Development and Psychometric Evaluation of a Reproductive Health need Assessment Scale for violated Women

Violence as a serious health problem and one of the main manifestations of gender inequality brings about adverse health effects for women. Therefore, it is of utmost importance to recognize the reproductive health status of women subjected to violence in order to provide the health services they need. Considering that one of the ways to determine reproductive health status is the use of valid questionnaires in this eld, the present study was conducted based on a mixed-method design. The rst part of the study (qualitative section) was conducted based on conventional content analysis. In this part, unstructured in-depth interviews were conducted with 18 violated women and 9 experts. In the next stage, the item pool was formed and the Reproductive Health Needs of Violated Women Scale (RHNVWS) was designed based on the review of the literature and the results of the qualitative section with 39 items using the Waltz approach. For psychometric assessment of the above instrument, face and content validity, item analysis, and construct validity were examined using exploratory factor analysis. Based on the results of factor analysis, the four following factors were extracted with a total variance of 47.62: "men's participation", "self-care", "support and health services", as well as "sexual and marital relationships. The internal consistency of the instrument was calculated at α = 0.70–0.89 and α = 0.94 for different constructs and the whole instrument, respectively. Moreover, intra-cluster correlation coecients were obtained at ICC = 0.96–0.99 and ICC = 0.98 for constructs and the whole instruments, respectively. Based on the results of the current study, the RHNVWS is a tool that specically assesses the reproductive health needs of violated women and has appropriate validity and reliability. The results of the assessment using the aforementioned instruments can be of great help in promoting the reproductive health of women subjected to violence.


Introduction
Various studies emphasize the high prevalence of intimate partner violence (IPV) and its adverse health effects [1,2]. In the United States, on average, one in three women experience physical abuse [2], and according to statistics, about 38% of female homicides are committed by intimate partners [3]. The The results of a study performed by Ludermir et al. (2010) on 1,045 women in northern Brazil pointed out that that the most common type of intimate partner violence was psychological violence (25.4%).
Moreover, it was found that psychological violence during pregnancy was positively and signi cantly associated with postpartum depression [18]. In the same context, Sudha et al. (2011) assessed 9,636 Indian married women and denoted that married women's experience of emotional, sexual, and physical intimate partner violence in the past 12 months was signi cantly correlated with the symptoms of sexually transmitted infections. Moreover, the experience of physical violence and acceptance of spousal physical violence was signi cantly associated with a lower odds ratio for care-seeking [11]. The identi cation of the relationship between intimate partner violence and reproductive health is of paramount importance for health promotion. In this regard, one of the primary healthcare services should be the empowerment of violated women to seek reproductive health care [2,11]. The impact of violence on health is widely recognized and the results of various studies pointed to the negative effect of violence on women's general health and reproductive health. Nonetheless, a few measures have been taken to shed light on the effect of violence on women's ability to achieve their health needs and health promotion [19,20].
It is essential to understand the reproductive health status of violated women and the health needs of these women to provide the required services, and one of the ways to promote this awareness is the use of valid questionnaires. Therefore, this mixed-method study aimed to determine the components of reproductive health in women subjected to violence and design a valid and reliable assessment tool.

Materials And Methods
The present study is the second stage of a mixed-method study.
First phase: In the rst part, a qualitative study was conducted based on conventional content analysis to explain the reproductive health needs of women with experiences of domestic violence. Participants in this section included 18 women within the age range of 20-49 years living in Tehran who were recognized as violated according to the Domestic Violence Questionnaire developed by the World Health Organization.
The inclusion criteria entailed: 1) absence of any physical or mental illnesses, 2) no history of substance abuse, 3) absence of AIDS and hepatitis. Moreover, nine experts in reproductive health, social health, and violence were also interviewed. The participants were selected by purposive sampling method which lasted for 9 months. The data were collected through unstructured individual interviews, observations, and eld notes. To increase the validity of the data, the researcher maintained a long presence with the participants and devoted su cient time to collect the data.
In addition, the presence of the research team and the close cooperation of three experts in the analysis and interpretation of data played an important role in improved validity of the information. Checking was performed in two ways, namely member check and peer check, by other members of the research team. In addition, we used a combination of data collection methods, such as in-depth individual interviews, observation, and led notes [21]. Apart from the research team, an external observer and code-recode method were used during data analysis to increase reliability or stability. Simultaneously with the interview, the qualitative content analysis was performed based on the Graneheim and Lundman method [22]using Max Kyoda software 2010.

Second phase
The inductive-deductive method was used to design the expressions and items of the tool. The items were formed based on the codes extracted from the rst phase of the study.
It was simultaneously attempted to nd another set of phrases related to this concept through an extensive review of related texts and tools, and nally, the item pool was formed. The instrument was reexamined by the research team, and similar items were removed. Face and content validity, item analysis, and construct validity were performed using exploratory factor analysis for psychometric assessment of the above instrument.

Face validity
In the present study, in order to evaluate the qualitative face validity, face-to-face interviews were conducted with 10 members of the target group, observing maximum diversity in terms of education, age, and violence severity. The di culty, relevance, and ambiguity of the items were assessed, and the required modi cations were made. In the next step, to eliminate or reduce some disproportionate items and determine the degree of importance of each phrase, the impact score of the item for each of the phrases was calculated based on a ve-point Likert scale with the participation of 10 participants. Finally, the items with an impact score of ≥ 1.5 were deemed appropriate, and the rest of the items were removed [23].

Content validity
Both qualitative and quantitative methods were used to determine content validity. For the assessment of the qualitative content validity, 10 experts and faculty members were asked to provide their corrective feedbacks on the following criteria: clarity and simplicity, grammar, wording, item allocation, and scaling. Content validity ratio (CVR) and content validity index (CVI) were calculated for quantitative content analysis.
To determine the content validity ratio, the theoretical and practical de nitions of each construct was provided. Moreover, 14 experts were asked to give their opinions about each of the items based on the three-point scale of "Essential", "Useful but not essential", and "Not necessary". In the current study, since 14 experts examined the primary tool, the minimum acceptable value of the CVR in the Lawshe table was regarded as 0.51 [24]. In order to calculate the CVI, 15 experts were asked to examine each item based on a four-point Likert scale; thereafter, the CVI was calculated. Items with a score greater than 0.79 were deemed appropriate and retained, items with a score within 0.7-0.79 were corrected and revised, and items with a score less than 0.7 were removed [25].

Content validity index and compliance with the chance agreement
The modi ed kappa formula was also used to eliminate the probability of chance agreement. Kappa coe cients between 0.4-0.59, 0.6-0.74, and > 0.74 were considered weak, good, and excellent, respectively [26].

Item analysis
Two methods of determining discrimination index and the loop method were used for item analysis. In the evaluation of the discrimination index, if the correlation coe cient between the item and the whole questionnaire was less than 0.3, the item was removed. In addition, if the correlation coe cient between the two items was more than 0.7, one of those items was removed. In the loop method, the reliability coe cient of all items was initially calculated. If deleting an item reduces reliability, it signi es that this item is well-coordinated with others; therefore, it is a good item [26]. At this stage, before construct validity, the questionnaire was provided to 50 women subjected to domestic violence who met the inclusion criteria, and the above coe cients were calculated for each item.

Construct validity
The construct validity was investigated by exploratory factor analysis (principal axis factoring and maximum likelihood factor analysis) using Varimax rotation. To select appropriate variables for factor analysis, correlation matrix, Kaiser-Meyer-olkin Measure of Sampling Adequacy (KMO) tests and Bartlett's test of sphericity were used. Eigenvalue and Scree plot were also used to determine the optimal number of factors.

Internal consistency
To determine internal consistency, Cronbach's alpha coe cient of items was determined. In addition, the total score and the score of each factor were separately calculated. Items with an alpha of ≥ 0.7 remained in the instrument [27].

Stability
In the current study, the test-retest method was used to evaluate stability. In this method, the tool was provided to 50 violated women who met the inclusion criteria in two stages with a two-week interval.
Thereafter, inter-class correlation analysis was performed between the scores of the two tests. The coe cients of < 0.5 /,0.5-0.75, and > 0.75 were considered weak, moderate, and good, respectively [28].

Sample
Inclusion criteria were the same in qualitative and quantitative part of the study. The research setting for data collection in psychometric assessment of the instrument was healthcare centers and comprehensive health centers in Tehran which were randomly selected. The sample size was determined based on the number of questionnaire items. As a general rule, the number of samples should be about 4 or 5 times the number of variables [29]. In the current study, the number of samples for exploratory factor analysis was obtained at 6 per item (i.e., 342 samples). Finally, 350 self-reporting questionnaires were provided to participants who were selected by convenience sampling. The obtained data were analyzed in SPSS software (version 22).

Results
The themes, classes, and subclasses extracted from the qualitative phase of the study are displayed in Table 1. In the next step, 116 items were extracted from qualitative interviews and also according to the review of related studies based on the concept of reproductive health needs of abused women, and the item pool was formed. After three revision sessions held by the research team, some items were merged or deleted, and some others were reviewed and edited. Finally, the number of items was reduced from 116 to 99 items, and a psychometric assessment was performed.
Face validity: Three items were revised in the assessment of face validity in the qualitative stage. The lowest impact factor was obtained at 1.8, and the items received a good score at this stage.
Content validity: At this stage, some items were removed, and some items were modi ed. For instance, the phrase "I care about suspicious sexual relationships of my husband" was modi ed to "I care about suspicious sexual relationships of my spouse due to the transmission of sexually transmitted diseases, AIDS and hepatitis." Finally, 83 items remained after merging and removing duplicate items. In the assessment of content validity, the numerical value of CVR was determined based on the Lawshe Table   for Minimum Values and the number of evaluators. If this value was less than 0.51, the items were removed, and the primary instrument with 72 items entered the CVI and kappa coe cient stage. Items with a CVI of ≥ 79 were retained, those which scores within 0.7-0.78 were corrected and revised, and the remaining items were removed. Moreover, the items with a kappa coe cient of < 0.74 were removed, and nally, 66 items remained. The CVI of the whole scale for the present tool was calculated at 0.905, which con rms this important index.
Item analysis: At this stage, the items whose correlation coe cient with the total score was less than 0.3 were deleted or edited. Moreover, in cases where the correlation coe cient between two items was > 0.7, they were merged and the number of items decreased from 66 to 57.
Construct validity: The demographic characteristics of the participant in the construct validity assessment are depicted in Table 2. The KMO index in the present study was 0.923, and the result of Bartlett's test of sphericity was signi cant (P < 0.000). To investigate the number of factors that made up the tool, eigenvalue and scree plot were used. The analysis was performed by considering the eigenvalue of > 1, and nally, four factors were account for 47.62% of variance as follows: men's participation: 20.51%, self-care: 13.24%, support and health services: 8.83%, sexual and marital relationships: 5.02% (Table 3).
Reliability: To calculate the internal consistency and relative stability of each factor, Cronbach's alpha coe cient and the intraclass correlation coe cient (ICC) calculated for 350 samples. Cronbach's alpha coe cient and ICC for the whole instrument were obtained at 0.94 and 0.98 respectively (Table 4).

Discussion
The results of the qualitative section indicated that the important health problems posed to violated women included lack of sensitivity to health, increased high-risk behaviors, ineffective life skills, insu cient family and social support, spouse's disregard for the physical, sexual, and mental health of women, as well as a health system and reproductive health unresponsive to the needs of violated women.
These problems result in special reproductive health needs at individual, family, and community levels. It is essential to identify the health needs of violated women in order to promote their general health and fertility. Therefore, in the second part of this study, based on the data obtained from the qualitative section, the Reproductive Health Needs of Domestic Violated Women scale (RHNVWS) was designed.
To design this questionnaire, all stages of psychometrics assessment, including face validity, content validity, item analysis, construct validity, and reliability were thoroughly performed. Based on the results of exploratory factor analysis and using Varimax rotation, four latent factors were extracted that explained 47.62% of the total variance of the reproductive health needs of women subjected to domestic violence. These factors included: men's participation, self-care, support and health services, and sexual and marital relationships, which yielded a 39-item questionnaire.
The items were scored on a ve-point Likert scale ranging from always (5) to never (1), and in some items, from very much (5) to very low (1). Cronbach's alpha coe cient for the whole instrument was calculated at 0.94. As evidenced by the results of the present study, the 12-item construct of men's participation had the highest variance (20.51) among the constructs of this tool. In this construct, the factor loading of items ranged from 0.41-0.73; therefore, it can be concluded that men play a key role in the promotion of women's reproductive health [30].
The results of a study carried out by Akhavan and Simbar (2016) demonstrated that men's involvement in reproductive health brings about the following consequences: raising awareness and information transfer to their wives, improving the coverage of family planning services, investing in prevention and control of sexually transmitted infections, increasing men's involvement in pregnancy issues, including early initiation of prenatal care and adoption of appropriate health behaviors [31]. These results largely overlap with the items of male participation in the present questionnaire and highlight the importance of assessing the needs of this area of reproductive health.
Self-care is a key strategy for health promotion and disease prevention [32]. In the RHNVWS, the 13-item construct of self-care accounted for 13.24% of the variance. In this construct, the factor loading of items varies from 0.43-0.64 and examines important aspects of self-care in the reproductive health of violated women. Another construct in the present instrument was support and health services. This 7-item construct accounts for 8.83% of the total variance. The factor loading of the items varies from 0.45-0.65, and the items include the support and services that violated women receive from their families, communities, and health centers.
The results of a study conducted by Khani et al. (2017) pointed out that the health systems that are unresponsive to sanitary needs and the non-response of the behavioral counseling centers to violence against women are among the important sexual and reproductive health needs of women [33]. The need to have access to a responsive health system in providing appropriate services and counseling is assessed in different items of the construct of support and health services. According to the World Health Organization, satisfying and safe sex life is one of the aspects of access to sexual rights [34]. The fourth 7-item construct of the present questionnaire is related to sexual and marital relationships which accounts for 5.02% of the variance of the instrument and assesses the various dimensions of sexual and marital relationships. In this construct, the factor loading of the items varies from 0.72 − 0.40.
The Centers for Disease Control and Prevention (CDC) has developed a tool to assess the reproductive health needs of con ict-affected women based on WHO and CDC studies. This instrument encompasses 10 sections, namely demographic characteristics, safe motherhood, family planning, marriage and marital life, sexual history, sexually transmitted infections, knowledge, beliefs and attitudes about HIV and AIDS, gender-based violence, female genital mutilation, and mental health [35].
Another questionnaire similar to the one used in the current study is the Reproductive Health Needs Assessment Questionnaire, developed in consultation with the International Organization for Migration and the United Nations Population Fund. The mentioned questionnaire consists of 114 items in the following sections: demographic information, safe motherhood, family planning, history and practice of sexual activity, sexually transmitted infections, HIV and AIDS, and gender-based violence [36].
The construct of self-care in the present study included items related to family planning, safe motherhood, sexually transmitted infections, HIV and AIDS which were assessed in separate sections in the two mentioned questionnaires. In addition, the sections on marriage, marital life, and sexual history in the two questionnaires were in line with the construct of sexual and marital relationships in the present questionnaire. The two aforementioned instruments have been prepared based on a review of various studies, and their psychometric assessment was limited to content validity. A large number of items and the method of recording the answers in the mentioned questionnaires made them time-consuming for both the respondent and researcher.
The instrument in the present study was exclusively designed for con ict-affected women, and its validity has been con rmed by examining the face validity, content, and constructs validity. Moreover, reliability was validated by Cronbach's alpha method. In addition, apart from using transcriptions and opinions of key participants, the items were extracted from interviews with violated women. The present researcher designed this questionnaire to represent reality from the participants' perspectives since the subjects of the present study are a certain group of women who have different experiences.
In the present study, it was attempted to design a valid questionnaire with minimum possible items. Therefore, these scales assess the reproductive health needs of women subjected to domestic violence in four important areas; moreover, it can be applied in a short time due to ease of answering using the Likert scale. One of the strengths of the current study is that the needs assessment was based on interviews with two target groups, including women subjected to intimate partner violence and key informants. The consideration of this critical issue in needs assessment increases the reliability of the results.
The results of the current study were con rmed in terms of face validity, content validity, construct validity, and reliability. Therefore, this instrument can be used as a valid and reliable tool for the assessment of the reproductive health needs of women with experience of domestic violence.

Conclusion
Public health improvement depends on the promotion of women's health. The violated women as a vulnerable group experience more serious health problems and have different health needs, compared to other women. The modi cation of reproductive health services of violated women and raising the awareness of women and men on reproductive health programs require the assessment of target group needs. In the present study, the use of experts and key informants' opinions and the experiences of violated women through in-depth and qualitative research resulted in the development of the RHNVWS questionnaire with such important features as appropriate reliability and validity, simplicity, and timeeffectiveness. Therefore, the present questionnaire, as a practical tool to assess the reproductive health needs of women with experience of domestic violence, can be of great help for health planning to provide the required services to the target group.