The present cross-sectional study evaluated female sexual dysfunction. The study population consisted of married women who were admitted during the study period for mother or child's health services in public and private health centers in Ilam, the Western Iran, from May to September 2014.

Sampling was conducted in multiple stages. In the first stage, the city was divided into six clusters. Subsequently, a public and a private clinic were randomly selected from each cluster. Overall, six public and six private health centers were selected from different regions of Ilam. In Iran, public health centers only attend to customers in the morning, whereas private health centers only serve customers in the afternoon. However, each week the collaborator midwife randomly attended a public health center each morning and private health centers each evening during the six days from Saturday to Thursday. Sampling was performed by a trained midwife. In order to gain access to all women that attended, the midwife was stationed at the customer reception. Eligible and willing participants were led to the rooms intended for study, and the required information was collected. In each center, consecutive sampling of eligible participants continued until the sample size was attained. Previous studies have reported the prevalence of female sexual dysfunction in women as 40%.¹⁶⁾ Considering P=0.4, d=0.05, α=5%, and 1-β=80% (expected power), a required sample size of 369 women was calculated. Approximately 20% of eligible women submitted incomplete questionnaires; therefore, the sample population included 444 eligible women.

Data were collected by a trained research midwife. Inclusion criteria stated that all the women were married, aged 15–55 years old and willing to participate in the present study. All women were asked to recall confounding factors associated with sexual function, such as debate and controversy over the last months, stress events in the past three months, physical and psychosocial disorders, and consumption of alcohol or nicotine. Repeat clients were excluded (Figure 1). The study was approved by the Institutional Committee of Ethical Practice of our institution (no. 909015). The aim of the study was outlined for all participants. All participants completed the informed consent form. To enhance confidentiality, all questionnaires were completed anonymously and only required information was collected.

Study data were collected demographically and according to the Female Sexual unction Index (FSFI) scale by a trained research midwife. The demographic questionnaire contained variables such as age, occupation, and education of the husband and wife, monthly income, age at marriage, menarche age, gravity, and parity. However, the FSFI questionnaire included 19 items that were divided into six domains: desire, arousal, lubrication, orgasm, satisfaction, and pain. Each domain contained 2–4 questions that were scored based on the Likert scale. The validity and reliability of the FSFI questionnaire has been confirmed in various countries^17,18,19) and in Iranian women.^20,21)

The total score for each woman was calculated based on the score obtained from the 19-item questionnaire, ranging from 2 to 36. A cutoff point of 28 was determined. Therefore, all women who achieved a total score of >28 were considered to have normal female sexual function, whereas all women who achieved a total score of <28 were considered to be suffering from female sexual dysfunction.

Data were presented as mean±standard deviation (SD) and percentages. Predicted probability for female sexual dysfunction was computed using a multivariate logistic regression model. Area under receiver operating characteristic curve was applied to compare the accuracy of the model. All statistical analyses were performed using SPSS ver. 16.0 software (SPSS Inc., Chicago, IL, USA).

A total of 444 women participated in the present study, excluding women with physical and psychosocial problems, medical illnesses and alcohol addiction, and those who consumed drugs that affect sexual function. Overall, 107 (24.1%) of all participants reported normal sexual function, whereas 337 (75.9%) suffered from female sexual dysfunction.

The mean±SD ages of subjects in the normal sexual function and sexual dysfunction groups were 32.93±8.66 and 28.85±6.42 years, respectively. A significant difference in the mean age of subjects was detected between the normal sexual function and female sexual dysfunction groups (r=-0.241, P<0.001).

Univariate logistic regression analysis indicated a significant association between female sexual dysfunction and age, menarche age (odds ratio [OR], 1.34; 95% confidence interval, 1.14 to 1.57), gravidity (OR, 4.38; multigravid vs. nulligravida), parity (OR, 3.14; multiparous vs. nulliparous) and education level (OR, 3.15; illiterate vs. college). No significant association was detected between female sexual dysfunction and the husband's age and family monthly income. Associations between female sexual function and the other variables assessed by univariate logistic regression analysis are presented in Table 1.

Multivariate logistic regression analysis demonstrated that menarche age (OR, 1.26), education level (OR, 1.71; illiterate vs. college), and gravidity (OR, 1.59; multigravid vs. nulligravida) were considered as independent predictive variables for female sexual dysfunction (Table 2). A multivariate logistic regression model for the probability of female sexual dysfunction and the other covariates was estimated as Ln (p/1–p)=-4.733+0.243 menarche age+0.537 education level+0.463 gravidity.

The coefficients of the logistic regression model indicated that education level (OR, 1.71; illiterate vs. college) was the variable that was most suitable for the prediction of female sexual dysfunction. To calculate the overall percentage of correct classifications of the model, the predicted values were compared to the observed values. The overall percentage of correct classification of the model was 77.9%. Therefore, when data on the menarche age, gravidity, and education level of the woman are obtained, the ability of the model to predict the actual sexual dysfunction category of the cases is 77.9%. Area under the receiver operating characteristic curve criterion was used to calculate the sensitivity and specificity of the model (Figure 2). The area under the curve represents the probability that the model predicted for a randomly chosen positive case will exceed the result for a randomly chosen negative case. Asymptotic significance was <0.05, which indicated that the model is a good method for the prediction of female sexual dysfunction.