-confidence during sildenafil treat- 716 Psychosomatic Medicine 65: (2003)

9 ERECTILE FUNCTION AFTER SILDENAFIL TREATMENT ment, together with the patient s interpretation that his partner wants treatment to be continued, can, to a fair extent, predict maintenance of treatment gains when medication is discontinued. Sexual self-confidence was also found to linearly predict sexual functioning after treatment cessation, as measured by other, continuous variables. These findings are consistent with the self-regulation aspect of the cognitive model of sexual dysfunction (9). When the man expects his genital response to occur, this expectation of sexual self-efficacy will motivate him not only to approach sexual situations with some confidence but also to remain psychologically engaged when his erection will occasionally prove not to develop as he expects. The contribution of the other partner-related cognition to this prediction is less straightforward. Believing that the partner wishes him to continue pharmacological treatment might increase the patient s motivation to remain engaged in sexual activity. Higher pretreatment sexual desire further increases the odds of improved erectile functioning after termination of medication. Higher sexual desire that has remained intact, notwithstanding the liabilities of erectile failure, obviously advances the person s motivation to engage in sexual interaction with a partner. The participant s experience of successful unassisted erectile functioning during the treatment period proved not to explain away the association between cognitive variables and the results at follow-up. In such cases, the increase in self-confidence could have been epiphenomenal and may simply reflect the success rate of unassisted erections, making the prediction of unassisted erectile functioning at follow-up tautological. Although it would not make much difference for clinical purposes, it would reduce the significance of the present findings for the cognitive model of sexual dysfunction. It was found that changes in sexual self-confidence at posttreatment assessment were associated with response at follow-up whereas changes in the patient s estimation of the probability of occurrence of erectile failure, the severity of its consequences, and the credibility of a proposition in which probability and problem severity in his own words were combined (as reflected in his scores on the visual analog scales) were not. The insignificant, low correlations between sexual self-confidence and visual analog scales suggest an absence of overlap between the constructs tapped by these measures. This may have been associated with different frames of reference that were evoked by the wording of the respective items. The answering categories of the sexual selfconfidence item ( much more confident, a little bit more confident, no change in my self-confidence, a little bit less confident, much less confident ) inherently instigated comparisons of the participant s present state with his previous state of self-confidence. The wording of the visual analog scales referred only to the present state. Although the researchers subsequently calculated difference scores between pre- and posttreatment administrations, these difference scores may not reflect participant s self-awareness of change in these domains. Such self-awareness of an increased ability to engage in sexual activity is hypothesized by the self-regulation model of sexual dysfunction (9) to be crucial to functional sexual performance. Limitations of the present study have to be acknowledged with regard to the stability and generalizability of the results. Thirty-seven percent of participants withdrew between medication start and final assessment. Although dropout rates of this magnitude are not uncommon in comparable studies (5, 21) of pharmacological treatment of ED, the possibility remains that results of the present study are based on a nongeneral sample of patients. The relatively small sample size, although found adequate to test the research questions of the present study, also constrains the generalizability of the results. Regression models tend to fit the data of the sample from which they are derived better than any other sample (22). Thus, the stability of prediction using the model generated from the present derivation sample awaits cross-validation in other groups of patients (17). The present results not only attest to the possibility of continued erectile improvement after discontinuation of medication but, even more interesting, show that this enduring gain can be predicted from cognitive changes before medication is withdrawn. Physicians who prescribe medication for psychogenic ED might ask patients to return after some time to discuss their progress. If patients report increased sexual self-confidence and hold their partner to evaluate treatment positively, and even more so when they reported substantial sexual desire before the start of treatment, the physician might suggest that they discontinue medication. Moreover, health insurance companies might be less hesitant in providing reimbursement for temporary sildenafil prescriptions. REFERENCES 1. Conti CR, Pepine CJ, Sweeney M. Efficacy and safety of sildenafil citrate in the treatment of erectile dysfunction in patients with ischemic heart disease. Am J Cardiol 1999;83:29C 34C. 2. Rendell MS, Rajfer J, Wicker PA, Smith MD. Sildenafil for treatment of erectile dysfunction in men with diabetes: a randomized controlled trial. JAMA 1999;281: Rowland DL, Boedhoe HS, Dohle G, Slob AK. Intracavernosal selfinjection therapy in men with erectile dysfunction: satisfaction and attrition in 119 patients. Int J Impot Res 1999;11: Sharlip ID. Does natural erectile function improve following intracavernous injections of vasoactive drugs? Int J Impot Res 1997;9: Christiansen E, Guirguis WR, Cox D, Osterloh IH. Long-term efficacy and safety of oral Viagra (sildenafil citrate) in men with erectile dysfunction and the effect of randomised treatment withdrawal. Int J Impot Res 2000;12: Cranston-Cuebas MA, Barlow DH, Mitchell W, Athanasiou R. Differential effects of a misattribution manipulation on sexually functional and dysfunctional men. J Abnorm Psychol 1993;102: Barlow DH. Causes of sexual dysfunction: the role of anxiety and cognitive interference. J Consult Clin Psychol 1986;54: Cranston-Cuebas MA, Barlow DH. Cognitive and affective contributions to sexual functioning. Annu Rev Sex Res 1990: Sbrocco T, Barlow DH. Conceptualizing the cognitive component of sexual arousal: implications for sexuality research and treatment. In: Salkovskis PM, editor. Frontiers of cognitive therapy. New York: The Guilford Press; 1996: Sakheim DK, Barlow DH, Abrahamson DJ, Beck JG. Distinguishing between organogenic and psychogenic erectile dysfunction. Behav Res Ther 1987;25: Rosen RC, Riley A, Wagner G, Osterloh IH, Kirkpatrick J, Mishra A. The Psychosomatic Medicine 65: (2003) 717

10 J.J.D.M. VAN LANKVELD et al. international index of erectile function (IIEF): a multidimensional scale for assessment of erectile dysfunction. Urology 1997;49: Althof SE, Corty EW, Levine SB, Levine F, Burnett AL, McVary K, Stecher V, Seftel AD. The EDITS: the development of questionnaires for evaluating satisfaction with treatments for erectile dysfunction. Urology 1999;53: Rust J, Golombok S. The GRISS: a psychometric instrument for the assessment of sexual dysfunction. Arch Sex Behav 1986;15: Ter Kuile MM, Van Lankveld JJDM, Kalkhoven P, van Egmond M. The Golombok Rust Inventory of Sexual Satisfaction (GRISS): psychometric properties within a Dutch population. J Sex Marital Ther 1999;25: Van Lankveld JJDM, Ter Kuile MM. The Golombok Rust Inventory of Sexual Satisfaction (GRISS): predictive validity and construct validity in a Dutch population. Pers Individ Dif 1999;26: Fugl-Meyer AR, Lodnert G, Bränholm IB, Fugl-Meyer KS. On life satisfaction in male erectile dysfunction. Int J Impot Res 1997;9: Tabachnick BG, Fidell LS. Using multivariate statistics. New York: HarperCollins; Hosmer DW, Lemeshow S. Applied logistic regression. New York: John Wiley and Sons; Goldstein I, Lue TF, Padma-Nathan H, Rosen RC, Steers WD, Wicker PA. Oral sildenafil in the treatment of erectile dysfunction. N Engl J Med 1998;338: Padma-Nathan H, Steers WD, Wicker PA. Efficacy and safety of oral sildenafil in the treatment of erectile dysfunction: a double-blind, placebo-controlled study of 329 patients. Int J Clin Pract 1998;52: Paige NM, Hays RD, Litwin MS, Rajfer J, Shapiro MF. Improvement in emotional well-being and relationships of users of sildenafil. J Urol 2001;166: Demaris A. Regression models. In: Wiederman MW, Whitley BE Jr, editors. Handbook for conducting research on human sexuality. Vol. 12. Mahwah (NJ): LEA; 2002: van Lankveld J JDM, van Koeveringe GA. Predictive validity of the Golombok Rust Inventory of Sexual Satisfaction (GRISS) for the presence of sexual dysfunctions within a Dutch urological population. Int J Impot Res 2003;15(2): Psychosomatic Medicine 65: (2003)