- Top of page
- SUBJECTS AND METHODS
- CONFLICT OF INTEREST
Although premature ejaculation (PE) has broad-ranging effects on both men and their partners [1,2], there is currently no simple instrument that is easy to use and interpret in clinical practice. PE is commonly diagnosed using the definition provided by the Diagnostic and Statistical Manual of Mental Disorders, fourth edition, text revision (DSM-IV-TR)  or similar criteria proposed by other professional or academic societies [2,4–8]. However, these professional organizations do not specify measurable outcomes that are meaningful measures of improvement or change in the condition from the patient’s perspective. For example, the second International Consultation defines PE according to three essential criteria: brief ejaculatory latency, loss of control, and psychological distress in the patient and/or partner , but does not provide guidance on how to assess these measures.
The assessment of PE in clinical trials has relied heavily upon intravaginal ejaculatory latency time (IELT), defined as the time between vaginal intromission and intravaginal ejaculation . However, latency time alone may not be sufficient for evaluating PE or its treatment, because the perception of latency time is not consistent across all men , and the impact of PE is multidimensional . In general, optimal patient-reported outcome (PRO) measures for clinical trials are brief, capture relevant concepts of a condition, have known validity and reliability, are responsive to treatment, and are easy to administer and interpret [10,11]. Several recent studies have used single-item PRO measures for evaluating PE and its treatment [1,12–14]. Although indices that add up to overall scores have been developed for PE [15,16], individual measures may be advantageous in that they are simpler and faster to administer, and easily interpretable response options (e.g. ‘good’) facilitate interpretation of the specific construct that they are intended to assess. Further, scoring an array of measures as a profile allows visualization and evaluation of the pattern of scores among the measures, which may be valuable in determining whether different subgroups of men with PE have different patterns of scores, or if patterns change differentially over time.
The Premature Ejaculation Profile (PEP) is a new instrument for the assessment of PE that deals with all domains of the condition as defined by the DSM-IV-TR: perceived control over ejaculation, personal distress related to ejaculation, and interpersonal difficulty related to ejaculation, as well as satisfaction with sexual intercourse, which is also felt to be important by experts in the field of sexual medicine . Each of these is assessed using a single item; responses are rated on a five-point scale; and higher scores indicate better functioning (Table 1) [13,14]. The measures of the PEP have been used in various studies of men with PE, including observational studies and clinical trials of treatments for PE [1,12,18], and have also been used to characterize the differences between men with and without PE [13,14,19]. Furthermore, the relationships between individual measures of the PEP have been evaluated using path analysis [18,19].
Table 1. Single-item outcome measures
|Measure||Concept||Question||Scores and response options|
|PEP||Perceived control over ejaculation Satisfaction with sexual intercourse||Over the past month, was your control over ejaculation during sexual intercourse*: Over the past month, was your satisfaction with sexual intercourse:||0: Very poor 1: Poor 2: Fair 3: Good 4: Very good|
|Personal distress related to ejaculation† Interpersonal difficulty related to ejaculation†||How distressed are you by how fast you ejaculate (come) during sexual (vaginal) intercourse?* To what extent does how fast you ejaculate (come) during sexual (vaginal) intercourse cause difficulty in your relationship with your partner?*||0: Extremely 1: Quite a bit 2: Moderately 3: A little bit 4: Not at all|
|PGI|| ||Compared to prior to the study, would you describe your condition as:||−3: Much worse −2: Worse −1: Slightly worse 0: No change 1: Slightly better 2: Better 3: Much better|
In this manuscript, approaches to using and scoring the PEP are proposed and validated. Depending on the specific clinical research objective, one can focus on the individual measures, use the PEP as a profile, or use an overall index score. Preliminary data on the measurement properties of the individual measures, in the form of reliability and known-groups validity, have been published elsewhere . The present report is a compilation of the validation evidence of all forms of the PEP and includes an assessment of the ability to detect an improvement in the PE condition.
SUBJECTS AND METHODS
- Top of page
- SUBJECTS AND METHODS
- CONFLICT OF INTEREST
Data were obtained from two large, community-based, observational studies of men and their partners (one in the USA (1587 men) , and the other in five European countries (1115 men) ), and a placebo-controlled phase III clinical trial (1238 men)  that evaluated the efficacy and safety of dapoxetine, a short-acting selective serotonin reuptake inhibitor (SSRI) in development for the treatment of PE. In the USA observational study, men were diagnosed as having PE by clinicians experienced or trained in using the DSM-IV-TR criteria. In the European observational study and the phase III study, men were diagnosed as having PE by clinicians experienced or trained in using the DSM-IV-TR criteria and by self-report of at least moderate personal distress and/or interpersonal difficulty related to ejaculation in response to the appropriate questions from the PEP. In the European study, diagnoses were not disclosed to subjects until after all study procedures were completed. Importantly, the PEP profile was not used to establish the PE diagnosis in any study.
In all studies, subjects were required to be aged ≥18 years, heterosexual, and in an ongoing, stable, monogamous, sexual relationship for at least 6 months. Subjects and their partners agreed to engage in sexual intercourse ≥2 times weekly in the observational studies. In the phase III study, couples agreed to engage in sexual intercourse ≥4 times during a 2-week baseline period and ≥6 times/month for the remainder of the study (9 weeks).
In all studies, subjects were ineligible if they had a physical or mental condition that interfered with typical sexual functioning, or if they or their partner reported a form of sexual dysfunction other than PE (including erectile dysfunction). Medications that may affect ejaculatory function were disallowed, including SSRI antidepressants, tricyclic antidepressants, phosphodiesterase type-5 inhibitors, and topical treatments for the prolongation of latency time.
All study protocols were reviewed and approved by the Institutional Review Boards (IRBs) of the participating study centres in accordance with the USA Code of Federal Regulations and IRB policies, or regional Ethics Committees outside the USA, and each study was conducted according to the guidelines of Good Clinical Practice.
Detailed methods for both observational studies have been published elsewhere [1,18]. Briefly, in both observational studies, men and their partners participated in three study visits: screening, initial assessment, and reassessment; in the USA study, these visits occurred at 0, 2, and 4 weeks, respectively, while in the European study, these visits occurred at 0, 4, and 8 weeks, respectively. The PEP was assessed at all visits. Subjects also reported their patient-reported global impression of change (PGI) in perceived control over ejaculation and satisfaction with sexual intercourse from the previous visit.
Detailed methods for the phase III clinical trial are reported elsewhere . Briefly, men with PE were randomized to receive placebo, dapoxetine 60 mg as-needed (prn), or dapoxetine 60 mg daily (qd) for 9 weeks. The PEP measures were reported at baseline and on Days 28 and 63 or study endpoint. The PGI, a measure of subject-perceived change in condition, was assessed on Days 28 and 63 or study endpoint (Table 1). Data from the trial were used to assess the sensitivity of the PEP items to change over time (related to treatment).
Only subjects with responses to all four PEP items at the same time point were included in these analyses.
The test-retest reliability of each PEP measure was calculated using data reported at the initial assessment and reassessment visits in the USA and European observational studies using the intraclass correlation coefficient (ICC) , which is a measure of consistency of the data between two time points (i.e. between Week 4 and Week 8) if there is no change expected in the condition (i.e. the subjects received no treatment). The ICCs were calculated using data from ‘stable’ men, who reported no change in their condition based on the following criteria: (i) for perceived control over ejaculation, stable subjects were those who reported ‘no change’ on a question assessing global impression of change in perceived control over ejaculation; (ii) for satisfaction with sexual intercourse, stable subjects were those who reported ‘no change’ on a question assessing global impression of change in satisfaction with sexual intercourse; (iii) for the PEP index, and personal distress and interpersonal difficulty related to ejaculation, stable subjects were those who reported a change in IELT of <30 s. Analyses from published studies of dapoxetine  showed that subjects who reported ‘no change’ in their condition with treatment using the PGI had a mean increase in IELT of 0.39 min, suggesting that a change in IELT of <30 s indicates no change in the condition over time.
Known-groups validity, or the ability of the measure to discriminate between two identifiable groups, was evaluated using data reported at the initial assessment visit in the USA and European observational studies. It was hypothesized that men with a diagnosis of PE would report poorer perceived control over ejaculation, lower satisfaction with sexual intercourse, and greater personal distress and interpersonal difficulty related to ejaculation than men without PE. These hypotheses were evaluated using a two-sided, independent-samples t-test. It was further hypothesized that compared with men without PE, men with shorter IELTs would also report poorer perceived control over ejaculation, lower satisfaction with sexual intercourse, and greater personal distress and interpersonal difficulty related to ejaculation. These hypotheses were evaluated by pooling IELTs into tertile subgroups (0–2 min, >2–5 min, and >5 min); independent sample t-tests were conducted to compare adjacent tertiles.
The ability of the PEP measures to detect change in the condition was assessed by evaluating the relationship between PGI and the mean change from baseline in scores for each of the PEP measures in the dapoxetine phase III clinical trial. Very few subjects reported a worsening in their condition, so only the levels of ‘no change’, ‘slightly better’, ‘better’, and ‘much better’ were included in the analysis. As the goal of these analyses was not to compare treatment response among treatment groups, but rather to evaluate the relationship between changes in PEP measures and patient-reported improvements in PE, analyses were conducted using all treatment groups combined. Effect sizes for change from baseline at study endpoint were calculated.
Known-groups validity of profiles, which is their ability to discriminate between two identifiable groups, was assessed using the same data sources, and evaluated to determine if the PEP profiles of men with and without PE differed, and whether there was any difference in the PEP profiles of the various IELT tertile subgroups. The ability of the PEP profiles to detect change was assessed by evaluating the differences between profiles at each level of PGI.
Several multivariate anova models were used with the four PEP measures as dependent variables, and independent variables of PE status, or IELT tertile subgroup, or response to PGI as appropriate to the particular analysis. Statistical tests were based on Wilks’ Lambda.
A PEP index score was derived as the average score of the four individual PEP items. The reliability and validity of the PEP index score was evaluated based on ICCs, known-groups validity, and the ability to detect change in PE, as described previously. Several principal components analyses were conducted on data from both observational studies to determine the adequacy of representing the information from the four PEP items as a single-index value. The Kaiser (1960) criterion  of retaining components with an eigenvalue of >1.0 was applied to reduce the dimensionality of the index.
- Top of page
- SUBJECTS AND METHODS
- CONFLICT OF INTEREST
The demographic characteristics of men in these studies have been published in detail elsewhere [1,18,20]. In the USA observational study, 207 men (12.7%) were diagnosed as having PE, and 1380 men (87.3%) did not receive a PE diagnosis ; demographic characteristics were similar between the groups. In the European observational study, 201 men (18.0%) were diagnosed as having PE, and 914 men (82.0%) did not receive a PE diagnosis ; demographic characteristics were similar between the groups. In the phase III clinical trial, of 1238 subjects enrolled, 245, 491, and 502 were randomized to receive 9 weeks of treatment with placebo, dapoxetine prn, or dapoxetine qd, respectively . Across the three treatment groups, 64–68% completed the treatment phase, and demographic characteristics were similar among groups; this high discontinuation rate probably resulted from the high respondent burden of the study, which included six different scales of mood and sexual function, and Holter monitoring. Most of the subjects in all studies were aged 30–40 years and Caucasian, although ≈25% of the subjects in the USA observational study were of other ethnicities (e.g. Black and Hispanic). Adverse events (AEs) during the treatment phase are reported in detail elsewhere . Treatment-emergent AEs were reported by 44.1% and 61.3% of subjects who received placebo and dapoxetine, respectively, in the double-blind treatment phase. The most common AEs were nausea, dizziness, headache, diarrhoea, fatigue, and insomnia, which were more common with dapoxetine than with placebo.
Reliability and validity of PEP measures
In an analysis of data from all men in the USA and European observational studies, for which PE status was expected to remain constant (i.e. there was no intervention), acceptable test-retest reliability was observed for the PEP measures (Table 2); ICCs among men who did not report a change in their condition between visits ranged from 0.66 to 0.74 in the USA observational study  and from 0.66 to 0.83 in the European observational study .
Table 2. Test-retest reliability of the PEP and PEP index score
|USA observational study ||European observational study |
|Perceived control over ejaculation*||0.70||0.66|
|Personal distress related to ejaculation†||0.74||0.70|
|Satisfaction with sexual intercourse‡||0.66||0.83|
|Interpersonal difficulty related to ejaculation†||0.73||0.67|
|PEP index score†||0.80||0.76|
Known-groups validity was assessed based on differences between men with and without PE, and among IELT tertiles. For each of the individual PEP items in both trials, the mean scores of men with PE indicated significantly worse functioning than those of men without PE (P < 0.001 for all; Fig. 1). For all measures of the PEP, lower scores indicate worse functioning. Men with IELTs of 0–2 min reported significantly poorer functioning on all PEP measures in comparison with those whose IELT was >2–5 min, who in turn reported significantly poorer functioning than those whose IELT was >5 min (P < 0.001 for all; Table 3).
Figure 1. Known-groups validity: PEP profiles of men with and without PE in the USA and European observational studies. The mean score of men with and without PE in the USA and European observational studies for each of the four individual PEP items. Each item was significantly lower (P < 0.001 for all) for men with PE in comparison with men without PE, indicating poorer sexual functioning in all four domains.
Download figure to PowerPoint
Table 3. Relationship among PEP measures, PEP index score, and IELT
|Observational study||Mean (sd)|
|Perceived control over ejaculation||Personal distress related to ejaculation||Satisfaction with sexual intercourse||Interpersonal difficulty related to ejaculation||PEP index score|
| A. 0–2 min||1.46 (1.12)||1.78 (1.21)||2.32 (1.16)||2.58 (1.24)||2.03 (1.00)|
| B. >2–5 min||2.41 (0.96)*||2.79 (1.07)*||3.02 (0.87)*||3.42 (0.90)*||2.91 (0.79)*|
| C. >5 min||3.12 (0.75)†||3.40 (0.85)†||3.39 (0.67)†||3.74 (0.65)†||3.41 (0.56)†|
| A. 0–2 min||1.55 (1.11)||2.04 (1.26)||2.13 (1.12)||3.04 (1.02)||2.19 (0.96)|
| B. >2–5 min||2.47 (0.98)*||3.12 (1.04)*||2.93 (0.81)*||3.64 (0.73)*||3.04 (0.74)*|
| C. >5 min||3.08 (0.75)†||3.57 (0.73)†||3.27 (0.65)†||3.90 (0.38)†||3.46 (0.47)†|
Among men in the phase III trial (Table 4), those who reported that their condition was ‘slightly better’, ‘better’, or ‘much better’ on the PGI assessment at the end of the trial also reported improvements from baseline in each of the four individual measures of the PEP. Furthermore, the improvement from baseline in the individual PEP measures was greatest for men who reported the greatest improvement in their PE on the PGI.
Table 4. Ability of PEP measures and PEP index score to detect change in PE*
|Variable||Change (n)||Mean (sd) change from baseline||Effect size|
|Perceived control over ejaculation||No change (327)||0.39 (0.69)|| |
|Slightly better (335)||1.26 (0.77)||1.26|
|Better (226)||1.99 (0.84)||2.32|
|Much better (172)||2.81 (0.96)||3.51|
|Personal distress related to ejaculation||No change (327)||0.40 (0.91)|| |
|Slightly better (335)||1.07 (0.91)||0.74|
|Better (226)||1.77 (0.98)||1.51|
|Much better (172)||2.46 (1.00)||2.26|
|Satisfaction with sexual intercourse||No change (327)||0.14 (0.83)|| |
|Slightly better (335)||0.84 (0.88)||0.84|
|Better (226)||1.57 (0.96)||1.72|
|Much better (172)||2.23 (1.04)||2.52|
|Interpersonal difficulty related to ejaculation||No change (327)||0.28 (0.98)|| |
|Slightly better (335)||0.78 (1.00)||0.51|
|Better (226)||1.33 (1.00)||1.07|
|Much better (172)||1.68 (1.10)||1.43|
|PEP index score||No change (327)||0.30 (0.58)|| |
|Slightly better (335)||0.99 (0.61)||1.19|
|Better (226)||1.66 (0.68)||2.34|
|Much better (172)||2.29 (0.70)||3.43|
Profile scoring allows visualization and evaluation of the pattern of scores among the four measures of the PEP, which may be valuable for determining whether different subgroups of men with PE have different patterns of scores, or if patterns change differentially over time. As was seen with each of the individual PEP measures, the profiles of PEP were significantly different (P < 0.001 based on Wilks’ Lambda) between men with and without PE in both the USA and European observational studies (Fig. 1). Moreover, the profiles of the PEP were significantly different (P < 0.001 for each comparison based on Wilks’ Lambda) between men in the different IELT tertile subgroups.
The shapes of the PEP profiles at study endpoint among men enrolled in the phase III trial are depicted graphically in Fig. 2 by their self-reported improvement in PE (based on the PGI). Notably, there is a lack of intersection among the profile lines, which also increase with each successive level of improvement reported on the PGI. Based on multivariate anova analyses (using Wilks’ Lambda), the PEP profile in patients reporting ‘no change’ in their PE was significantly (P < 0.001) different from that in patients reporting that their PE was ‘slightly better’. Similarly, the PEP profile for subjects reporting their PE was ‘better’ was significantly different (P < 0.001) from those reporting ‘slightly better’, and the profile for men reporting that their PE was ‘much better’ was significantly different (P < 0.001) from those reporting ‘better’.
Figure 2. Relationship between the PEP profile and PGI in PE in the phase III trial. Profile of the mean scores of each of the items of the PEP at each level of PGI. There were significant effects in both elevation and shape (P < 0.001 for both tests based on Wilks’ Lambda). The effect of elevation can be seen from the lack of intersection among the profile lines, and the effect of shape can be seen from the fact that the successive improvements in perceived control over ejaculation with each level of improvement in PGI are of a greater magnitude than for personal distress related to ejaculation and satisfaction with sexual intercourse, which are, in turn, of a greater magnitude than that for interpersonal difficulty related to ejaculation.
Download figure to PowerPoint
Results of the principal components analyses were consistent across the PE and non-PE populations of both observational studies and supported the use of an index score. In all cases, only the first principal component showed an eigenvalue of >1.0, and was retained, accounting for ≈60% of the total variation. Moreover, the four PEP items were weighted almost equally. Thus, a PEP index score, representing general severity of PE, was derived as the average score of the four individual PEP items.
The reliability and validity of the PEP index score was evaluated using methods identical to those used for the individual PEP items. As shown in Tables 2, 3 and 4, and Figs 1 and 2, the PEP index score showed acceptable reliability and validity, and ability to detect change in PE, similar to what was observed for each individual item; as expected, the test-retest reliability of the index was greater than that of the individual items.
- Top of page
- SUBJECTS AND METHODS
- CONFLICT OF INTEREST
Results from these analyses have shown that the PEP (including the individual measures, the index score, and the profile) is a valid and reliable instrument for the assessment of PE. Responses to the four PEP measures were significantly different between men with and without PE, and among varying levels of IELT. Each of the four PEP items provided reproducible results in the absence of a change in the condition, and showed measurable improvement in men who reported an improvement in their PE. Further analyses have shown the validity of the PEP when used either as a profile or using an index score.
The present analyses showed the reliability and validity of three different ways to use the PEP, each of which may suit specific research purposes. First, one may use the four individual measures. With this approach, use of all four measures is not required (e.g. the researcher may only wish to include measures of control and distress if those are the only domains of interest). The second and third approaches require administration of all four measures that may then be scored as either a profile or an index. The advantage of profile scoring is that it allows visualization and evaluation of the pattern of scores among the four measures, which may be of interest, for example, if the researcher is trying to determine if different subgroups of men with PE have different patterns of scores or if the patterns change differentially over time. The advantage of the index score is that it provides a summary of all four domains captured by the PEP. While index scoring obscures specific changes in any one domain, it may be appropriate when a summary score is required or when the research is focused on capturing all four domains without interest in a specific domain. An index is particularly useful for diagnostic purposes, because it allows flexibility in how subjects assign priority to the different domains of the PEP.
Regardless of the approach used to score the PEP, its potential usefulness for practising clinicians is that it captures key elements of PE that are useful in its diagnosis and in monitoring treatment outcomes. Most consensus approaches to the diagnosis of PE, such as application of the criteria found in the DSM-IV-TR , consider it a multidimensional condition characterized by dimensions intrinsic to the condition (such as latency time and perceived control over ejaculation) and dimensions related to the negative impact of the condition (such as personal distress, interpersonal difficulty, and low satisfaction with sexual intercourse). The PEP was designed specifically to capture the domains specified in the DSM-IV-TR definition of PE, which has been commonly used in clinical research. The relevance of these domains to men with PE was evaluated using qualitative methods. The measures of the PEP may be better suited to the assessment of PE than indices that use multiple questions to assess a single construct, as they are brief, capture relevant concepts, are easy to administer and interpret, and, as shown here, have known validity and reliability [10,11]. Other instruments that have been used for the assessment of PE, such as the Golombok-Rust Inventory of Sexual Satisfaction (GRISS) , the Self-Esteem and Relationship (SEAR) questionnaire , and the Index of Premature Ejaculation , employ multiple questions to assess individual domains of PE, leading to a summary or index score that may be cumbersome to patients and time-consuming to clinicians to use during routine clinical practice. In contrast, the usefulness of the PEP lies in its interpretability: the meaning of each PEP item and response is readily apparent to the investigator or physician, as is the PEP profile as a whole.
A key advantage of the PEP is that it has been used in studies of PE. Recent analyses have shown that perceived control over ejaculation and personal distress related to ejaculation are the most influential PEP items in determining PE status based on the DSM-IV-TR criteria , and that all four items are moderately (but significantly) correlated to overall scores from the GRISS and SEAR . In addition, the exploration of relationships among the PEP items through a path analysis (a specific type of regression analysis) showed that the man’s sense of low perceived control over ejaculation mediates whether he also reports personal distress related to ejaculation or low satisfaction with sexual intercourse [18,19]. The slopes of the relationships among the four PEP measures between men with PE and men without PE (Fig. 1), as well as the relationships among the four PEP measures between subjects with and without improvement in their PE in the phase III trial (Fig. 2) are consistent with these findings, in that there was less impairment and improvement in PE for the more distal measures.
Limitations of the PEP include that it was not designed as a diagnostic tool, and has not been evaluated for this purpose. Also, the PEP does not include a measure of ejaculatory latency, which must be considered separately from these results. The primary limitation of the present study was that IELT was not restricted in assigning the diagnosis of PE. The International Society for Sexual Medicine has recently proposed new diagnostic criteria for PE that include an IELT of ≈1 min, a lack of control over ejaculation, and negative consequences resulting from the condition, such as distress . However, this definition did not exist when the present study was conducted, using the best available definition at that time. Because an IELT threshold was not a component of the PE diagnosis in the present study, there are an insufficient number of subjects with an IELT of ≈1 min for evaluation.
In conclusion, these analyses have shown that the PEP measures, the PEP profile, and the PEP index score are valid and reliable measures of PE.