Both the 5-item short version of the International Index of Erectile Function (IIEF-5) and the Expanded Prostate Cancer Index Composite (EPIC) have been used to assess erectile function. In this study, the authors compared various definitions of potency according to the IIEF-5 and the EPIC.
Patients with prostate cancer who had completed the IIEF-5 and the EPIC within 7 days of each other were included. The Spearman correlation coefficient (rho) was calculated to assess the relation between IIEF-5 and EPIC sexual domain scores. Concordance of potency rates by IIEF-5 and EPIC was assessed in cross-tabulations. By calculating the area under the receiver operator characteristics (ROC) curve (AUC), the authors ascertained the discriminative ability of the IIEF-5 score to identify potent men as defined by the EPIC.
Analyzing 102 questionnaire pairs, IIEF-5 and EPIC domain scores were found to be highly correlated (rho = 0.776). EPIC sexual domain scores ≥60 had high concordance with IIEF-5 scores ≥17 (98%) and with nearly all single-item definitions of potency (≥71%). However, an EPIC sexual domain score ≥80 was a very strict definition of potency, and only 54% of patients with IIEF-5 scores ≥22 met this threshold. On the basis of ROC analysis (AUC = 0.90), an IIEF-5 score ≥20 was identified as the ideal cutoff for defining potency and corresponded with an EPIC sexual domain score ≥60.
Erectile dysfunction (ED) has a great impact on health-related quality of life (HRQoL). In addition to physiologic impairment, ED is related to impairment across a broad range of psychosocial domains, including—among others—affect, loneliness, psychologic adjustment, marital happiness, and depression.1–3 Therefore, erectile function is crucial to an individual's sense of identity and how he relates to his significant other.4
Consequently, assessment of pretherapy and post-therapy sexual function is a key element in evaluating HRQoL in men with prostate cancer.5 Multiple instruments have been used in the past to measure disease-specific HRQoL, particularly sexual function, in patients with prostate cancer, including the Functional Assessment of Cancer Therapy-Prostate module, the University of California-Los Angeles Prostate Cancer Index (UCLA-PCI), and the Expanded Prostate Cancer Index Composite (EPIC).6 The foundation for the EPIC, as its name suggests, is the UCLA-PCI; and, as such, their psychometric properties are similar and are specifically validated in prostate cancer.7 The International Index of Erectile Function (IIEF), however, was developed in patients with ED that resulted from multiple etiologies.8 Accordingly, prostate cancer researchers generally use one of the prostate cancer-specific questionnaires, whereas the andrology literature tends to favor the IIEF, which makes the comparison of results challenging.
In an attempt to bridge the gap between assessments of prostate cancer related ED and ED related to other causes, we sought to evaluate various definitions of potency by using the IIEF 5-item short version (IIEF-5) and the EPIC. Thus, in this report, we provide information to help with the interpretation of outcomes data based on either of these instruments.
MATERIALS AND METHODS
All patients had been diagnosed with localized prostate cancer before their inclusion in the study. Patient data were collected as part of the ongoing longitudinal assessment of patients with prostate cancer and were entered into the Institutional Review Board-approved Duke Prostate Center database. Patient characteristics (age, race, prostate-specific antigen level, disease stage, Gleason grade, treatment received, marital status, education, employment status, and household income) were abstracted from the database. After obtaining informed consent, patients were asked to fill out the EPIC questionnaire.7 Clinicians also used the IIEF-5,9 which also was entered into the database. Therefore, the database was queried for pretherapy and post-therapy patients who had answered both an IIEF-5 and an EPIC questionnaire within 7 days of each other. Because each patient served as his own control, and the purpose of this study was to compare the 2 instruments and not to evaluate ED in this patient cohort, patients who were using phosphodiesterase type-5 inhibitors or other erectile aids were not excluded.
Erectile function was measured with both the EPIC and the IIEF-5. The EPIC is a validated instrument that measures HRQoL in patients with prostate cancer in 4 domains (urinary, bowel, sexual, and hormonal).7 It is based on the UCLA-PCI but was expanded to include items that assess irritative voiding symptoms and symptoms of androgen ablation therapy.6 Each of the 4 EPIC domain summary scores demonstrated robust internal consistency and test-retest reliability during the validation in patients with prostate cancer.7 We calculated sexual domain, sexual function, and sexual bother scores for the EPIC questionnaire as described previously.10 These scores can range from 0 to 100, with higher scores representing better HRQoL.
The IIEF-5 includes questions (Q) 15, Q2, Q4, Q5, and Q7 of the IIEF8 and sometimes also is referred to as the Sexual Health Inventory for Men (SHIM).11, 12 It has demonstrated the ability to discriminate very well between men with and without ED and has an estimated area under the receiver operator characteristics (ROC) curve (AUC) of 0.97.9 We calculated sum scores by adding the responses for all 5 questions. Thus, sums could range from 1 to 25, with higher scores representing a higher degree of erectile function. Scores ≥22 generally are interpreted as absence of ED, whereas scores between 17 and 21 are considered mild ED.9, 12
Definitions of Potency
To compare various definitions of potency according to the EPIC or the IIEF-5, we defined potency first by composite scores: 1) EPIC sexual domain score ≥80, 2) EPIC sexual domain score ≥60, 3) EPIC sexual function score ≥80, or 4) EPIC sexual function score ≥60, based on previously suggested cutoff values for the UCLA-PCI.13 Alternatively, we defined potency by using individual items from the 2 instruments as men who 5) had an erection hard enough for penetration ‘most of the time’ or ‘always’ (IIEF Q2), 6) had an erection ‘firm enough for intercourse’ (EPIC Q30), 7) had an erection ‘whenever I wanted one’ (EPIC Q31), 8) had any sexual intercourse during the last 4 weeks (EPIC Q34; answered with ‘less than once a week,’ ‘about once a week,’ ‘several times a week,’ or ‘daily’), or 9) considered their sexual function ‘good’ or ‘very good’ (EPIC Q35).
Differences in EPIC and IIEF-5 scores were evaluated with the Wilcoxon rank sum test. We cross-tabulated the potency rates according to the various definitions to determine the degree of agreement or disagreement. We assessed the correlation between IIEF-5 and EPIC sexual domain or function scores for sexually active patients (defined according to IIEF Q2) with scatter plots, by Loess and linear regression, and by calculating Spearman correlation coefficient (rho). In addition, we calculated Spearman rho to quantify the relations between single-item responses and IIEF-5 and EPIC scores.
To define the ideal cutoff value for the IIEF-5 score for potency, as defined by an ideal EPIC sexual domain or function score, we plotted ROC curves.14 The ability of the IIEF-5 score to discriminate between potent and nonpotent patients was assessed by calculating the AUC.
Questionnaires from patients who had undergone any treatment for prostate cancer before filling out the questionnaire (n = 30 men) were excluded from the primary analysis. However, to assess whether our findings from the pretherapy patients would hold true for the entire cohort, we performed sensitivity analyses in which both pretherapy and post-therapy patients were included. All statistical analysis was performed with R (version 2.6.2)15 with the packages Design16 and ROCR.14
Main Outcome Measures
The concordance of potency rates by various definitions based on IIEF-5 and EPIC data and the discriminative ability of the IIEF-5 score to identify potent patients, as defined by an optimal EPIC sexual domain or function score, were the main outcome measures.
One hundred two questionnaire pairs representing 99 patients were included in the analysis. Questionnaires were collected between July 2005 and February 2008. None of the patients underwent treatment in the interim between filling out the IIEF-5 and the EPIC. The sociodemographic and cancer characteristics of these men are shown in Table 1. The median age was 60 years, and the vast majority of patients (80%) were married. Nearly half had a college education or higher. Most patients had clinical stage T1c disease, and the majority had Gleason scores ≤6.
Table 1. Sociodemographics and Cancer Characteristics of 99 Patients With Prostate Cancer
Treatment (n=30 patients who filled out questionnaire pairs after treatment)
Radical retropubic prostatectomy
Robot-assisted radical prostatectomy
Overall, 72 questionnaire pairs were completed a median 15 days (interquartile range, 8 days-23 days) before any treatment was received, and an additional 30 questionnaire pairs were completed after patients had received treatment. For 72 questionnaire pairs (71%), both instruments were answered on the same day.
The EPIC and IIEF-5 scores for the entire cohort and for the pretreatment and post-treatment patients are listed in Table 2. Post-therapy scores were significantly lower than pretherapy scores. Moreover, sexual function scores were significantly lower than sexual bother scores in both subgroups (P < .001).
Table 2. Overall Scores on the International Index of Erectile Function 5-Item Short Version and the Expanded Prostate Cancer Index Composite
First, by examining scatter plots, we observed a linear relation between IIEF-5 scores and EPIC sexual domain or function scores in pretherapy patients who were sexually active (rho = 0.766 and rho = 0.772, respectively) (Fig. 1). Similar results were obtained when all sexually active patients were included (rho = 0.826 and rho = 0.817, respectively; data not shown). On the basis of linear regression, EPIC sexual domain scores could be estimated by multiplying IIEF-5 scores by 3.270 and then subtracting 0.79. Similarly, EPIC sexual function scores could be estimated by multiplying IIEF-5 scores by 3.343 and subtracting 7.03. Nevertheless, the EPIC seemed to provide more detailed information regarding sexual function in men with relatively little impairment, because men who had an IIEF-5 score of 25 had EPIC sexual function scores that ranged from 61.1 to 91.7 (Fig. 1B). Similarly, the EPIC discriminated better among men who were not sexually active according to the IIEF-5, because those 6 men had a broad range of EPIC domain scores (10.3-64.7).
We also explored the relation between the instruments by comparing commonly used definitions of potency according to the IIEF-5 (ie, score ≥22 or score ≥17) with commonly used definitions of potency according to the EPIC (ie, score ≥80 or score ≥60). According to Table 3, an EPIC sexual domain score ≥80 is a very strict definition of potency, and only 54% of patients with IIEF-5 scores ≥22 met this definition. An EPIC sexual function score ≥80 is even stricter, and only 24% of patients with IIEF-5 scores ≥22 met this definition. However, concordance was much higher when potency was defined by an EPIC sexual domain score ≥60, because 71% of patients with IIEF-5 scores between 17 and 21 and 89% of patients with IIEF-5 scores between 22 and 25 met this threshold (Table 3). Similarly, 41 of 42 men (98%) with EPIC sexual domain scores ≥60 had IIEF-5 scores ≥17.
Table 3. Correlation of Definitions for Potency According to the International Index of Erectile Function 5-Item Short Version and the Expanded Prostate Cancer Index Composite Sexual Domain and Sexual Function Scores (Pretherapy Patients Only)*
No. of Patients (%)
No. of Patients (%)
EPIC Sexual Domain
EPIC Sexual Function
IIEF-5 indicates International Index of Erectile Function 5-item short version; EPIC, Expanded Prostate Cancer Index Composite.
Three patients had missing EPIC domain scores, although their EPIC function scores and IIEF-5 scores were available.
Table 4 shows the concordance of various definitions of potency by single items with the IIEF-5 score. IIEF-5 scores from 22 to 25 had concordance rates of 89% to 100% regardless of the single-item definition used to define potency. IIEF-5 scores from 17 to 21 related well to the classic definitions of potency (ie, having ‘erections sufficient for penetration’ and ‘firm enough for intercourse,’) but less concordance was observed regarding ‘erections whenever I wanted one’ or ‘good’ or ‘very good’ sexual function (Table 4). Patients with IIEF-5 scores ≤16 generally had little potency. However, having had any sexual intercourse in the last 4 weeks was a relatively lax definition of potency, and nearly half of the patients with IIEF-5 scores from 12 to 16 answered in the affirmative, but answers to this item still correlated well with the IIEF-5 score (rho = 0.533).
Table 4. Correlation of Definition of Potency by the International Index of Erectile Function 5-Item Short Version Scores and by Selected Single Items (Pretherapy Patients Only)
Table 5 shows a similar analysis correlating EPIC domain scores with single-item responses. Overall, correlation was high, with rho ranging from 0.701 to 0.854. An EPIC sexual domain score ≥80 ensured virtually 100% potency regardless of the single-item definition used. An EPIC sexual domain score ≥60 performed well in relation to all 5 items, with the lowest concordance being 71% for ‘good’ or ‘very good’ sexual function. Potency clearly was impaired in patients with sexual domain scores <60, regardless of the single item used. Again, having an erection sufficient for penetration ‘most of the time’ or ‘always’ and self-classification of the patient's sexual function as ‘good’ or ‘very good’ were more stringent definitions of potency than having had any sexual intercourse during the last 4 weeks. Sensitivity analysis for all 3 cross-tabulations in which both pretherapy and post-therapy patients were included showed very similar concordance and correlations (data not shown).
Table 5. Correlation of Definition of Potency by Extended Prostate Cancer Index Composite Sexual Domain Score and by Selected Single Items (Pretherapy Patients Only)*
We sought to define the ideal cutoff value for the IIEF-5 score in defining potency in men with prostate cancer by plotting ROC curves (Fig. 2). On the basis of the analysis described above, we defined patients as potent based on an EPIC sexual domain score ≥60. By using this definition, a cutoff for the IIEF-5 score of ≥20 was ideal for defining potent patients with true-positive rates of 88.1% and 84.4%, false-positive rates of 18.5% and 9.4%, and accuracies of 85.5% and 87.8% for pretherapy patients and all patients, respectively. The IIEF-5 score was very good in discriminating potent and impotent patients with an AUC of 0.900 and 0.926 for pretherapy patients and all patients, respectively. Similarly, when potency was defined using a sexual function score ≥60, an IIEF-5 score ≥20 was identified again as the ideal cutoff value (ROC curve not shown). In this case, true-positive rates were 92.7% and 84.8%, false-positive rates were 19.4% and 10.7%, and accuracies were 87.5% and 87.3% in pretherapy patients and all patients, respectively.
We observed that responses to the IIEF-5 and the EPIC were highly correlated but that potency rates varied widely, depending on the definition of potency. An EPIC sexual domain score ≥60 corresponded well with individual definitions of potency that were familiar conceptually to physicians and patients. On the basis of this finding, we identified an IIEF-5 score ≥20 as the ideal cutoff for defining potency, because this corresponded to an EPIC sexual domain or function score ≥60.
Because most studies in the andrology literature use the IIEF, whereas most studies in the prostate cancer field use the EPIC or a related instrument, our results can help bridge the gap between assessing primary ED and ED in men with prostate cancer. Moreover, for clinicians, our results provide meaning to otherwise difficult to understand IIEF-5 and EPIC sexual domain composite scores.
The American Urological Association Prostate Cancer Guideline Update Panel recently concluded that clinical studies reporting erectile function outcomes demonstrate inconsistent measurement and widely disparate rates of ED.17 Various studies have reported erectile function outcomes after radical prostatectomy,18 brachytherapy,19 or external-beam radiotherapy with or without brachytherapy20, 21 using the IIEF-5. Many more studies have used the EPIC to assess ED in prostate cancer patients. For example, outcomes after laparoscopic,22 robot-assisted,10 retropubic,23 and perineal24 radical prostatectomy have been reported based on the EPIC instrument. Similarly, the EPIC has been used to follow patients after external-beam radiotherapy or brachytherapy for prostate cancer.23 Recently, Sanda et al used a 26-item short version of the EPIC to analyze HRQoL outcomes in patients with prostate cancer.25 The data presented in our study allow for better comparison of these reports. For example, on the basis of our results, the mean pretreatment IIEF-5 score of 10.5 in a study that followed patients after brachytherapy by Fujioka et al21 would be comparable to an EPIC sexual domain score of approximately 33.5. This is much lower than the mean pretherapy EPIC sexual domain score (48.6) in another study on brachytherapy,23 suggesting that there was a lower incidence of baseline ED in the latter study.
We observed that an EPIC sexual domain score ≥80 was a very strict definition of potency. This is in line with previously reported results, which indicated that 98% of patients with an UCLA-PCI sexual function composite score ≥80 had erections firm enough for intercourse and that 100% of those patients had intercourse in the last 4 weeks.13 Conversely, having erections firm enough for intercourse was not a stringent definition of potency, because substantial numbers of patients with IIEF-5 scores from 12 to 16 or with EPIC sexual domain scores from 40 to 59.9 responded in the affirmative to this question in our study. This is in accordance with data from an analysis of the UCLA-PCI in which 33% of patients with sexual function scores from 40 to 59 reported erections firm enough for intercourse.13 Nevertheless, this does not mean that this definition is unreasonable. For an individual patient, it may be most important to be able to have satisfactory intercourse with his partner, and a composite score may not prove useful for patient counseling.26 Similarly, some men may be more interested in orgasm, masturbation, or oral intimacy. Therefore, our results mainly provide useful information for the interpretation of outcomes data and suggest that finding a single, ideal definition of potency may not be a reasonable goal.
Regarding pretherapy potency rates in patients with prostate cancer, we observed that 71% had any sexual intercourse in the previous 4 weeks (Table 4), whereas only 30% had an EPIC sexual domain score ≥80 (Table 5). When an IIEF-5 score ≥22 was used to define pretherapy potency, 51% of men met this definition (Table 4). This rate is very similar to the previously reported IIEF-based 50% and 43% potency rates in a prostate cancer screening population and in candidates for bilateral nerve-sparing radical prostatectomy, respectively.27, 28
Our study had several limitations. The retrospective design did not allow us to control or assess the sequence in which patients responded to the IIEF-5 and the EPIC instruments. Moreover, selection bias, for which we could not control, may have been present. Nevertheless, our analyses included patients with a wide range of erectile function, with sexual domain scores on the EPIC ranging from 10 to 94, and the overall prevalence of pretherapy ED was similar to the prevalence reported in other published series.27, 28
Some of the absolute numbers of patients in certain cells of our cross-tabulations were relatively low. Therefore, the percentages of patients meeting a certain definition of potency provided more an estimate for the strictness of that definition than an accurate point-estimate for the pretherapy prostate cancer population. However, the purpose of this study was to provide insight into various definitions of potency according to the IIEF-5 and the EPIC and not to analyze prevalence of ED in the pretherapy prostate cancer population.
It is important to note that the EPIC focuses on assessing sexual function over the last 4 weeks, whereas the IIEF-5 assesses erectile function over a much longer time frame. This may account for some of the discordances we observed.
We acknowledge that the IIEF-5 and the EPIC are only 2 of a multitude of available instruments that assess sexual function.29 In addition, event logs such as the Sexual Encounter Profile are used commonly to assess the effect of treatment interventions for ED.30 Moreover, both the IIEF-5 and the EPIC assess erectile function, but another important endpoint is satisfaction with treatment for ED, which is measured commonly using the Erectile Dysfunction Inventory of Treatment Satisfaction (EDITS) instrument.31 Thus, our comparison of the IIEF-5 and EPIC is to be considered as a first step toward cross-validation of different instruments, and we do not suggest that these 2 instruments are superior to other instruments that were not evaluated in the current study.
In summary, both the IIEF-5 and the EPIC are valuable instruments for the assessment of erectile function. Advantages of the IIEF-5 include its shortness and its ease of interpretation in everyday clinical use. Conversely, the EPIC is longer but provides information not only on erectile function but also on overall sexual function and sexual bother, which can be very valuable in certain clinical and research settings. In fact, the men in our study had bother scores that were significantly higher than their function scores, suggesting that not every patient is bothered by a poor sexual function.
In conclusion, we observed that responses to the IIEF-5 and the EPIC were highly correlated but that potency rates varied widely, depending on the definition of potency. Moreover, an IIEF-5 score ≥20 was identified as the ideal cutoff for defining potency corresponding with an EPIC sexual domain or function score ≥60. Our results provide valuable information that may help with the interpretation of sexual function outcomes data and provide a first step toward bridging the gap between the assessment of ED in andrology and prostate cancer research.