The objective of this study was to investigate the predictive validity of the Dutch translation of the Golombok Rust Inventory of Sexual Satisfaction (GRISS) for the presence of clinically relevant sexual dysfunctions and patient's need of help. A total of 57 male urological outpatients (age 50.7±13.2 y; range 27–77 y) completed the GRISS, a 28-item self-report questionnaire assessing sexual dysfunctions and sexual satisfaction in heterosexual relationships. GRISS subscales were found to differentiate multivariately between men with and without sexual dysfunctions. The erectile dysfunction (In the original English version of the GRISS, the term ‘impotence’ was used instead of ‘erectile dysfunction’), dissatisfaction, and infrequency subscales also differentiated univariately between these groups. The predictive validity for the presence of sexual dysfunctions and for the patients' need for professional help was investigated. Prediction models, derived by means of logistic regression analysis, were tested in a crossvalidation sample. Sensitivity and specificity for the presence of sexual dysfunctions, as well as the predictive values positive and negative were found to be satisfactory. The predictive validity of the GRISS was found equally satisfactory for the patients' need for professional help. The discriminant and predictive validity of the GRISS in men with and without sexual dysfunctions appeared satisfactory. Routine use of the GRISS appears warranted for the screening of sexual dysfunctions in new urological patients.
The Dutch translation of the Golombok Rust Inventory of Sexual Satisfaction (GRISS) is a short paper-and-pencil questionnaire that can be used to evaluate sexual functioning and sexual satisfaction.1,2,3 Although some instruments were available for the assessment of sexual functioning and sexual dysfunctions for the Dutch language area: the Sexual Experience Scales, the Intimate Bodily Contact Scales, there was an expressed need for a shorter Dutch language instrument for screening sexual dysfunctions and to measure therapy outcome.4,5
Ter Kuile, Vroege, and van Lankveld kept the Dutch translation of the GRISS as close as possible to the original English version. (Translation of the GRISS and its use for research purposes have been performed in agreement with the author and the Publisher of the original English version, NFER-Nelson, Darville House, 2 Oxford Rd East, Windsor, Berkshire SL4 1DF, UK. The Dutch translation is available for inspection purposes through the author.)6 The questionnaire has separate forms for men and women. Each form consists of 28 items and covers the most frequently occurring sexual dysfunctions of heterosexual persons with a steady partner. Completion of the questionnaire requires between 5 and 10 min.7 It provides a total score of the person's satisfaction with sexual functioning within the relationship. In total, 12 subscale scores are given for erectile dysfunction, premature ejaculation, female anorgasmia, vaginismus, infrequency of sexual contact, sexual noncommunication, male and female dissatisfaction, male and female nonsensuality, and male and female avoidance of sex. Partners' aggregated scores constitute a profile of the sexual functioning of both partners within their relationship. According to its designers, the GRISS can be applied as a screening instrument and for the measurement of therapy outcome.2,3,8
The factor structure, internal consistency, and stability of the Dutch adaptation have been examined and found satisfactory.7 On the basis of these findings, it was considered justified to maintain the original subscale structure. The construct validity of the GRISS has been investigated in both male and female sexological patients and in gynaecological patients and their male partners in The Netherlands.9 Convergent validity of the GRISS was found with the sexual dissatisfaction subscale of the Maudsley Marital Questionnaire. Divergent validity was considered to be satisfactory, when GRISS scores were tested against instruments measuring psychological distress (SCL-90), relationship dissatisfaction (Maudsley Marital Questionnaire), and the personality traits of neuroticism, psychoticism, extraversion (Eysenck Personality Questionnaire—Revised). Correlations of GRISS subscales with social desirability (Lie Scale of the EPQ-R) were absent.
The predictive validity of the GRISS constructs for the presence of sexual dysfunctions was investigated in the same samples.9 Patients' self-report of the presence of sexual dysfunction was used as the criterion against which prediction was tested. Sexual dysfunction was recorded as such if a patient-reported sexual problem was found to meet the clinical criteria of sexual disorder of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV).10 These criteria require not only the presence of sexual difficulty, but also the report of the patient or the patient's partner of substantial distress because of the sexual difficulty. In male sexology patients, the sensitivity of the GRISS to detect the presence of sexual dysfunctions was 71.6%, whereas specificity was 91.4% (overall correct prediction 84.5%) when only the men's scores were considered. When GRISS scores of both partners were used for prediction, the overall correct prediction for sexual dysfunction in men increased to 87.7% (sensitivity: 68.1%; specificity: 94.2%). For the identification of sexual dysfunctions in female sexology outpatients, scores of female partners alone correctly predicted 84.8% (sensitivity: 96.8%; but specificity: 15.1%). With both partners' scores being used, the overall correct prediction increased to 90.3%, with the sensitivity remaining high at 96.5%, but with specificity rising to 62.0%. Prediction of sexual dysfunctions in female gynaecology patients was also found to be satisfactory when both partners' scores were available. With only female scores, the overall correct prediction was 79.2% (sensitivity: 40.0%; specificity: 94.4%). With both partners' scores, overall correct prediction was 79.3%, with sensitivity increasing to 62.5%, while specificity remained high at 90.0%. The least satisfactory prediction was revealed in male partners of gynaecological patients. Although overall correct prediction in this group appeared to be sufficient (89.7), this finding was entirely accounted for by specificity (99.2%), whereas sensitivity was very low (17.7%). Using both partners' GRISS scores, it was not possible to raise sensitivity to an acceptable level (41.7%). Van Lankveld and ter Kuile suggested that this low validity was attributable to fact that the data were collected within a gynaecological outpatient clinic.9 In all, 75% of the male partners presenting a sexual dysfunction had a female partner who also had a sexual dysfunction. Their presence in the gynaecology outpatient clinic was hypothesised to reflect a shared view of both partners that the female sexual problem had priority. Most of these men merely accompanied their partner during her visit to the gynaecologist and did not seek help for their own sexual dysfunction.
The present study aims to assess the predictive power of the GRISS for the presence of sexual dysfunctions of men within a medical setting. For this purpose, male patients of a urological outpatient clinic who came for their first visit were asked to participate. The urological outpatient clinic probably presents a more adequate setting with respect to the occurrence of male sexual dysfunctions. Slatford and Currie11 found a 10.6% prevalence of sexual dysfunctions in new patients of a urological outpatient clinic. Benet and Melman12 published a review of the epidemiological literature of erectile dysfunction. They found a 34% prevalence of erectile dysfunction in a study that investigated patients of a general hospital outpatient clinic. They estimated that a majority of the new patients in a more specific urological setting would suffer from erectile dysfunction.
The GRISS was completed by individuals with and without sexual dysfunctions who came for their first visit to the urology outpatient clinic of the Maastricht University Medical Centre. Subjects who currently had a heterosexual relationship were asked to complete the GRISS. In all, 57 male patients participated. Subjects were required to have thorough mastery of the Dutch language, to be judged by the consulting urologist, which was deemed necessary to be able to answer all relevant questions. Demographic characteristics of the entire study sample are displayed in the first column in Table 1. For the assessment of predictive validity, patients were individually included in the contrasted groups, respectively, with and without sexual dysfunctions, depending on whether or not a sexual dysfunction was reported. Group membership then served as a criterion variable against which prediction was tested.
Upon their presentation at the outpatient clinic's desk, new patients received a leaflet with information about the study. It explained that a paper-and-pencil questionnaire was to be completed and that the doctor would ask a number of questions on sexual functioning and possible sexual complaints. Complete privacy was assured, and voluntariness of participation was emphasised. After having read the information, the patient was asked if he wanted to participate. If so, a brief sexual history form was passed to the urologist before the start of his consultation (see Appendix 1), who remained blind with respect to the patient's GRISS scores. At the end of the consultation, the urologist completed the brief sexual history form. If, at that time, the patient had not presented a sexual dysfunction at his own initiative, the urologist posed all relevant questions and completed the form. After leaving the doctor's office the patient was given the questionnaire (GRISS) and asked to complete it in a quiet corner of the waiting room, allowing for sufficient privacy.
To test the discriminant validity of the GRISS scales against the criterion of the presence or absence of sexual dysfunctions, one-way MANOVA was performed to enable multivariate comparison. Univariate comparisons were made by means of t-tests for independent samples. For the evaluation of predictive validity, multivariate analyses were performed by means of logistic regression using the maximum likelihood method.13 Based on the models produced by the logistic regressions, clinical decision analysis was performed to examine the classificatory qualities of the GRISS in our study sample. Sensitivity and specificity indicate the abilities of the instrument to correctly detect the presence or absence of sexual dysfunctions in these study samples by means of the optimal linear combination of selected subscale scores as calculated by the logistic regression. Sensitivity and specificity are defined as the proportion of subjects with and without respectively, sexual dysfunctions, which has been correctly identified by the regression model of their GRISS scores. Of further importance for the qualifications of the GRISS as a clinical screening instrument are the predictive values positive and negative, which were calculated on the basis of the same logistic regression model. Predictive value positive is defined as the proportion of positive predictions (the GRISS scores predict the subject to have a sexual dysfunction) that coincide with the observed presence of a sexual dysfunction. Predictive value negative is defined as the proportion of negative predictions (the GRISS scores predict the subject not to have a sexual dysfunction) that coincide with the observed absence of a sexual dysfunction. Since regression models tend to fit the data of the sample from which they are extracted better than any other sample, the validity of these predictions is questionable.14 To enhance validity, randomly split sample halves of the data set were used to, respectively, derive and crossvalidate predictors.
Before conducting these analyses the data sets were screened for missing data, multivariate outliers and other assumptions. Groups were independently examined for missing data and multivariate outliers. No multivariate outliers (Mahalanobis distance: χ2(7)>24.322, P=0.001) were identified after performance of linear regression.14 Results of the further evaluation of assumptions showed no threats to the assumptions of logistic regression. A constant factor was included in every model. Regression models were calculated with forced entry of all GRISS subscales at the first step. Subsequently, predictors were individually evaluated for their contribution to the regression model and removed if found to be superfluous on the basis of the likelihood-ratio test. For the purpose of crossvalidation, the sample was randomly split into two equal parts. From the first sample half, prediction equations were derived. These equations were subsequently crossvalidated in the other sample half. All statistics were tested against α=0.05.
In all, 57 male patients (age 50.7±13.2 y; range 27–77 y) participated. Age of men with and without sexual dysfunctions differed significantly (t (55)=2.8, P<0.01). Men with sexual dysfunctions were older (age: 56.3±12.6 y) than functional men (age: 47.0±12.4 y) (see Table 1). Sexually functional and dysfunctional men did not differ with respect to other demographic features than age. Of 23 men (40.4%), who were found to have sexual dysfunctions, six men (10.5%) reported this problem at their own initiative. 30% declared having problems with sexual functioning only when specifically questioned by the urologist. Self-reported sexual dysfunctions were erectile dysfunction (n=5), premature ejaculation (n=2), and male dyspareunia (n=1). Of patients reporting sexual dysfunctions (both self-reported and clinician-elicited), eight (14.0%) reported two different sexual dysfunctions. All of these eight men reported erectile dysfunction as their primary complaint. Seven of them had orgasmic disorder as secondary problems, mainly inability to reach orgasm and to ejaculate (n=6). Distribution of sexual dysfunctions (both primary and secondary) is shown in Table 1. Prevailing sexual dysfunctions were erectile dysfunction (30%) and orgasmic/ejaculatory disorders (14%), most prominently retarded or absent ejaculation. Of 23 men presenting sexual dysfunctions, two did not want to receive any professional assistance when asked. In all, 18 (32%) men wished to receive help, one hesitated, and two were found to already receive professional help.
Discriminant validity of the GRISS subscales
The ability of the GRISS to discriminate between groups of subjects with and without sexual dysfunctions was first investigated by means of MANOVA. Independent variable was the presence of sexual dysfunction (yes vs no). Raw scores of the seven GRISS subscales were entered as dependent variables. A significant group effect was found of the combined subscale scores (F (7, 49)=5.09; P=0.000; η2=0.421; power=0.993). GRISS subscales further proved to differentiate univariately between men with and without sexual dysfunctions (Table 2). Although all mean subscale scores were higher in dysfunctional men, as compared to functional men, only scores on the erectile dysfunction, dissatisfaction, and infrequency subscales were significantly higher (two-tailed, with Bonferroni correction) in men with sexual dysfunctions.
Another analysis was performed to evaluate the discriminative validity of GRISS subscales with regard to the desire of patients for professional assistance. For this purpose, a group desiring professional help was defined comprising patients who indicated having a sexual dysfunction and answering affirmatively to the urologist's question as to whether or not they wanted professional help, those still hesitating, and the patients already receiving such assistance. All other patients were allocated to the group not desiring help. Independent variable was the wish to receive help (yes vs no). Raw scores of the seven GRISS subscales were entered as dependent variables. A significant group effect was found of the combined subscale scores (F (7, 49)=4.03; P=0.001; η2=0.365; power=0.971). In a subsequent univariate comparison, the GRISS erectile dysfunction subscale proved to differentiate between men desiring and not desiring help. Compared to men, who did not want to receive further assistance, scores on this subscale were significantly higher (two-tailed, with Bonferroni correction of α) in men wanting to receive help. The dissatisfaction (F (1, 55)=4.31; P=0.043; η2=0.073; power=0.245) and infrequency subscales (F (1, 55)=5.53; P=0.022; η2=0.091; power=0.336) were found to have marginally significant contributions to the main effect.
Logistic regression and clinical decision analyses were performed to evaluate the ability to identify patients suffering from sexual dysfunctions by means of their individual GRISS subscale scores. The results of this evaluation are displayed in Table 3 he presence of erectile dysfunctions. The logistic regressions and clinical decision analyses on male subjects' data were performed between observed presence of a sexual dysfunction as dependent variable and GRISS erectile dysfunction, dissatisfaction, and infrequency as independent variables.
A significant regression model was produced after backward removal of nonsignificant subscales (Model χ2(1)=10.05; P=0.002). Significantly contributing GRISS scales were erectile dysfunction and dissatisfaction. The optimal combination of subscales correctly predicted the overall presence or absence of sexual dysfunctions in 82.1% of the cases in the derivation sample and 79.3% in the crossvalidation sample. Test sensitivity and specificity of this regression model were 63.6% in the derivation sample half (91.7% in the crossvalidation sample half) and 94.1% (70.6%), respectively, while predictive values positive and negative were 87.5% (68.8%) and 80.0% (92.3%). This finding implies that the odds are for a new male urology patient with high scores on the erectile dysfunction and dissatisfaction subscales of the GRISS to admit the presence of a sexual dysfunction to the urologist, either at his own initiative or when he is directly questioned on this matter.
A logistic regression analysis was then performed to investigate whether desire for help could also be predicted from individual GRISS scores. The variable to be predicted was the wish to receive help (yes vs no). GRISS erectile dysfunction, dissatisfaction, and infrequency subscales were entered as predictor variables. A significant regression model was produced after backward removal of nonsignificant subscales (Model χ2(1)=10.05; P=0.002). Significantly contributing GRISS scales were erectile dysfunction and dissatisfaction, (see Table 3). The optimal combination of subscales correctly predicted the overall presence or absence of the patient's wish to receive help for his sexual dysfunction in 82.1% of the cases in the derivation sample and 72.4% in the crossvalidation sample. Test sensitivity and specificity of this regression model were 63.6% (90.0%) and 94.1% (63.2%), respectively, while predictive values positive and negative were 87.5% (56.3%) and 80.0% (92.3%). This finding implies that the odds are for a new male urology patient with high scores on the erectile dysfunction and dissatisfaction subscales of the GRISS to reply affirmatively that he wants to receive professional help for his sexual dysfunction when the urologist directly questions him on this matter. Low scores on these subscales predict absence of need for professional assistance.
An exploratory analysis was conducted to examine the predictive abilities of the GRISS for the presence or absence of erectile dysfunction, which was found to be the most commonly reported sexual dysfunction in this sample (30%). All GRISS subscales were entered at the first step. After backward removal of nonsignificant subscales, a significant regression model was produced (Model χ2(1)=26.68; P=0.000). Significantly contributing GRISS scales were erectile dysfunction (B=0.47; R=0.40) and dissatisfaction (B=0.23; R=0.17). A constant (−7.72) was added to this equation. The optimal combination of subscales correctly predicted the overall presence or absence of erectile dysfunction in 87.3% of the cases. Test sensitivity and specificity of this regression model were 75.0 and 92.3%, respectively, while predictive values positive and negative were 80.0 and 90.0%. This finding implies that the same GRISS subscales, that were found to be predictive of sexual dysfunctions in general, can be used to more specifically detect erectile dysfunctions in new patients. When only the erectile dysfunction subscale (B=0.44; R=0.42) was entered in the logistic regression, an equally significant regression model appeared (Model χ2(1)=22.45; P=0.000). However, although overall correct prediction remained at 87.3%, sensitivity fell to 68.8%, whereas specificity rose to 94.9%. Predictive values positive and negative were 84.6 and 88.1%. Thus, the dissatisfaction subscale was found to be relevant in addition to the erectile dysfunction subscale to improve sensitivity. This finding converges with the intuition that the presence of erectile dysfunctions alone, when not giving rise to dissatisfaction or distress, is an insufficient basis for sexual complaint presentation.
In our small sample of male patients of a university hospital's urological outpatient clinic, a rather high prevalence of sexual dysfunctions (40.4%) was found. The larger part (30%) only reported these dysfunctions after the urologist specifically informed whether such problems were experienced. Without these explicit questions only 10.5% of the patients self-presented any sexual dysfunctions. This finding corroborates earlier contentions that the clinician should actively ask for the existence of sexual dysfunctions, because patients tend not to forward their sexual dysfunctions all by themselves despite their apparent need for professional assistance.15,16 Our prevalence findings are in line with Benet and Melman,12 but exceed the 10.6% prevalence that Slatford and Currie11 found in new patients of a genitourinary clinic. The divergent findings with the latter study may be explained by selection differences. Patients in the Slatford and Currie11 study had a lower mean age (75% were between 20 and 34 y of age), and most were single, whereas the mean age of the men in the present study was around 50 y of age, and 95% were in a steady relationship with their female partners.11 Of patients in our sample who presented sexual dysfunctions, 20 (87%) either wanted to receive professional assistance or already received help.
GRISS subscales were found to differentiate between patients with and without sexual dysfunctions. Specifically, men with sexual dysfunctions had higher scores on the subscales of erectile dysfunction, dissatisfaction, and infrequency. Moreover, the ability of the GRISS scales to predict the presence of sexual dysfunctions in individual men was also found satisfactory, with predictive values positive and negative ranging from 68 to 92% in both derivation and crossvalidation sample halves. The findings in the present urological sample thus converge with earlier GRISS validation studies in other samples.9 Moreover, the finding that both problems with erectile dysfunction and experienced dissatisfaction contributed to prediction converges with the intuition that the mere presence of erectile dysfunctioning is insufficient to produce patients' self-evaluation as having a relevant sexual dysfunction. The accuracy of prediction of the patients' wish to receive professional assistance worked out equally well with predictive values positive and negative ranging from 56 to 92% in both sample halves. The presence in patients of the most commonly reported sexual dysfunction in our sample, erectile dysfunction, could be predicted from individual scores on the erectile dysfunction and dissatisfaction subscales. This conclusion, however, can only be drawn with caution because the regression model could not be crossvalidated, because of limited sample size. The small number of other categories of sexual dysfunction in our sample limits the prediction of their presence by means of GRISS scores. The small sample might also have raised barriers to the validation of other GRISS subscales for the prediction of the presence of sexual dysfunction. The erectile dysfunction and dissatisfaction subscales that were found to contribute significantly to the above-mentioned prediction models, together comprise eight questions. Although some gain of time might be obtained by omitting other items from the questionnaire, it remains an empirical issue whether a shortened GRISS would yield equally strong predictions.
For ethical reasons no demographic, urological, or sexual data could be collected on nonresponders. This, of course, entails a risk of volunteer bias of our data and conclusions. However, the results of earlier research work on this issue suggest that the generalizability of data from the present study does not necessarily suffer from self-selection bias.17,18
In sum, it appears warranted to recommend the use of the GRISS as a routine instrument in urological outpatient clinics to detect patients with sexual dysfunctions, as well as to identify those patients who are in need of professional assistance or guidance. The instrument can be of help to call forth the clinician's attention to a patient's sexual dysfunction, which then can further be explored.
Rust J, Golombok S . The Golombok Rust Inventory of Sexual Satisfaction (GRISS). Br J Clin Psychol 1985; 24: 63–64.
Rust J, Golombok S . The GRISS: a psychometric instrument for the assessment of sexual dysfunction. Arch Sex Behav 1986; 15: 157–165.
Rust J, Golombok S . The Golombok Rust Inventory of Sexual Satisfaction. Manual. NFER-Nelson: Windsor, 1986.
Frenken J, Vennix P . Sexuality Experiences Scales (SES); Manual. Lisse: Swets & Zeitlinger, 1981.
Vennix P . De ILKS-gedachte en Andere Operationalisaties. NISSO: Zeist, 1983.
Ter Kuile MM, Vroege JV, van Lankveld JJ . The Golombok Rust Inventory of Sexual Satisfaction; Dutch Translation and Adaptations. Leiden University Medical Centre: Leiden, 1993.
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 Marit Ther 1999; 25: 59–71.
Collier JL . The use of the GRIMS and the GRISS in the assessment and outcome of sexual problems: are questionnaires of more value than a clinical interview? Sex Marit Ther 1989; 4: 11–16.
Van Lankveld JJDM, Ter Kuile MM . The Golombok Rust Inventory of Sexual Satisfaction (GRISS): predictive validity and construct validity in a Dutch population. Pers Indiv Diff 1999; 26: 1005–1023.
Association AP . Diagnostic and Statistical Manual of Mental Disorders, 4th edn (DSM-IV). APA: Washington DC, 1994.
Slatford K, Currie C . Prevalence of psychosexual problems in patients attending a genitourinary clinic. Br J Vener Dis 1984; 60: 398–401.
Benet AE, Melman A . The epidemiology of erectile dysfunction. Urol Clin N-Amer 1995; 22: 699–709.
Hosmer DW, Lemeshow S . Applied Logistic Regression. John Wiley & Sons: New York, 1989.
Tabachnick BG, Fidell LS . Using Multivariate Statistics. HarperCollins: New York, 1989.
Bachmann GA, Lweiblum SR, Grill J . Brief sexual inquiry in gynecologic practice. Obst Gynec 1989; 73: 425–427.
Van Lankveld JJ, Ter Kuile MM, Kenter GG, Van Hall EV, Weijenborg PTM . Seksuele klachten en ervaringen met seksueel en fysiek geweld bij gynaecologische patiënten. Ned Tijdschr Geneesk 1996; 140: 1903–1906.
Trivedi N, Sabini J . Volunteer bias, sexuality, and personality. Arch Sex Behav 1998; 27: 181–195.
Van Lankveld JJDM, Grotjohann Y, Van Lokven BME, Everaerd W . Characteristics of couples applying for bibliotherapy via different recruitment strategies: a multivariate comparison. J Sex Marit Ther 1999; 25: 197–209.
About this article
Cite this article
van Lankveld, J., van Koeveringe, G. 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 15, 110–116 (2003). https://doi.org/10.1038/sj.ijir.3900966
- self-report questionnaire
- sexual assessment
- sexual satisfaction
- sexual dysfunction
- predictive validity
- clinical decision analysis
International Journal of Sexual Health (2019)
Validity and Reliability of the Golombok-Rust Inventory of Sexual Satisfaction (GRISS) in Patients With Type 2 Diabetes
Journal of Sex & Marital Therapy (2019)
The Journal of Sex Research (2018)
Journal of Sex & Marital Therapy (2015)
The Journal of Sexual Medicine (2015)