RAPID COMMUNICATION
`
`THE INTERNATIONAL INDEX OF ERECTILE FUNCTION
`(liEF): A MULTIDIMENSIONAL SCALE FOR ASSESSMENT OF
`ERECTILE DYSFUNCTION
`
`RAYMOND C. ROSEN, ALAN RILEY, GORM WAGNER, IAN H. OSTERLOH, JOHN KIRKPATRICK,
`AND A VANISH MISHRA
`
`ABSTRACT
`Objectives. To develop a brief, reliable, self-administered measure of erectile function that is cross-culturally
`valid and psychometrically sound, with the sensitivity and specificity for detecting treatment-related changes
`in patients with erectile dysfunction.
`Methods. Relevant domains of sexual function across various cultures were identified via a literature search of
`existing questionnaires and interviews of male patients with erectile dysfunction and of their partners. An initial
`questionnaire was administered to patients with erectile dysfunction, with results reviewed by an international
`panel of experts. Following linguistic validation in 10 languages, the final 15-itein questionnaire, the International
`Index of Erectile Function (liEF), was examined for sensitivity, specificity, reliability (internal consistency and test(cid:173)
`retest repeatability), and construct (concurrent, convergent, and discriminant) validity.
`Results. A principal components analysis identified five factors (that is, erectile function, orgasmic function, sexual
`desire, intercourse satisfaction, and overall satisfaction) with eigenvalues greater than 1.0. A high degree of
`internal consistency was observed for each of the five domains and for the total scale (Cronbach's alpha values
`of 0.73 and higher and 0.91 and higher, respectively) in the populations studied. Test-retest repeatability cor(cid:173)
`relation coefficients for the five domain scores were highly significant. The liEF demonstrated adequate construct
`validity, and all five domains showed a high degree of sensitivity and specificity to the effects of treatment.
`Significant (P values ~ 0.0001) changes between baseline and post-treatment scores were observeq across all
`five domains in the treatment responder cohort, but not in the treatment nonresponder cohort.
`Conclusions. The liEF addresses the relevant domains of male sexual function (that is, erectile function, orgasmic
`function, sexual desire, intercourse satisfaction, and overall satisfaction), is psychometrically sound, and has been
`linguistically validated in 10 languages. This questionnaire is readily self-administered in research or clinical set(cid:173)
`tings. The liEF demonstrates the sensitivity and specificity for detecting treatment-related changes in patients
`with erectile dysfunction. UROLOGY 49: 822-830, 1997. © 1997, Elsevier Science Inc. All rights reserved.
`
`Erectile dysfunction (ED), defined by a National
`
`Institutes of Health (NIH) Consensus Devel(cid:173)
`opment Conference as the inability to achieve or
`
`This research was supported by a grant from Pfizer Inc.
`From Ou Center for Sex and Mmital Health, University of
`Medicine and Dentistry of New Jersey, Robert Wood johnson
`Medical Scltool, Piscataway, New jersey; Human Sexuality Unit,
`Harewood House, Springfield University Hospital, London,
`United Kingdom; Department of Medical Physiology, Pamun In(cid:173)
`stHute, University of Copenhagen, Copenhagen, Denmarh; and
`Pfizer Central Research, Sandwich, United Kingdom, and Pfizer
`Central Research, Groton, Connecticut.
`Reprint requests: Raymond C. Rosen, Ph.D., Department of
`Psychiatl)', Robert Wood johnson Medical Sclwol, 675 Hoes
`Lane, Piscataway, N] 08854
`Submitted: December 11, 1996, accepted (with revisions):
`Febntal)' 24, 1997
`
`maintain an erection sufficient for satisfactory sex(cid:173)
`ual performance, 1 is estimated to affect as many as
`30 million men in the United States 2 The problem
`is strongly age-related, with an approximately two(cid:173)
`fold to threefold increase in the prevalence of
`moderate-to-severe ED between the ages of 40 and
`70 years 2 A variety of medical, psychologic, and
`lifestyle factors have been implicated in the etiol(cid:173)
`ogy of ED, 2
`4 which impacts negatively on self(cid:173)
`-
`esteem, quality of life, and interpersonal relation(cid:173)
`ships.'
`laboratory-based diagnostic proce(cid:173)
`Although
`dures are available, it has been proposed that sex(cid:173)
`ual function is best assessed in a naturalistic set(cid:173)
`ting with patient self-report techniques 5
`6 For this
`·
`purpose, multidimensional instruments are more
`
`© 1997, ELSEVIER SCIENCE INC.
`822. ALL RIGHTS 1:\L'SJ::RVED
`
`0090-4295/97/$17.00
`PI! 50090-4295(97)00238-0
`
`INTELGENX 1010
`
`

`
`sensitive than unidimensional scales in the evalu(cid:173)
`ation of treatment outcomes, and they are more
`psychometrically valid. 7 Multidimensional scales
`also provide greater potential for use in a clinical
`setting. Self-report methods are preferable to pa(cid:173)
`tient interview techniques, particularly in multi(cid:173)
`center, multinational clinical trials.
`Existing self-report measures of male sexual
`11 have several limitations, including ex(cid:173)
`function8
`-
`cessive length or complexity, unacceptable patient
`burden, an overly narrow or restrictive focus, and
`inadequate psychometric, cultural, or linguistic
`validation. None of the current measures has been
`demonstrated to have adequate discriminant valid(cid:173)
`ity or to provide sufficient sensitivity in evaluating
`treatment outcomes in multinational clinical tri(cid:173)
`als. Additionally, factor analytic methods were not
`used in the development of existing measures. De(cid:173)
`spite these limitations, self-report measures pro(cid:173)
`vide essential data on male sexual function in both
`research and clinical settings. 5 A strong recom(cid:173)
`mendation of the NIH Consensus Conference was
`to develop better and more reliable methods for
`assessing the symptoms of ED and relevant treat(cid:173)
`ment outcomes. 1
`The objective of the present research was to de(cid:173)
`velop a brief and reliable measure of erectile func(cid:173)
`tion that is culturally, linguistically, and psycho(cid:173)
`metrically valid. State-of-the-art methods for
`questionnaire development were used, and a mul(cid:173)
`tidimensional measure was designed to provide
`sensitive and specific outcome assessments in clin(cid:173)
`ical trials of ED. Finally, the goal was to develop
`a self-administered questionnaire that would be
`suitable for use by clinicians and researchers, one
`that would be minimally burdensome to patients.
`
`METHODS
`
`PHASE 1: ITEM 5ELECTJON
`Using multiple sources, relevant domains of male sexual
`function were identified across various cultures. A compre(cid:173)
`hensive review of the literature was conducted, and existing
`questionnaire instruments were evaluated. Detailed inter(cid:173)
`views of male patients with ED (n = 3 7) and their partners
`(n = 7) were also conducted in five countries. In this phase,
`four dimensions of male sexual function were identified: erec(cid:173)
`tile function, orgasmic function, sexual desire, and sexual sat(cid:173)
`isfaction. In a phase II trial of 351 patients with ED, an initial
`version of the questionnaire was administered and found to
`have a high degree of internal consistency among items
`(Cronbach's alpha statistic12 greater than 0.85) and excellent
`treatment sensitivity (P <0.01).u An exploratory factor anal(cid:173)
`ysis was performed that indicated a robust factor struc(cid:173)
`ture.n·14 The results were reviewed by an international panel
`of experts who made recommendations for item modification
`and the development of additional items.
`
`International Index of Erectile Function (IIEF) questionnaire
`in less than 15 minutes and reported little or no difficulty in
`comprehending the items. Linguistic validation of the instru(cid:173)
`ment was conducted in lO languages (Danish, Dutch, English
`[American, Australian, and British], Finnish, French,
`German, Italian, Norwegian, Spanish, and Swedish)* in 12
`countries by the MAPI Research Institute in Lyon, France.
`This process included forward and back translations of the
`items and comprehensive testing of the final item pool. Inter(cid:173)
`national harmonization techniques were used to ensure cross(cid:173)
`cultural equivalence of the items in the targeted languages.
`
`PHASE 3: RELIABILITY, CONSTRUCT VALIDITY, AND
`TREATMENT RESPONS1VENESS
`The final 15-item questionnaire (see Appendix) was ad(cid:173)
`ministered in a large-scale clinical trial of patients with ED
`(study A), a comparison group of functional, age-matched
`volunteers (study B), and a clinical validation study that in(cid:173)
`cluded both patients with ED and normal volunteers (study
`C). The designs of the studies and subject characteristics are
`summarized in Table I. Each study protocol was approved by
`the institutional review board at the participating site. All par(cid:173)
`ticipants in the studies gave written informed consent. Men
`aged 18 years or older with a clinical diagnosis of ED of broad(cid:173)
`spectrum etiology and of at least 6 months' duration (studies
`A and C) or normal volunteers (studies Band C) were eligible
`for enrollment. Patients with penile anatomic defects, uncon(cid:173)
`trolled major medical illnesses or psychologic disorders, or
`known drug or alcohol dependence were excluded from the
`studies.
`Study A This study consisted of a 2 to 4-week run-in phase,
`followed by a 12-week, double-blind, placebo-control1ed
`phase in which 111 patients with ED of broad-spectrum eti(cid:173)
`ology were randomized to receive either placebo or 25 mg
`(one capsule) of sildenafil (VIAGRA; Pfizer Inc.). Sildenafil is
`an oral medication that is being evaluated for the treatment
`of ED. 15
`16 The placebo or sildenafil dose could be increased
`•
`to 50 mg (two capsules) and then to 100 mg (four capsules)
`if a patient's response was suboptimal. The liEF was self-ad(cid:173)
`ministered at the screening visit (week -4 or -2), at the end
`of the run-in phase (week 0), and at the end of 2, 4, 8, and
`12 weeks of double-blind treatment. A global efficacy question
`("Did the treatment improve your erections?") was asked at
`the end of the double-blind treatment phase. The sensitivity,
`specificity, and reliability (internal consistency and test-retest
`repeatability) of the 15-item questionnaire were determined
`as follows. Each patient was designated as a "responder" or
`"nonresponder," based on his response to the end-of-treat(cid:173)
`ment global efficacy question. Within each cohort, the mean
`and median baseline-to-end point changes in response values
`for each question were calculated. The sensitivity of the IIEF
`was assessed by evaluating the clinical relevance and statisti(cid:173)
`cal significance of the changes in the responder cohort. Speci(cid:173)
`ficity was assessed in the same manner in the nonresponder
`cohort. Internal consistency was evaluated by calculating
`Cronbach's alpha statistic on the item domains and the total
`scale. 12
`Study B. This study assessed the response to the liEF ques(cid:173)
`tionnaire in 109 male volunteers (controls) without any his(cid:173)
`tory of male ED. These volunteers were age-matched to the
`patients randomized in study A (Table I). The liEF was self(cid:173)
`administered, with the results in these controls compared
`with those obtained in men with ED in study A using be-
`
`PHASE 2: CULTURAL AND L1NGU1ST1C EVALUATION
`Pilot testing of the instrument was conducted in 14 men
`with ED in the United Kingdom. All patients completed the
`
`*Additional validation studies of other languages (for example,
`Arabic, Chinese, Mandwin, and Portuguese, among others) in
`Asia and Latin America are ongoing.
`
`UROLOGY 49 (6), 1997
`
`823
`
`INTELGENX 1010
`
`

`
`TABLE I. Study designs and baseline characteristics of individuals enrolled in validation studies
`Study C
`Patients
`with ED
`
`Study A (Patients with ED)
`
`Study B
`(Controls)
`
`Controls
`
`Study Design
`Treatments
`
`Duration of study
`Timing of liEF self-administration
`Other relevant assessments
`
`Sildenafil (25, 50, or 100 mg) or
`placebo
`12 weeks
`Week -4 or -2, 0, 2, 4, 8, and 12
`Global efficacy question: final visit
`
`None
`
`1 day
`Day 1
`
`None
`
`4 weeks
`Week 0 and 4
`Clinical interview:
`Week 0 and 4
`Locke-Wallace Scale:
`WeekO
`Marlowe-Crowne Scale:
`WeekO
`
`Patient characteristics
`n
`Mean age, yr (range)
`Mean duration of ED, yr (range)
`Primary etiology*
`Organic
`Psychogenic
`Mixed
`Unknown
`
`111
`56 (29-89)
`4.61 (1-37)
`
`21%
`40%
`37%
`3%
`
`KEY: ED = erectile dysfunction; llEF = lntemational Index of Erectile Function.
`*Percentages do not lota/100 due to rounding.
`
`109
`55 (29-76)
`
`37
`53 (29-71)
`5.9 (1-18)
`
`21
`58 (37-76)
`
`14%
`49%
`38%
`0%
`
`tween-groups discriminant analysis (analysis of covariance
`controlling for age) and post hoc comparison of group differ(cid:173)
`ences on individual items.
`Study C. This 4-week study evaluated the construct validity
`and test-retest repeatability of the liEF in 37 patients with
`male ED and in 21 age-matched controls (Table I). The liEF
`was self-administe~ed at week 0 and week 4. In this study,
`blinded clinical interviews of patients were conducted at week
`0 to evaluate the convergent validity of the measure (that is,
`concordance with an independent method of assessment). In
`addition, patients completed measures of marital satisfaction
`(Locke-Wallace scale17
`) and social desirability (Marlowe(cid:173)
`Crowne scale18
`) to assess divergent validity (that is, separate(cid:173)
`ness from overlapping or related constructs) at week 0. Test(cid:173)
`retest reliability of the total and individual item scores of the
`liEF were assessed by calculating the Pearson product-mo(cid:173)
`ment correlation coefficient19 for each group (patients and
`controls). Internal consistency was evaluated using the Ku(cid:173)
`der-Richardson formula. Discriminant validity was assessed
`using repeated-measures analysis of variance, with subject
`group as the between-groups variable, time (week 0 and week
`4) as the repeated-measures factor, and study measure as the
`outcome variable.
`
`RESULTS
`
`FACTOR ANALYSIS AND DOMAIN SCORING
`A principal components analysis (with varimax
`rotation) was performed to investigate the factor
`structure of the final 15-item questionnaire (see
`Appendix). Five factors with eigenvalues' greater
`than 1.0 were identified (Table ll). Final item se-
`
`t Eigenvalue is a statistical measure of the relative explanatory
`power of i11dividual factors in a factm: analysis.
`
`lection for each factor was based on a combination
`of statistical and clinical considerations 20 Based
`on results of the confirmatory factor analysis, to(cid:173)
`gether with clinical interviews and expert panel
`consultation, the responses to individual items of
`the questionnaire were assigned to five separate
`domains of sexual function: (l) erectile function,
`(2) orgasmic function, (3) sexual desire, (4) inter(cid:173)
`course satisfaction, and (5) overall satisfaction.
`Domain scores were computed by summing the
`scores for individual items in each domain. The
`system of domain scoring and resulting interdo(cid:173)
`main correlations are presented in Table Ill.
`
`SCALE RELIABILITY
`Two separate aspects of scale reliability were
`evaluated, namely, internal consistency and test(cid:173)
`retest repeatability. Internal consistency ( Cron(cid:173)
`bach's alpha) was computed separately for the five
`domains and for all items combined in each of the
`three test samples. Responses in the erectile and
`orgasmic function domains were highly consis(cid:173)
`tent, with alpha values greater than 0.90 (Table
`IV). A satisfactory degree of consistency also was
`observed for items in the other domains (alpha
`values greater than 0. 70) and for the total scale
`(alpha values greater than 0.90) in each of the test
`samples.
`Test-retest repeatability was assessed in study C
`by computing correlations between the domain
`scores and total scale scores at baseline and week
`
`824
`
`UROLOGY 49 (6), 1997
`
`INTELGENX 1010
`
`

`
`TABLE II. Principal components analysis with varimax rotation of 15 questions of International
`Index of Erectile Function: factor loadings*
`Factor 1
`Factor 3
`Factor 2
`
`Item
`
`Factor 4
`
`Factor 5
`
`1. Erection frequency
`2. Erection firmness
`3. Penetration ability
`4. Maintenance frequency
`5. Maintenance ability
`6. Intercourse frequency
`7.
`Intercourse satisfaction
`8. Intercourse enjoyment
`9. Ejaculation frequency
`10. Orgasm frequency
`11. Desire frequency
`12. Desire level
`13. Overall satisfaction
`14. Relationship satisfaction
`15. Erection confidence
`
`Eigenvalue
`
`0.77
`0.92
`0.89
`0.82
`0.68
`0.10
`0.61
`0.53
`0.26
`0.23
`0.06
`0.04
`0.29
`0.18
`0.65
`
`4.72
`
`*Items with t/Je highest loadings within each factor are bolclfaccd.
`
`0.03
`0.12
`0.16
`0.26
`0.39
`-0.02
`0.28
`0.39
`0.20
`0.25
`-0.01
`0.26
`0.76
`0.83
`0.53
`
`2.22
`
`0.31
`0.20
`0.15
`0.13
`0.09
`0.11
`0.31
`0.18
`0.89
`0.87
`0.15
`0.07
`0.28
`0.21
`0.01
`
`2.03
`
`0.17
`0.08
`0.06
`-0.02
`0.07
`0.34
`-0.13
`0.01
`0.10
`0.18
`0.88
`0.87
`0.15
`0.14
`0.01
`
`1.81
`
`-0.05
`0.04
`0.14
`0.22
`0.41
`0.79
`0.48
`0.53
`0.13
`0.12
`0.16
`0.08
`-0.01
`0.13
`0.07
`
`1.47
`
`4 visits. As shown in Table IV, test-retest repeat(cid:173)
`ability was relatively high for the erectile function
`(r = 0.84) and intercourse satisfaction (r = 0.81)
`domains, as well as for the total scale scores (r =
`0.82). Moderately high correlations were observed
`for the other domains (r values of 0.64 to 0. 77).
`
`DISCRIMINANT VALIDITY
`Discriminant validity, or the ability of the liEF
`scale to discriminate reliably between clinical and
`nonclinical populations, was assessed by compar(cid:173)
`ing the responses from patients with ED with those
`from controls in two studies. As shown in Table
`V, highly significant differences were observed be(cid:173)
`tween the the patients with ED and age-matched
`controls for most domains. Differences between
`domain scores between these two groups were
`greatest for
`the erectile function domain (P
`sO.OOOl), followed by intercourse satisfaction (P
`sO.OOl) and overall satisfaction (P sO.OOl). The
`least degree of difference between patients and
`controls was seen for the sexual desire domain,
`with results failing to reach statistical significance
`in study C. This result is not surprising because
`all patients were recruited for a clinical trial of ED
`and were excluded for concomitant sexual disor(cid:173)
`ders, such as hypoactive sexual desire.
`
`CONVERGENT AND DNERGENT VALIDITY
`To demonstrate construct validity of a new mea(cid:173)
`sure, it is important to show that scale scores are
`positively correlated with independent measures
`of the same or similar domains (convergent valid(cid:173)
`ity). Conversely, there should be minimal associ(cid:173)
`ation with measures that do not directly assess the
`
`domains in question (divergent validity). In study
`C, domain scores were compared with blinded, in(cid:173)
`dependent clinician ratings of sexual functioning
`and with scales that measure marital adjustment
`(Locke-Wallace) and social desirability (Marlowe(cid:173)
`Crowne). Significant positive correlations were
`observed between independent clinician ratings
`and subscale scores for all five domains (Table VI).
`In contrast, none of the correlations between do(cid:173)
`main scores and measures of marital adjustment
`or social desirability reached statistical signifi(cid:173)
`cance.
`
`SENSITIVITY AND SPECIFICITY
`To evaluate the sensitivity of the liEF, a com(cid:173)
`parison was made between mean pretreatment and
`post-treatment domain scores of patients who
`were self-rated as treatment responders in study A.
`Specificity was assessed by comparing the pretreat(cid:173)
`ment and post-treatment domain scores in patients
`rated as nonresponders in the same study. Patients
`were defined as responders or nonresponders
`based on their response to the end-of-treatment
`global efficacy question. All five domains of the
`liEF demonstrated a high degree of sensitivity and
`specificity to the effects of treatment (Table VII).
`Although the magnitude of change was greatest for
`the erectile function domain, significant changes
`were observed across all five domains in the treat(cid:173)
`ment responder group. The lowest magnitude of
`change was noted for the sexual desire domain. In
`contrast, none of the comparisons in the treatment
`nonresponder group approached significance (P
`values of O.ll to 0.79).
`
`UROLOGY 49 (6), 1997
`
`825
`
`INTELGENX 1010
`
`

`
`TABLE Ill.
`
`Items
`
`I, 2, 3, 4, 5, 15
`9, 10
`II, 12
`6, 7,8
`13, 14
`
`liEF domain scoring and intercorrelations
`Domain Scoring
`Score
`Minimum
`Range
`Score
`0 (or 1)-5
`0-5
`1-5
`0-5
`1-5
`
`Maximum
`Score
`30
`10
`10
`15
`10
`
`I
`0
`2
`0
`2
`
`EF
`1.00
`0.55
`0.30
`0.76
`0.60
`
`Domain lntercorrelations
`SD
`
`IS
`
`OF
`
`1.00
`0.39
`0.47
`0.53
`
`1.00
`0.35
`0.37
`
`1.00
`0.53
`
`OS
`
`1.00
`
`Domain
`EF
`OF
`SD
`IS
`OS
`
`EF
`OF
`SD
`IS
`OS
`
`KEr: EF = erectile Junction; liEF = lntcmationallndex of Erectile Fm1ction; IS = intercourse satisfaclion; OF = OJgasmic
`function; OS = overall satisfaction; SD = se:..11al desire.
`
`COMMENT
`
`A 15-item, self-administered questionnaire scale
`was developed for the assessment of erectile func(cid:173)
`tion. This instrument (the liEF) was developed in
`several stages, including initial pretesting with se(cid:173)
`lected patient groups and expert panel consult(cid:173)
`ants, followed by an intensive linguistic validation
`process. Based on a principal components analysis
`with varimax rotation, five factors or response do(cid:173)
`mains were identified: (1) erectile function, (2) or(cid:173)
`gasmic function, (3) sexual desire, ( 4) intercourse
`satisfaction, and (5) overall satisfaction. The high(cid:173)
`est degree of positive correlation was between
`erectile function and intercourse satisfaction (r =
`0.76), with two items (items 7 and 8) showing
`positive loadings on both factors. This is not sur(cid:173)
`prising because a primary outcome of ED for most
`patients is the inability to achieve satisfactory sex(cid:173)
`ual intercourse. 1
`Psychometric validation of the final instrument
`was addressed in three major areas: (1) test relia(cid:173)
`bility, (2) construct validity, and (3) treatment re(cid:173)
`sponsiveness. Adequate performance in each of
`these areas should be demonstrated before a new
`scale is accepted for general research or clinical
`23 For the liEF, analyses were performed in
`use 21
`-
`each of these areas in two separate samples of pa(cid:173)
`tients with ED and age-matched controls. Overall,
`the liEF was shown to have strong internal con(cid:173)
`sistency, measured in terms of both the total scale
`and individual domain scores, and adequate test(cid:173)
`retest repeatability. Although some variation in the
`degree of internal consistency was noted between
`samples, all of the values obtained were greater
`than 0. 70 and more than half were greater than
`
`0.90. Test-retest repeatability correlation coeffi(cid:173)
`cients ranged from 0.64 to 0.84, and all were
`highly significant.
`Construct validity (that is, whether the instru(cid:173)
`ment actually measures what it was designed to
`assess) is normally accomplished by experimental
`testing of a priori questions or hypotheses, such
`as: (1) Will the test reliably differentiate between
`clinical patients and age-matched controls? (dis(cid:173)
`criminant validity); (2) Can a positive association
`be shown with alternative measures of the same
`construct or domains? (convergent validity); and
`(3) Are the results influenced by related, but con(cid:173)
`ceptually independent, variables? (divergent valid(cid:173)
`ity). In the present study, adequate construct va(cid:173)
`lidity was established in each of these three areas.
`Discriminant validity was demonstrated by a com(cid:173)
`parison of baseline scores between patients and
`controls. In the larger sample (studies A and B),
`between-group differences were highly significant
`(P values s0.01) for all five domains. In the
`smaller sample (study C), differences between
`groups were significant (P values sO.Ol) for all
`domains, with the exception of sexual desire (P =
`0.72). In this study, patients and controls were
`closely matched on sexual desire, perhaps reflect(cid:173)
`ing a high level of sexual motivation in patients
`seeking treatment in a clinical trial of ED. Tests of
`convergent and divergent validity were similarly
`confirmatory. First, a significant positive associa(cid:173)
`tion was shown with independent clinician ratings
`for each of the major response domains. As ex(cid:173)
`pected, the highest correlation was observed for
`the domain of erectile function (r = 0. 75). This
`association might have been even higher, except
`for the fact that clinician interview ratings took
`
`826
`
`UROLOGY 49 (6), 1997
`
`INTELGENX 1010
`
`

`
`TABLE IV.
`
`All items
`Erectile function
`Orgasmic function
`Sexual desire
`Intercourse satisfaction
`Overall satisfaction
`
`Internal Consistency*
`Study A
`Study C
`Study B
`
`fiEF domain characteristics: reliability
`Test-Retest
`Repeatability'
`Study C
`0.82
`0.84
`0.64
`0.71
`0.81
`0.77
`
`0.91
`0.92
`0.92
`0.77
`0.73
`0.74.
`
`0.96
`0.96
`0.99
`0.82
`0.87
`0.87
`
`0.91
`0.93
`0.93
`0.91
`0.88
`0.86
`
`KEY: IIEF = IHtemational Index of Erectile Function.
`* Cmnbach's alpha.
`.
`1 PearsonJJmduct-mmncnt C01Tclation coefficient.
`
`TABLE V.
`
`liEF domain characteristics: discriminant validity
`Study A and Study B
`Study C
`Patients
`Patients
`Controls
`Controls
`Mean± SD Mean± SD
`PValue* Mean± SO Mean± SD
`13.5 ± 8.1
`26.9 ± 5.3
`7.3 ± 3.5
`9.5 ± 2.2
`7.2 ± 1.5
`7.0 ± 1.9
`6.0 ± 4.5
`10.8 ± 4.8
`5.5 ± 2.4
`9.0 ± 1.6
`
`10.7 ± 6.5
`5.3 ± 3.2
`6.3 ± 1.9
`5.5 ± 3.0
`4.4 ± 2.3
`
`25.8 ± 7.6
`8.8 ± 2.9
`7.0 ± 1.8
`10.6 ± 3.9
`8.6± 1.7
`
`"0.0001
`"0.001
`"0.01
`"0.001
`"0.001
`
`Domain
`
`Erectile function
`Orgasmic function
`Sexual desire
`Intercourse satisfaction
`Overall satisfaction
`
`PValue*
`
`"0.0001
`:50.01
`0.72
`"0.0003
`"0.0001
`
`I<Er: liEF = Inlernatimwl Index of Erectile Function.
`* P values cL~scsscd using repeat-measures, bctwceu-graups analysis of variance metlwd.
`
`TABLE VI.
`
`Domain
`
`Erectile function
`Orgasmic function
`Sexual desire
`Intercourse satisfaction
`Overall satisfaction
`
`Clinical Interview
`Pearson r
`P Value
`0.75
`0.51
`0.61
`0.45
`0.63
`
`<0.0001
`<0.001
`<0.0001
`<0.005
`<0.001
`
`liEF domain characteristics: convergent and divergent validity
`Validation Measure (Study C)
`Marital Adjustment
`(Locke-Wallace)
`Pearson r
`PValue
`-0.08
`-0.21
`0.16
`-0.05
`0.31
`
`Social Desirability
`(Marlowe-Crowne)
`Pearson r
`P Value
`-0.07
`-0.13
`0.24
`-0.02
`0.17
`
`0.63
`0.45
`0.15
`0.78
`0.31
`
`0.62
`0.23
`0.36
`0.89
`0.07
`
`KEY: liEF= lntemational Judex of Erectile ftmcUon.
`
`into account both past history and current sexual
`performance ratings, whereas the questionnaire
`assessed only the latter. Second, measures of social
`desirability and marital adjustment were not sig(cid:173)
`nificantly correlated with any liEF domain scores.
`This suggests that liEF scores are highly indepen(cid:173)
`dent of social desirability and marital adjustment
`influences.
`A final area of test validation concerns treatment
`responsiveness, or the sensitivity and specificity of
`the instrument, which was evaluated by compar(cid:173)
`ing the change between baseline and end point
`scores in treatment responders and nonresponders
`(study A). A high degree of sensitivity and speci-
`
`ficity was demonstrated for each of the domains of
`the liEF. For the responder group, highly signifi(cid:173)
`cant changes between baseline and end point
`scores were observed in each domain. The mean
`change in scores was highest for the erectile func(cid:173)
`tion domain and lowest for the sexual desire do(cid:173)
`main. These results are not surprising because the
`study drug, sildenafil, is an agent with a peripheral
`site of action and proerectile effects. 15
`16 Treatment
`•
`response specificity was demonstrated by the rel(cid:173)
`ative lack of change between baseline and end
`point scores in the nonresponder group. Taken to(cid:173)
`gether, these findings indicate that the liEF is a
`highly sensitive and specific instrument for de-
`
`UROLOGY 49 (6), 1997
`
`827
`
`INTELGENX 1010
`
`

`
`TABLE VII.
`
`Domain
`Treatment responders
`Erectile function
`Orgasmic fUnction
`Sexual desire
`Intercourse satisfaction
`Overall satisfaction
`Treatment nonresponders
`Erectile function
`Orgasmic function
`Sexual desire
`Intercourse satisfaction
`Overall satisfaction
`
`liEF domain characteristics: sensitivity and specificity (study A)
`t Statistic
`Mean Change*
`SEM
`n
`
`50
`50
`49
`48
`49
`
`42
`42
`42
`42
`42
`
`12.80
`3.44
`1.12
`4.63
`3.47
`
`0.88
`0.70
`-0.52
`0.10
`0.57
`
`1.2
`0.5
`0.3
`0.6
`OA
`
`0.8
`0.6
`0.3
`0.4
`0.3
`
`10.6
`6.4
`4.5
`8.4
`8.4
`
`1.07
`1.25
`-1.55
`0.27
`1.65
`
`PValue
`
`"'0.0001
`"'0.0001
`"'0.000 1
`"'0.000 1
`"'0.0001
`
`0.67
`0.36
`0.32
`0.79
`0.11
`
`KEl': liEF = Intcnwtional Index of Erectile Function.
`*Mean difference between prelrealmenL score and post-treatment scores.
`
`tecting changes in erectile function in response to
`treatment.
`Other advantages of this new scale are worth
`noting. First, all of the major aspects of the NIH
`definition are addressed by individual items in
`the erectile function domain. A patient's ability
`to achieve or maintain an erection sufficient for
`intercourse are addressed separately (items 3
`and 4, respectively), as is the degree of satisfac(cid:173)
`tion achieved (item 7). The liEF also addresses
`the ability to achieve erections independent of
`intercourse (items 1 and 2). Furthermore, the
`psychologic dimension of erectile confidence is
`assessed (item 15), which has been shown to be
`related to treatment outcome in other con(cid:173)
`texts24 Finally, the brevity and ease of compre(cid:173)
`hension of the measure provide important prac(cid:173)
`tical advantages. For example, the liEF may be
`ideally suited for use in studies assessing the
`prevalence of ED in different countries.
`Limitations of the instrument are the sole fo(cid:173)
`cus on current sexual functioning, the superfi(cid:173)
`cial assessment of nonerectile components of
`sexual response, and the limited assessment of
`the partner relationship. Although the liEF pro(cid:173)
`vides a broad measure of sexual function across
`five domains, it should be viewed as an adjunct
`to, rather than a substitute for, a detailed sexual
`history. The liEF was designed as an assessment
`measure for ED, and it is not intended for use as
`a primary measure of premature ejaculation or
`hypoactive sexual desire. Finally, the liEF has
`not been evaluated in long-term follow-up stud(cid:173)
`ies or in the patient subpopulations that were
`excluded from the clinical trials described, such
`as those with anatomic deformities (for example,
`Peyronie's disease). Thus, further studies would
`be needed to determine whether this instrument
`is valid in these instances.
`
`CONCLUSIONS
`The liEF, a 15-item questionnaire, has been de(cid:173)
`veloped and validated as a brief and reliable self(cid:173)
`administered scale for assessing erectile function.
`This instrument is psychometrically sound and
`easy to administer in research and clinical settings.
`The liEF currently is available in 10 languages for
`use in multinational clinical trials, and it demon(cid:173)
`strates adequate sensitivity and specificity for de(cid:173)
`tecting treatment-related changes in erectile func(cid:173)
`tion in patients with ED.
`
`AcKNO\VLEDGMENT. To Drs. Pierre Wicker, Frances Quirk,
`Mike Hodges, Murray Maytom, David Cox, and Fidela Mo(cid:173)
`reno, and to Mike Smith, Andrew Lee, Michelle Cuddigan,
`jennifer Gill, and Claire Hargreaves for their valuable support
`and contributions, and to Dr. Patricia Leinen for her assis(cid:173)
`tance in the preparation of the manuscript.
`
`REFERENCES
`L NIH Consensus Development Panel on Impotence: Im(cid:173)
`potence. JAMA 270: 83-90, 1993.
`2. Feldman HA, Goldstein I, Hatzichristou DG, Krane Rj,
`and McKinlay JB: Impotence and its medical and psychosocial
`correlates: results of the Massachusetts Male Aging Study. J
`Urol151: 54-61, 1994.
`3. Saenz de Tejada I, Goldstein I, Azadzoi K, Krane Rj,
`and Cohen RA: Impaired neurogenic and endothelium-me(cid:173)
`diated relaxation of penile smooth muscle from diabetic men
`with impotence. N Eng1J Med 320: 1025, 1989.
`4. Shabsigh R, Fishman l], Schum C, and Dunn JK: Cig(cid:173)
`arette smoking and other risk factors in vasculogenic impo(cid:173)
`tence. Urology 38: 227, 1991.
`5. Conte HR: Development and use of self-report tech(cid:173)
`niques for assessing sexual functioning: a review