Content uploaded by Ellen Laan
Author content
All content in this area was uploaded by Ellen Laan on Feb 09, 2016
Content may be subject to copyright.
Journal of Sex & Marital Therapy, 32:289–304, 2006
Copyright © Taylor & Francis Group, LLC
ISSN: 0092-623X print
DOI: 10.1080/00926230600666261
The Female Sexual Function Index (FSFI)
and the Female Sexual Distress Scale (FSDS):
Psychometric Properties within
a Dutch Population
MONIEK M. TER KUILE and MARIEKE BRAUER
Department of Psychosomatic Gynaecology and Sexology, Leiden University Medical Center,
Leiden, The Netherlands
ELLEN LAAN
Department of Sexology and Psychosomatic Obstetrics and Gynecology,
Academic Medical Center, University of Amsterdam, The Netherlands
The aim of the present study was to investigate the psychomet-
ric properties of the Female Sexual Function Index (FSFI; Rosen
et al., 2000) and the Female Sexual Distress Scale (FSDS; Derogatis,
Rosen, Leiblum, Burnett, & Heiman, 2002) within a Dutch popu-
lation of approximately 350 women with and without sexual com-
plaints. The main conclusions of this study are that the multidimen-
sional structure of the FSFI and the unidimensional structure of the
FSDS are fairly well replicated in a Dutch sample. The amount of
variance explained by confirmatory and exploratory factor analy-
ses was good. Internal consistency and stability of the FSFI and its
subscales and the FSDS are satisfactory to good, and the subscales
are reasonably stable across demographic variables. The discrim-
inant validity and the ability of the scales to predict the presence
or absence of sexual complaints was excellent. The convergent and
divergent construct validity of the FSFI and the FSDS was good.
These results support the reliability and psychometric validity of
the FSFI and the FSDS in the assessment of dimensions of female
sexual functioning and sexual distress in clinical and nonclinical
samples.
Address correspondence to Moniek M. ter Kuile, Outpatient Clinic for Psychosomatic Gy-
naecology and Sexology, Poortgebouw-Zuid, Postbus 9600, Leiden University Medical Center,
PO Box 9600, 2300 RC Leiden, The Netherlands. E-mail: m.m.ter kuile@lumc.nl
289
290 M. M. ter Kuile et al.
Research into the diagnosis and treatment of female sexual dysfunction (FSD)
has rapidly advanced over the past few years. This has created an increased
need for psychometric instruments for diagnosing FSD and for effectively
monitoring treatment-induced changes (Meston & Derogatis, 2002). Some
new inventories for assessing FSD have been published recently and seem
promising in the assessment of FSD (Daker-White, 2002). One of the recent
developed inventories for assessing FSD is the Female Sexual Function Index
(FSFI; Rosen et al., 2000). The FSFI is a multidimensional self-report question-
naire for assessing sexual function in women (Rosen et al., 2000); it consists of
19 items that assess sexual desire, arousal, lubrication, orgasm, satisfaction,
and pain. The factor structure and internal consistency have been exam-
ined and found to be satisfactory. Discriminant validity testing confirmed the
ability to discriminate between women with and without sexual complaints
(Meston, 2003; Rosen et al., 2000; Wiegel, Meston, & Rosen, 2005).
Recent consensus-based characterizations of FSD have emphasized per-
sonal distress as an essential component of the definition of FSD (Basson
et al., 2000). Because none of the other inventories for sexual dysfunction
(LIT) (Daker-White, 2002) also assesses personal distress, the Female Sexual
Distress Scale (FSDS) has developed (LIT). The FDS is a 12 item self-report
assessment questionnaire. Derogatis et al. (2002) developed a 12-item self-
report questionnaire to assess sexuality related personal distress. Principal
components analysis identified one factor accounting for 73% of the total vari-
ance. The consistency and test-retest reliability coefficients were satisfactory.
The scale showed a high degree of discriminative ability in distinguishing
between sexually dysfunctional and sexually functional women and proved
to be sensitive for treatment response. Moderate positive correlations were
observed with other conceptually related measures of nonsexual distress,
which supported the construct validity of the scale (Derogatis et al., 2002).
Overall, the FSDS appears to be a valid and reliable measure for assessing
sexuality related personal distress in women.
The primary objective of the present study is to investigate the psy-
chometric proporties of the FSFI and the FSDS within a Dutch population
regarding (a) factor structure, internal consistency, and stability; (b) the dis-
criminative value of the FSFI dimensions and the FSDS for the presence of
sexual problems; (c) the divergent and convergent validity of the FSFI di-
mensions and the FSDS.
METHOD
Subjects
The participants consisted of 342 women: 234 women with a sexual com-
plaint (FSD group) and 108 women without a sexual complaint (control
group). To be included in this study, the women needed to be in a hetero-
sexual relationship. The sexual complaint group consisted of women with a
FSFI and FSDS within a Dutch Population 291
sexual problem who sought therapy at an outpatient clinic for sexology of
a university medical hospital. In this patient sample, 173 (74%) had a sexual
pain disorder, 43 (18%) had a hypoactive desire disorder, and 17 (8%) of the
sample presented with different sexual problems (lubrication, orgasm, and
dissatisfaction problems). The mean duration of the sexual complaint was
4.6±4.7years. The women completed the questionnaires at the end of the
first visit. The second group consisted of women with no sexual complaints
who responded to an advertisement in a local newspaper. We conducted the
initial screening on inclusion and exclusion criteria during a first telephone
contact. Subsequently, we invited who eligible and interested women to on
outpatient clinic of sexology for a sexual interview and to complete the FSFI
and FSDS. They received a compensatory fee of 7.50 Euro ($12.40 U.S. dol-
lars). A subsample of the control group (N=75) was invited to participate
also in the test-retest reliability of the FSDS and FSFI. If they agreed, they re-
ceived an envelope that contained the FSFI and FSDS and a return envelope.
Detailed demographic data for the control group and the FSD group
are provided in Table 1. We found that the women in the FSD group
were significantly older (t=−2.5,p=0.01) and had a significantly longer
TABLE 1. Subjects Characteristics for Women with and without Sexual Complaints
Characteristic
With sexual
problem
(n=234)
M(±SD)
Without sexual
problem
(n=108)
M(±SD)
Total
(N=342)
M(±SD)
T-value,
Z-value or
Chi-square1
Age women (yrs) 30.0 (10.0) 27.1 (9.4) 29.1 (9.9) −2.5a
Duration relationship (yrs) 7.7 (9.3) 4.5 (6.7) 6.7 (8.7) −3.2a
Duration complain 4.6 (4.7) —
Recruitment/referral (%)
Self-referred 2 100 33
Family doctor 51 34
Gynecologist 42 29
Other 5 4
Education (%) 0.15
Primary school—lower secondary 18 4 14
Higher secondary school 46 71 54
College-university 36 25 32
Living with partner (% yes) 76 38 64 46.7b
Has a child (% yes) 13 16 21 2.4
Sexual and physical abuse (%) 0.8
Any sexual abuse 29 24 28
Genital touched 25 20 23
Coerced to sexual acts 15 10 13
Coerced to intercourse 11 8 10
Physical assault 16 19 17 0.5
Sexually active, last month (% yes) 75 98 82 25.5b
Note. Maximum nis indicated; Outpatient clinic =women who solicited for therapy at an outpatient
clinic for sexology of a university medical hospital; M=Mean; SD =Standard deviation; yrs =years;
1=Observed two-tailed significance; ap<0.05; bp<0.001.
292 M. M. ter Kuile et al.
partner relationship duration (t=−3.2,p=0.002) than women in the con-
trol group. We found the effect size of the differences to be small (Cohen,
1977). Moreover, a significantly larger percentage of the women in the FSD
group lived with their partner (χ2=46.7,p<.001) and reported not to have
been sexually active in the last month (25% versus 2%) (χ2=25.5,p<.001)
compared to women in the control group. We found no differences between
the FSD group and control group with regard to education level, having
children, and a history of sexual or physical abuse.
Measures
The original versions of the FSFI and the FSDS were translated into Dutch
by three independent persons. When the translators disagreed, the final text
was arrived at after an extended discussion. The final version was back-
translated by a native English speaker; we observed no substantial loss of
information. On the basis of some minor differences between the original
and retranslated versions of the questionnaires, we made additional changes
in the Dutch translation. Higher scores on the FSFI indicate better sexual
functioning, and higher scores on the FSDS indicate more sexual distress.
MEASURES USED IN THE CONSTRUCT VALIDATION
We collected data only in the FSD group (N=234) for the divergent an
convergent construct. We used the following measures.
The Golombok Rust Inventory of Sexual Satisfaction (GRISS; Rust &
Golombok, 1985; Rust & Golombok, 1986) contains 28 items and covers the
most frequently occurring sexual complaints of heterosexual persons with a
steady partner. It provides an overall score for the person’s satisfaction with
sexual functioning within the relationship. In addition, seven subscale scores
can be derived: anorgasmia, vaginismus, in frequency of sexual contact, sex-
ual noncommunication, dissatisfaction, nonsensuality, and avoidance of sex.
The psychometric qualities of original GRISS (Rust & Golombok, 1985, 1986),
and the Dutch version were found to be satisfactory (Ter Kuile, Van Lankveld,
Kalkhoven, & Van Egmond, 1999; Van Lankveld & Ter Kuile, 1999). Higher
scores indicate more dissatisfaction.
The Maudsley Marital Questionnaire (MMQ; Crowe, 1978) is a 20-item
self-report instrument measuring dissatisfaction with the general relationship,
with the sexual relationship, and with life in general. The MMQ has shown
good reliability and validity. The psychometric qualities of the Dutch version
of the MMQ were also found to be satisfactory (Arrindell, Boelens, & Lambert,
1983). Higher scores represent larger dissatisfaction.
To measure psychological distress, we used the Symptom Checklist-90
(SCL-90; Derogatis & Cleary, 1977). The Dutch version of the SCL-90 (Arrindell
&Ettema, 1981, 1986) was used as an index of psychological distress. The
FSFI and FSDS within a Dutch Population 293
psychometric qualities of the Dutch version were found to be satisfactory.
Higher scores represent greater psychological distress.
STATISTICAL ANALYSIS
To investigate how much variance of the total variance in FSFI scores was
accounted for by the six components of Rosen et al. (2000), we performed
a simultaneous component analysis (SCA) (Nunnally, 1978; Kiers, 1990) on
the correlation matrix of the 19 items for the women in the FSD group and
for the women in the control group separately. In an SCA analysis, the com-
ponents are computed as sums of variables that belong to a particular subset
only. In this study, the allocation of the items is based on the six components
of Rosen et al. (2000): that is, a matrix with binary elements only, such that
each variable has a weight of 1 on the component of the group of which
it is supposed to be a member (e.g., desire) and 0 for the other 5 groups.
We performed actual calculations involved using the SCA computer program
(Kiers, 1990). To investigate how much variance of the total variance in FSFD
scores was accounted for by one factor, we conducted a principal component
analysis (PCA) on the correlation matrix of the 12 items in both groups sep-
arately. We determined homogeneity of the subscales by calculating mean
interitem correlations and internal consistency by Cronbach’s coefficient al-
pha. We assessed the stability of the FSFI and the FSDS in the control group
by calculating the test-retest reliability. To test the discriminant validity of the
FSFI subscales and FSDS against the criterion of the presence or absence of
sexual problems, we made univariate comparisons by means of t-tests for in-
dependent samples. For the evaluation of predictive validity, we performed
logistic regression analysis using the maximum likelihood method (Hosmer
&Lemeshow, 1989). On the basis of the model produced by the logistic re-
gression, we performed clinical decision analysis to examine the classificatory
qualities of the scales in the study sample. The scale scores, as we calculated
by a logistic regression analysis, demonstrate sensitivity and specificity, which
indicates the ability of the instruments to correctly demonstrate the presence
or absence of sexual problems in this study sample. Of further importance
for the qualifications of the FSFI and FSDS as clinical screenings instruments
are the positive and negative predictive values, which we calculated on the
basis of the same logistic regression model. A positive predictive value is
defined as the proportion of positive predictions (the subject is predicted by
the scale scores to have a sexual problem), which coincides with the ob-
served presence of a sexual problem. Negative predictive value is defined as
the proportion of negative predictions (the subject is predicted by the scale
scores not to have a sexual problem), which coincides with the observed
absence of a sexual problem. We further examined the diagnostic properties
of the subscales of the FSFI and FSDS using receiver operating characteris-
tics (ROC) analyses, which provide the areas under the curves (AUCs). To
create the ROC of the FSFI subscales together, we modeled the predicted
294 M. M. ter Kuile et al.
TABLE 2. Mean (M) and Standard Deviations (SD) for the FSFI Subscales, FSFI Total and the
FSDS Scores and Discriminant Statistics for Women with and without Sexual Complaints
With sexual Without sexual
problem N=234 problem N=108
MSDMSD T 1d
FSFI subscales
Desire 4.5 2.0 6.7 1.3 11.9a1.2
Arousal 10.1 6.3 17.6 2.8 15.0a1.4
Lubrication 11.4 7.4 19.0 3.2 13.1a1.2
Orgasm 8.1 5.7 12.8 2.8 9.9a1.0
Satisfaction 8.0 4.0 13.4 2.1 16.6a1.5
Pain 2.0 4.3 14.2 1.9 33.8a3.3
FSFI total 45.4 24.0 83.7 10.7 20.3a1.9
FSDS 27.3 10.8 5.1 6.4 −23.5a2.3
Note. FSFI =Female Sexual Function Index (Rosen et al., 2000); FSDS =Female Sexual Distress Scale
(Derogatis et al., 2002); Cases with missing data were excluded for analysis by analysis; Maximum Nper
subgroup is indicated; 1 =Observed two-tailed significance; d=Cohen’s d: effect size; a=p<0.001.
probabilities of the presence of a sexual dysfunction that calculated through
logistic regression analysis.
We further studied the divergent and convergent contruct validity using
a higher order principal components analysis with varimax rotation of the
scores on the FSFI subscales and FSDS together with the scores on (sub-
scales of) standardized psychometric instruments measuring theoretically re-
lated constructs: sexual dissatisfaction (GRISS) marital, sexual and general
life maladjustment (MMQ), and psychological distress (SCL-90).
RESULTS
FSFI
ASCA analysis revealed that in the FSD group1the a priori matrix in binary
form representing the six factor structure of Rosen et al. (2000) accounted
for 88.6% of the total variance; that is a loss of only 0.7% varience, compared
with a 6-factor solution on a basis of an explorative PCA (81.6%). More-
over, the SCA accounted for 80.5% of the variance in the control group. That
is a loss of 1.1% variance compared with a 6-factor solution on the basis
of an exploratory PCA (81.6%). Within both groups, except for one item,
each item had the highest component loading on the a priori component
(range .27–.82), and component loadings on the other five components were
1Because 23% of the women in the FSD group reported not to be sexually active in the
past month, we repeated the SCA analysis in the group of women who reported that they had
been sexually active in the past month. The results of this secondary analysis were comparable
with the results reported here for the whole FSD group.
FSFI and FSDS within a Dutch Population 295
considerably lower. In both groups, item 14 did not have the highest com-
ponent loading on the a priori component (satisfaction) but had the highest
loading on the arousal component. Because the six-factor solution of the
original FSFI replicated fairly well, we constructed six subscales comprising
the same items as originally selected (Rosen et al., 2000). In Table 2, the
means and the standard deviations of the six subscale scores and the total
score are presented for women with and women without sexual complaints.
As can be seen in Table 3, Cronbach’s alpha’s (range .72–.98) of the
subscales were satisfactory (>.70; Nunnally, 1978), and all items manifest a
TABLE 3. Reliablility Coefficients for the FSFI Subscales and the FSDS for Women with Sexual
Complaints (FSD group: N=232) and Women without Sexual Complaints (Control group:
N=108)
Subscale
group Alpa Range rr
test-retest
FSFI desire
Control group .72 .56 .72
FSD group .90 .82
Total group .90 .81
Aroursal
Control group .88 .57–.73 .85
FSD group .96 .78–.89
Total group .96 .83–.90
Lubrication
Control group .96 .80–.94 .82
FSD group .96 .81–.91
Total group .97 .85–.93
Orgasm
Control group .83 .51–.82 .71
FSD group .95 .84–.90
Total group .95 .82–.90
Satisfaction
Control group .80 .41–.88 .90
FSD group .81 .50–.86
Total group .87 .59–.91
Pain
Control group .84 .53–.78 .97
FSD group .94 .82–.88
Total group .98 .93–.95
Total score
Control group .93 .37–.82 .93
FSD group .95 .26–.90
Total group .97 .34–.95
FSDS
Control group .93 .32–.85 .93
FSD group .93 .26–.74
Total group .97 .56–.85
Note. FSFI =Female Sexual Function Index (Rosen et al., 2000). FSDS =Female Sexual Distress Scale
(Derogatis et al., 2002); FSD group =female sexual dysfunction group; Alpha =Cronbach’s coefficient
alpha; Range rii =iter- item-correlation; rtest-restest Pearon product moment correlation coefficient between
test-retest in the control sample (N=68); Maximum Nper subgroup is indicated.
296 M. M. ter Kuile et al.
TABLE 4. Product-Moment Intercorrelation Coefficients of FSFI subscales
FSFI Subscales
Desire Arousal Lubrication Orgasm Satisfaction Pain
Total group (N=342)
Desire —
Arousal .73 —
Lubrication .59 .88 —
Orgasm .55 .82 .79 —
Satisfaction .69 .84 .75 .70 —
Pain .42 .54 .53 .42 .56 —
FSD group (N=234)
Desire —
Arousal .66 —
Lubrication .48 .86 —
Orgasm .46 .81 .76 —
Satisfaction .59 .79 .67 .67 —
Pain .03#.19 .21 .20 .21 —
Control group (N=108)
Desire —
Arousal .48 —
Lubrication .37 .44 —
Orgasm .26 .30 .24 —
Satisfaction .38 .52 .37 .23 —
Pain .04#.04#.16#.09#.17#—
Note. FSFI =Female Sexual Function Index (Rosen et al., 2000); Cases with missing data were ex-
cluded for analysis by analysis; Maximum Nper subgroup is indicated; Observed two-tailed significance;
#=p>0.05 (Not statically significant).
significant and positive correlation with the other items of a subscale (range
.41–.95). The correlation-coefficients between pre and retest, ranging from
.71 to .90, were never more than 0.20 lower than Cronbach’s alpha of a
subscale (Nunnally, 1978). We observed no significant differences between
test and retest. These results indicate that the FSFI subscales have appropri-
ate levels of stability over a period of 2.8 weeks of time (SD =1.7; range
1–8 weeks).
We computed interscale correlations for both groups as an internal cri-
terion for the validity of the six subscales. As can be seen in Table 4, nearly
all the subscales were significantly and positively correlated, with medium
to large effect sizes (Cohen, 1977). For the purpose of interpretation, Cohen
(1977) considered |.10|<r<|.30|as small, |.30|<r<|.50|as medium, and
r>|.50|as large. These results indicate that the FSFI subscales do not mea-
sure totally independent constructs.
The significant correlations (−.12 <r<.32), which we found between
the FSFI subscales on the one hand and demographic and complaint variables
on the other, had nearly all a small effect size (see Table 4). Therefore, it
can be concluded that the FSFI subscales scores are rather independent of
biographic and complaint variables.
FSFI and FSDS within a Dutch Population 297
TABLE 5. Product-Moment Correlation Coefficients between the FSFI subscales and the FSDS
and Biographic and Complaint Variables within the Total Group (N=342)
FSFI–subscales
Desire Arousal Lubrication Orgasm Satisfaction Pain FSDS
Age −.18b−.26b−.30b−.20b−.22b−.10 .09
Education level .08 .10 .10 .08 .09 .08 −.15a
Relationship duration −.17a−.24b−.31b−.19b−.22b−.12a.13a
Duration of complaint1−.32b−.25b−.17a−.20a−.17a−.10 .08
Note. FSFI =Female Sexual Function Index (Rosen et al., 2000); FSD =Female Sexual Distress; Cases
with missing data were excluded for analysis by analysis; Maximum Nis indicated.
1N=169; Observed two-tailed significance.
ap<0.05.
bp<0.01.
All the six FSFI subscales scores and the FSFI-total score were signifi-
cantly lower in the sample of women with a sexual complaint (FSD group)
than in the sample of women without sexual complaints (control group see
Table 2). Medium to large effect sizes were found for the differences in the
mean FSFI subscale scores between women with and without sexual com-
plaints (see Table 2).
We performed a logistic regression analysis and a clinical decision anal-
ysis with the presence of a sexual complaint as criterion and the six FSFI
subscales as independent variables. Three hundred thirty-five cases were
entered in the analysis. Inclusion of the six FSFI subscales did produce
asignificant regression model (Model: χ3(8) =102.0p<.001). The fol-
lowing FSFI scales significantly contributed to the model: pain (Wald (1)
=25.7,p<.001), lubrication (Wald (1) =12.8,p<.001), and satisfaction
(Wald (1) =15.5,p<.001). This model correctly predicted the overall pres-
ence or absence of sexual problems in 93.7% of the cases (See Table 6).
Sensitivity of the combined use of the six FSFI subscales was 94.7%, and
the specificity was 91.6%. The positive and negative predictive values of
the combined use of the six FSFS subscales were, respectively, 96.0% and
89.1%. The ROC curve for the six FSFI subscales had an area under the
curve of 0.98 (95% CI: 0.96–0.99). A subsequent logistic regression analysis
to predict the presence of a sexual complaint versus no sexual complaint on
the basis of only the FSFI total score (Model χ2(8) =100.0p<.001) cor-
rectly classified 93.8% of the cases (sensitivity: 96.5%; specificity: 88.0%). The
positive and negative predictive values of the FSFI total score were, respec-
tively, 94.5% and 92.2%. The ROC curve of the FSFS total score had an area
under the curve of 0.97 (95% CI: 0.95–0.99). It can be concluded that the
discriminative value of the FSFI total score is as good as that of the com-
bined use of the six subscales of the FSFI, suggesting that the use of only
the total score may be more convenient without any loss of predictive value
(see Table 6).
298 M. M. ter Kuile et al.
TABLE 6. Overall Correct Prediction of the Presence Versus of the Absence of Sexual Problems
of the Six- FSFI Subscales Scores, FSFI Total Score and the FSDS Score: Test-Sensitivity, Test-
Specificity, Positive and Negative Predictive Values, and the Areas under the Curve
6FSFI subscales FSFI total score FSDS score
Number of subjects 335 335 339
Overall correct prediction 93.7 93.8 87.3
Sensitivity 94.7 96.5 92.2
Specificity 91.6 88.0 76.6
Positive predictive value 96.0 94.5 89.5
Negative predictive value 89.1 92.2 82.0
Area under the curve (95% CI) 0.98 (0.96–0.99) 0.97 (0.95–0.99) 0.96 (0.94–0.98)
Note. FSFI =Female Sexual Function Index (Rosen et al., 2000); FSDS =Female Sexual Distress Scale
(Derogatis et al., 2002); CI =Confidence Interval.
FSDS
We performed a PCA with a one-factor solution2on the correlation matrix
of the 12 FSDS items within the sample of women with sexual complaints
(FSD group) and within the sample of women without sexual complaints
(control group) separately. The one-factor solution accounted for 56.9% of
the total variance in the FSD group and 57.8% of the total variance in the
control group. In both groups, all of the 12 items had a component loading
in the range of .65–.81. Because the one-factor solution of the original FSDS
replicated fairly well, we computed a FSDS score by adding the 12 items.
In Table 2, the mean and the standard deviation of the FSDS score is pre-
sented for women with and women without sexual complaints. As can be
seen in Table 3, Cronbach’s alpha (.93) of the scale was good (Nunnally,
1978), and all items manifested a significant and positive correlation with the
other items of the scale (range .39–.77). Moreover, the correlation coefficient
between test and retest was .93 and comparable with Cronbach’s alpha of the
FSDS (Nunnally, 1978). This result indicates that the FSDS has an appropriate
level of stability over a period of approximately 3 weeks. FSDS scores were
significantly higher, with a large effect size, in the sample of women with
sexual complaints than in the sample of women without sexual complaints
(see Table 2). We performed a clinical decision analysis for the FSDS score
in a sample of 339 subjects. The FSDS score correctly predicted the overall
presence or absence of sexual problems in 87.3% of the cases. Sensitivity of
the FSDS was 92.2%, and specificity was 76.6%. The positive and negative
predictive values of the FSDS were 89.5% and 82.0%, respectively. The ROC
2The eigenvalue one criterion suggested in both groups a two-factor model that accounted
for 66% of the total variance in the FSD group and 68% of the total of variance in the control
group. However, these two factor models were conceptually not well interpretable. Further-
more, the eigenvalue one criterion, suggested a one-factor model in the total group (FSD +
control group) that accounted for 73% of the total variance.
FSFI and FSDS within a Dutch Population 299
curve of the FSDS had an area under the curve of .96 (95% CI: 0.94–0.98) (see
Table 6). These values are within the range of acceptable limits for diagnostic
classifications.
Because we found significant differences in demographic variables be-
tween the women with and without sexual complaints, we cannot exclude
that the predictive value of the FSFI and FSDS in predicting sexual complaints
is a direct consequence of these differences in demographic variables. In or-
der to test for the independent predictive power (yes versus no complaints)
of FSFI total score and the FSDS score beyond the effect of the demographic
variables, we also performed logistic regression analyses and clinical decision
analyses for the FSFI total score and FSDS score while controlling for demo-
graphic variables (age, duration of complaints, and living with the partner)
in the first step. In the first step, the three demographic variables did not
produce a significant regression model (Model; χ2(8) =7.1,p=.53). The
second step including the FSFI total score or the FSDS score produced sig-
nificant models, with comparable values as reported when not controlling
for the demographic variables. Therefore, we can conclude that the predic-
tive value of the FSFI total score and the FSDS score is not a direct conse-
quence of the differences in demographic variables found between the two
groups.
Construct Validity of the FSFI and FSDS
To inspect the association of the FSFI subscales and the FSDS with the GRISS
subscales, MMQ subscales, and SCL-90 total score, we performed a higher-
order principal components analysis with varimax rotation on the scores of
women of the FSD group. Cases with missing values were deleted listwise,
which reduced the number of patients in the analysis to 199.
A PCA analysis suggested a four-factor solution that accounted for 65.0%
of the total variance (see Table 7). We found that the first factor accounted
for 39.9% of the variance and interpreted it as sexual dysfunction.Wefound
high loadings on this factor for the four FSFI subscale, arousal, orgasm, lubri-
cation, and satisfaction and for the GRISS subscale anorgasmia. The second
factor, accounting for 10.9% of the total variance, is interpreted as sexual
maladjustment.Wefound high loadings on this factor for the GRISS sub-
scales noncommunication, avoidance, and non-sensuality and for the FSFI
subscale desire. The third factor, accounting for 7.8% of the total variance,
is interpreted as psychological distress.Weobserved high loadings on this
factor for the SCL-90 total score the MMQ subscale scores for general life
maladjustment and relationship maladjustment, and the FSDS. The fourth
factor, accounting for 6.3% of the total variance, is interpreted as sexual pain
disorder.Wefound high loadings on this factor for the FSFI subscale pain
and for the GRISS subscale vaginismus.
300 M. M. ter Kuile et al.
TABLE 7. Principal Component Analysis with Varimax Rotation of the FSFI
Subscales and FSDS Scale and Other Scales for Women with Sexual Com-
plaints. (N=199); Listwise Deletion of Cases with Missing Values. Eigenvalues
and Percentages of Explained Variance of the Unrotated Solution
Rotated factor matrix I II III IV
FSFI
Desire −.76∗
Aroursal .76∗−.50∗
Lubrication .76∗
Orgasm .89∗
Satisfaction .61∗−.52∗
Pain −.82∗
GRISS
Infrequency −.41 .53∗
Non-communication .73∗
Dissatisfaction −.48 .44
Avoidance .70∗
Non-sensuality −.43 .58∗
Vaginismus .77∗
Anorgasmia −.70∗
FSDS
Sexual dissatisfaction .51∗
MMQ
Marital maladjustment .66∗
Sexual maladjustment −.63∗.43
General life maladjustment .76∗
SCL-90
Psychological distress .81∗
Eigenvalue 7.2 2.0 1.4 1.1
% explained variance 39.9 10.9 7.8 6.3
Note. FSFI =Female Sexual Function Index (Rosen et al., 2000); FSD =Female Sexual
Distress; GRISS-Golombok Rust Inventory for Sexual Satisfaction (Rust & Golombok,
1985, 1986); MMQ-Maudsley Marital Questionnaire (Crowe, 1978); SCL-90-Symptom
Check List 90 (revised Derogatis & Cleary, 1977);
∗(rotated factor loadings of >.35 are reported) Rotated factor loadings of >.50.
DISCUSSION
The aim of the present study was to investigate the psychometric properties
of the FSFI and the FSDS in a Dutch population. The main conclusions of this
study are that the multidimensional structure of the FSFI and the unidimen-
sional structure of the FSDS are fairly well replicated in a Dutch sample. The
amount of variance explained by confirmatory, compared with exploratory,
factor analyses was good. Internal consistency and stability of the FSFI and
its subscales and of the FSDS are satisfactory to good, and the subscales
are reasonably stable across demographic variables. The discriminant valid-
ity and the ability of the scales to predict the presence or absence of sexual
complaints was excellent. The convergent and divergent construct validity of
the FSFI and the FSDS was good.
FSFI and FSDS within a Dutch Population 301
With respect to the dimensional structure of the FSFI a confirmatory
factor analysis revealed that the original 6-factor solution of Rosen et al.
(2000) resulted in an acceptable loss of only 1% in accounting for the
total variance in comparison to a 6-factor explorative PCA. The correla-
tions between most of the six subscales, however, were rather high. Be-
cause Cohen (1977) considered r>.5aslarge, four of the six subscales
(i.e. arousal, lubrication, orgasm and satisfaction) were highly interrelated
(range .70 <r<.80) in women with sexual complaints. Only the pain
subscale appears to measure a relatively independent dimension. A sec-
ond order factor analysis did support this finding, because four of the six
subscales of the FSFI loaded high on the same dimension. Also, in the
American validation studies, high intercorrelations were observed within dif-
ferent samples of women with and without sexual complaints. In fact, the
comorbidity of female sexual dysfunctions is well documented (Derogatis
&Conklin-Powers, 1998; Talakoub et al., 2002). The results of this study
and earlier studies support the idea that the separate categorical classifica-
tion of female sexual dysfunctions has to be reconsidered (Basson et al.,
2003).
The reliability of the subscales as assessed with Cronbach’s coeffi-
cient alpha was satisfactory to good, and the values found in this study
were comparable with the values found in American FSFI validation stud-
ies (Meston, 2003; Rosen et al., 2000; Wiegel et al., 2005). The test-retest
reliability of the FSFI scales was sufficient and comparable with the val-
ues found by Rosen et al. (2000). In addition, the FSFI subscales were
stable with respect to age, duration of the relationship, and level of
education.
The six FSFI subscales and the FSFI total score discriminated very well
between women with and without sexual problems. The six FSFI subscales
predicted the overall presence or absence of sexual problems in nearly 94%
the cases. The pain, lubrication, and satisfaction FSFI subscales significantly
contributed to this prediction. Furthermore, we found that the total FSFI
score predicted the overall presence or absence of sexual problems in nearly
94% the cases, as well. Because the discriminative value of the total FSFI
score is as good as that of the combined use of the six subscales of the
FSFI, it can be concluded that the use of only the total score may be more
convenient without any loss of predictive value. This finding is in line with
the U.S. validation studies (Wiegel et al., 2005). The overall FSFI total score
showed the best prediction profile, as reflected by the largest area under the
curve (AUE: .899), on the basis of which we propose to use the total FSFI
score for the prediction of sexual problems in general (Wiegel et al., 2005).
The individual FSFI subscales, however, may be more sensitive to treatment
intervention effects.
With respect to the unidimensional structure of the FSDS, an one-factor
analysis accounted for a satisfactory amount, 68% of total variance in the total
302 M. M. ter Kuile et al.
sample and 57%–58% of the total variance in the two samples separately.
The internal consistency and stability of the FSDS was very good and was
reasonably stable across demographic variables. The FSDS correctly predicted
the overall presence or absence of sexual problems in 87% of the cases
(sensitivity: 92.2%, specificity: 76.6%). This percentage is comparable with
the predictive values of the FSDS found in different populations in the United
States (88%–93%; Derogatis et al., 2002) and support the value of the FSDS
as a screening instrument.
The convergent and divergent construct validity of the FSFI subscales
and the FSDS was evaluated by higher-order principal component analysis.
Four of the FSFI subscales were shown to represent an independent dimen-
sion, which we labeled sexual disfunction and which was relatively inde-
pendent of (a) sexual maladjustment as measured by the different subscales
of the GRISS; (b) psychological and relational maladjustment as measured by
MMQ and SCL-90; and (c) sexual pain, as measured by the FSFI pain scale
and the GRISS vaginismus scale. The FSFI subscale desire loaded on the
same dimension as the GRISS subscales for noncommunication, avoidance
of sexuality, and non-sensuality and was labeled as sexual maladjustment.
Because the arousal and satisfaction FSFI subscales of the sexual disfunction
dimension also had relatively high loadings on the sexual maladjustment di-
mension, this pattern of loadings may indicate that sexual disfunction and
sexual maladjustment refer to interrelated dimensions. The FSFI subscale pain
loaded on the same dimension as the construct-related GRISS subscale for
vaginismus, and these scales seem to constitute a relatively independent con-
struct, with little overlap with other female sexual dysfunctions as measured
by the GRISS and the FSFI. Of note is that female sexual distress as measured
with the FSDS seems unrelated to sexual dysfunctions and sexual maladjust-
ment as measured by the FSFI and GRISS and loaded on the same dimension
as psychological distress (SCL-90) and marital distress (MMQ). These results
are in line with the study results of Bancroft, Loftus, and Long (2003), who,
in a national survey of heterosexual women, also found that the best predic-
tors of sexual distress were markers of general emotional wellbeing and the
emotional relationship with the partner. Physical aspects of sexual response
in women, including arousal, vaginal lubrication, and orgasm, were poor
predictors of female sexual distress. This finding indicates the importance
of the woman’s mental health and of the quality of the relationship with
the partner for the experience of distress about sexuality (Bancroft et al.,
2003).
In conclusion, the satisfactory psychometric properties of the FSFI and
FSDS seem fairly well replicated in a Dutch sample. Further research is nec-
essary to investigate the psychometric properties of the FSFI and the FSDS
in different populations with different sexual complaints and to investigate
sensitivity of the scales for treatment response.
FSFI and FSDS within a Dutch Population 303
REFERENCES
Arrindell, W. A., Boelens, W., & Lambert, H. (1983). On the psychometric properties
of the Maudsley Marital Questionnaire (MMQ)—Evaluation of self-ratings in dis-
tressed and normal volunteer couples based on the Dutch version. Personality
and Individual Differences, 4, 293–306.
Arrindell, W. A., & Ettema, H. (1981). Dimensionele structuur, betrouwbaarheid en
validiteit van de Nederlandse bewerking van de Symptom Check List (SCL-90):
Gegevens gebaseerd op een fobische en een ‘normale’ populatie [Dimensional
structure, realibility and validity of the Dutch adaptation of the Symptom Check
list (SCL-90): Data from a phobic and a ’normal’ population]. Nederlands Tijd-
schrift voor de Psychologie, 36, 77–108.
Arrindell, W. A., & Ettema, H. (1986). Handleiding bij een multidimensionele psy-
chopathologie indicator [Manual for multi-dimensional psychopathology indica-
tor]. Lisse: Swets & Zeitlinger.
Bancroft, J., Loftus, J., & Long, J. S. (2003). Distress about sex: A national survey of
women in heterosexual relationships. Archives of Sexual Behavior, 32, 193–208.
Basson, R., Berman, J., Burnett, A., Derogatis, L., Ferguson, D., Fourcroy, J.,
Goldstein, I., Graziottin, A., Heiman, J., Laan, E., Leiblum, S., Padma-Nathan,
H., Rosen, R., Segraves, K., Segraves, R.T., Shabsigh, R., Sipski, M., Wagner, G.,
& Whipple, B. (2000). Report of the international consensus development con-
ference on female sexual dysfunction: Definitions and classifications. Journal of
Urology, 163, 888–893.
Basson, R., Leiblum, S., Brotto, L., Derogatis, L., Fourcroy, J., Fugl-Meyer, K.,
Graziottin, A., Heiman, J. R., Laan, E., Meston, C., Schover, L., van Lankveld,
J., & Schultz, W. W. (2003). Definitions of women’s sexual dysfunction recon-
sidered: Advocating expansion and revision. Journal of Psychosomatic Obstetrics
and Gynecology, 24, 221–229.
Cohen, J. (1977). Statistical power analysis for the behavioral sciences. (Rev. ed.).
New York: Academic Press.
Crowe, M. J. (1983). Conjoint marital-therapy-controlled outcome study. Psychologi-
cal Medicine, 8, 623–636.
Daker-White, G. (2002). Reliable and valid self-report outcome measures in sex-
ual (dys)function: A systematic review. Archives of Sexual Behavior, 31, 197–
209.
Derogatis, L. R., & Cleary, P. A. (1977). Confirmation of dimensional structure of
SCL-90—Study in construct-validation. Journal of Clinical Psychology, 33, 981–
989.
Derogatis, L. R., & Conklin-Powers, B. (1998). Psychological assessment measures of
female sexual functioning in clinical trials. International Journal of Impotence
Research, 10 (Suppl. 2), S111–S116.
Derogatis, L. R., Rosen, R., Leiblum, S., Burnett, A., & Heiman, J. (2002). The female
sexual distress scale (FSDS): Initial validation of a standardized scale for assess-
ment of sexually related personal distress in women. Journal of Sex & Marital
Therapy, 28, 317–330.
Hosmer, D. W., & Lemeshow, S. (1989). Applied logistic regression. New York: John
Wiley & Sons.
304 M. M. ter Kuile et al.
Kiers, H. A. L. (1990). User’s manual. SCA: A program for simultaneous components
analysis of variables measured in two or more populations. Groningen: iec Pro-
gramma.
Meston, C. M. (2003). Validation of the Female Sexual Function Index (FSFI) in
women with female orgasmic disorder and in women with hypoactive sexual
desire disorder. Journal of Sex & Marital Therapy, 29, 39–46.
Meston, C. M., & Derogatis, L. R. (2002). Validated instruments for assessing female
sexual function. Journal of Sex & Marital Therapy, 28, 155–164.
Nunnally, J. C. (1978). Psychometric theory (2nd ed.). New York: McGraw-Hill.
Rosen, R., Brown, C., Heiman, J., Leiblum, S., Meston, C., Shabsigh, R., Ferguson, D.,
& D’Agostino, R. (2000). The Female Sexual Function Index (FSFI): A multidi-
mensional self-report instrument for the assessment of female sexual function.
Journal of Sex & Marital Therapy, 26, 191–208.
Rust, J., & Golombok, S. (1985). The Golombok-Rust Inventory of Sexual Satisfaction
(GRISS). British Journal of Clinical Psychology, 24, 63–64.
Rust, J., & Golombok, S. (1986). The GRISS—A psychometric instrument for the
assessment of sexual dysfunction. Archives of Sexual Behavior, 15, 157–165.
Talakoub, L., Munarriz, R., Hoag, L., Gioia, M., Flaherty, E., & Goldstein, I. (2002).
Epidemiological characteristics of 250 women with sexual dysfunction who pre-
sented for initial evaluation. Journal of Sex & Marital Therapy, 28, 217–224.
Ter Kuile, M. M., Van Lankveld, J. J. D. M., Kalkhoven, P., & Van Egmond, M. (1999).
The Golombok Rust Inventory of Sexual Satisfaction (GRISS): Psychometric
properties within a Dutch population. Journal of Sex & Marital Therapy, 25,
59–71.
Van Lankveld, J. J. D. M., & Ter Kuile, M. M. (1999). The Golombok Rust Inventory of
Sexual Satisfaction (GRISS): Predictive validity and construct validity in a Dutch
population. Personality and Individual Differences, 26, 1005–1023.
Wiegel, M., Meston, C., & Rosen, R. (2005). The Female Sexual Function Index (FSFI):
Cross-validation and development of clinical cutoff scores. Journal of Sex &
Marital Therapy, 31, 1–20.
