Internal Structure and Consistency of the Spanish Version of the Six-Item Female Sexual Function Index

Objective : The purpose of this paper is to evaluate the structure of the Spanish version of the FSFI-6 in climacteric Colombian women. Methods : The validation study involved 1,427 sexually active women aged between 40 and 59 years. Cronbach’s alpha, McDonald’s omega and confirmatory factor analysis were calculated. Results : The Cronbach’s alpha was 0.876 and the McDonald’s omega was 0.886. The one- dimension model did not fit, chi-square=582.04, df=9, p<0.001, RMSEA=0.211 (90%CI 0.197-0.226), CFI=0.888 and TLI=0.813. Conclusion : The FSFI-6 presents a one-dimensional structure with high internal consistency in climacteric women from the Caribbean coast of Colombia. But, the data’s goodness of fit is inadequate. These findings need to be verified in other populations.


Introduction
Sexual dysfunctions are a mixed group of clinical syndromes, typically characterized by a clinically important impairment in a person's ability to experience sexual pleasure or respond sexually. Female sexual dysfunction, as defined by the latest version of the Diagnostic and Statistical Manual of Mental Disorder (DSM-5), has been separated into several categories, including orgasmic disorder, sexual interest/arousal disorder, Genito-pelvic pain/penetration disorder, substance/medication induced sexual dysfunction, other specified sexual dysfunction, and unspecified sexual dysfunction [1]. Female sexual dysfunction is highly prevalent, but, according to the background and measurement instrument, this prevalence is variable around the world [2,3]. Several factors including physiological, anatomical, and socio-cultural aspects are related to female sexual dysfunction [4,5], the most prevalent of which is hypoactive desire disorder [2]. In Latin America, female sexual dysfunction is close to 20% in Colombia and reaches 98% in Ecuador [6][7][8].
The psychometric performance of scales varies according to the population's characteristics. However, a similar functioning in different population groups is an important empirical approximation to the validity and reliability of a measurement [9]. Factor analysis is frequently used to test the dimensionality of health measurement instruments and thereby to indirectly demonstrate the construct validity [10]. Rosen et al. [11] introduced the Female Sexual Function Index  to evaluate the sexual response over the previous four weeks. Subsequently, a short version (FSFI-6) was designed by Isidori et al. [12], who observed acceptable internal consistency (Cronbach's alpha=0.79) and stability (Pearson correlation=0.95) in a sample of 184 women aged between 21 and 41. However, the dimensionality of the FSFI-6 was omitted. Subsequently, other investigations replicated the high values of internal consistency and the convergent, divergent, discriminant and nomological validity of the instrument [13][14][15][16].
Finally, using a sample of 307 women from the general population and another of 68 from the clinical practice, aged between and 21 and 66, the one-dimensional structure of the Portuguese version of the FSFI-6 was tested using confirmatory factor analyses. Several translations of both FSFI-6 and FSFI-19 are now available, not all of them have been adequately analyzed and there is little information regarding the ability of those versions to reproduce the dimensionality and then the construct [17,18]. Although, a Spanish version of the FSFI-6 has been applied to assess sexual problems, its dimensionality and construct validity are still unknown [13,14]. The purpose of this study was to evaluate the dimensionality and internal consistency of a Spanish version of the FSFI-6 in a large sample of Colombian Caribbean women, to deepen the psychometric performance of the index given the need to corroborate the validity and reliability of a health measurement instrument in different contexts. In addition, the McDonald omega was calculated as a complementary measure of internal consistency reliability.

Methods
A psychometric or validation study was designed, and approval was obtained from the research ethics board. Participants gave their informed consent. The information was collected after a pilot test applied to a group of twenty women. Questionnaires from this group of women are not included in the present analysis. The scales were applied at home. A trained health professional applied the FSFI-6.

Instrument
Women completed the six items of the Spanish version of the FSFI-6. They were asked about their sexual performance over the past 4 weeks. Each item provides six response options that are rated from zero to five, where zero represents the poorest function and five, optimal function. Total scores range from zero to thirty [12]. The items are: [A] How would you rate your level ("degree") of sexual desire or interest? [B] How would you rate your level of sexual arousal ("turn on") during sexual activity or intercourse?
[C] How often did you become lubricated ("wet") during sexual activity or intercourse? [D] When you had sexual stimulation or intercourse, how often did you reach orgasm? [E] How satisfied have you been with your overall sex life? [F] How often did you experience discomfort or pain during vaginal penetration?

Statistical analysis
The internal or dimensional structure of the FSFI-6 was explored through confirmatory factorial analysis (CFA), using the maximum likelihood method. To find out whether the group of items had a latent factor, Bartlett´s test of sphericity [19], and Kaiser Meyer Olkin measure of sampling adequacy (KMO) [20] were both calculated. These indicators suggest that the factor analysis should be followed, but there is no guarantee of finding a satisfactory dimensional structure. Several goodness of fit indicators was calculated in the CFA: chi square test with degrees of freedom (df) and probability value (p), and the RMSEA coefficient (Root Mean Square Error of Approximation Of the approximation error) with a 90% confidence interval (CI90%), the Comparative Fit Index (CFI), the Tucker-Lewis Index (TLI) and SRMS (Standardized Mean Square Residual). These indicators are acceptable if the chi square shows the probability value as being greater than 5%; for RMSEA, less than 0.06; and CFI and TLI, values greater than 0.89. The internal consistency of the index was estimated using Cronbach's alpha [21] and McDonald's omega [22]. The analysis was performed using STATA [23].

Result
Factor analysis showed a Bartlett´s test chi-square of 5,095.15, df=15 and value p<0.001 and KMO of 0.841. Following this, one dimension was retained giving an Eigen value of 3.8, which explained 63.3% of the total variance. In the CFA, the model did not fit perfectly, and had a chi square=582.04, df=9, p<0.001, RMSEA=0.211 (90%CI 0.197-0.226), CFI=0.888 and TLI=0.813. Commonalities and loadings are presented in the . Several one-or two-dimensional models and different item numbers were tested without adequate goodness of fit. The FSSI-6 showed high internal consistency with both measurements, Cronbach's alpha was 0.876 (95%CI 0.867-0.885) and McDonald omega was 0.886.

Discussion
In this study, it was found that a FSFI-6 showed poor construct validity, despite the high internal consistency among climacteric women from three Caribbean cities in Colombia. In this study, poor construct validity was observed for the FSFI-6. This finding differs from that of Santos-Pechorro et al. [24], who using a Portuguese version, found good indicators for construct validity in several coefficients using CFA. However, they reported unsatisfactory RMSEA and omitted chi square and the p value for the chi-square. The construct validity of the scales should be repeated in different population groups [10,25].
In this study, the internal consistency for FSFI-6 was high using two different measures: Cronbach's alpha and McDonald's omega. Previous papers have also reported high internal consistency; for example, Isidori et al. [12] obtained a Cronbach's alpha of 0.79, both Chedraui et al. [13] and Perez et al. [14] obtained 0.91, and Lee et al. [15] found 0.89. However, it is evident that the internal consistency can change by population, meaning that every time the scale is applied to a sample, this must be reported [9]. The FSFI-6 is a recently introduced instrument with few items and easy scoring and interpretation [12]. The performance of short scales is as good as that of extensive instruments that purported a comprehensive dysfunction" derive from the bases of "normality" for female sexual function as they evaluate sexual function comparing it to male sexual response. Also, the construct does not consider that changes in female sexual response through life, do not necessarily represent a sexual dysfunction [2,27]. The present study is a contribution to the knowledge of the psychometric performance of the FSFSI in Colombian Caribbean climacteric women, especially considering the calculation of CFA [28], and McDonald omega as a measure of internal consistency, which has been omitted in previous studies [22,29,30]. Nevertheless, the study presents the limitation inherent to this type of research, which does not allow generalizations as psychometric performance changes according to populational characteristics [9].

Conclusion
To conclude, the FSFI-6 presents a one-dimensional structure with high internal consistency in climacteric women from Colombian Caribbean coast. However, the data's goodness of fit is inadequate. New research is needed to verify the construct validity of the FSFI-6 in other populations.

Conflicts of Interest
The authors have no potential conflict of interest to disclose