Validation of women abuse screening tool (WAST-S) in the Mozambique National Health Service (significance for HIV prevention and ART treatment Adherence): A quasi-experimental crossover study

Intimate partner violence (IPV), includes a wide range of abusive behaviors perpetrated by someone involved or who was involved in an intimate relationship with the victim. It is a serious and preventable public health problem globally. This article reports on the validation of an adaptation of the Women Abuse Screening Tool (WAST-S) in Mozambique and recommends its application for more effective prevention and ART treatment. The study utilized a quantitative approach in a prospective quasi-experimental crossover design, in which clusters, 4 health care units, were allocated to two intervention approaches in reverse order. IPV was measured using a translated, culturally-adapted version of WAST-S. This was compared to a standard clinical interview that included questions on IPV. There was a good agreement between the scales of WAST-Short and the standard clinical interview: 0.235 (95% CI: 0.219 - 0.250). The WAST-Short was found to be more effective than the standard clinical interview as an IPV screening tool. It provides information about intimate relationships and can measure the presence of violence. The standard clinical interview is less effective at obtaining this information. WAST-S proved to be a more reliable instrument than a clinically adapted interview, and can be applied in the screening of IPV in high patient volume settings and efficiently implemented in primary health care units which can then refer patients for specialized care. The two questions of the WAST-S provide insight into the intimate relationship measuring the presence of violence, information not easily accessible by the standard clinical interview. This information when provided to clinicians has the potential for assuring more effective prevention and treatment adherence.


INTRODUCTION
The World Health Organization classifies intimate partner violence (IPV) as a subset of gender-based violence, in partner of the victim (Righi et al., 2019).IPV is a serious and preventable public health problem worldwide (Tjaden et al., 2000;Yount et al., 2022).While the global prevalence of IPV is estimated at about 30%, it is particularly prevalent in sub-Saharan Africa, where the overall prevalence is 36%.Compared to the rest of the world, women in sub-Saharan Africa are more affected by domestic violence than men (Jethá et al., 2021).In some southern African countries, levels of sexual abuse of women are even higher (Ahinkorah et al., 2018).For instance, in Mozambique, 40% of women of reproductive age (15-49 years) reported experiencing physical, sexual, or emotional violence perpetrated by an intimate male partner in the last 12 months (Tura and Licoze, 2019).
Generally, IPV disproportionately affects more women than men.Worldwide, one in three women has been physically and/or sexually abused by her intimate partner or family members at some point in her lifetime, compromising their physical, mental, and reproductive health (Jethá et al., 2021).Violence in intimate partner relationships is usually attributed to gender norms that promote male dominance over women and women's acquiescence to male power (Shai et al., 2019).Indeed, gender theories suggest that IPV results from the way women are socialized to accept being sexually passive, dependent on men for protection and economic survival, and even being disciplined by their partners when they fail to behave according to their traditional roles (Shamu et al., 2018).Abuse by male partners results from the violation of such norms (Tura and Licoze, 2019), and the context in which IPV takes place can be of significant importance (Shai et al., 2019).
IPV affects one-third of all women, and the data from the majority of studies on screening and interventions for IPV are based on interview-based reports, such as those by Tura and Licoze (2019), which rely on self-reported violence.Their main limitation is that they do not involve specific instruments for IPV screening and diagnosis.Another limitation is that cases of IPV may be underreported and neglected (Jethá et al., 2021) due to concealment, subtle or absent overt signs of violence, or failure to recognize that a relationship is abusive.Nevertheless, the trauma inflicted on the victims, ranging from gastrointestinal disturbances, chronic pain, and cardiovascular problems to spontaneous abortions, suicides, and depression (Stephenson et al., 2013;Tura and Licoze, 2019), suggests the need for proper screening of all patients (universal screening), including asymptomatic individuals whose suffering may have been triggered by IPV.Clinical and laboratory screening tests are especially important in the absence of signs of an underlying disease (Eluf-Neto and Wünsch-Filho, 2000).Healthcare professionals should routinely ask all women about domestic violence, and the screening might encourage women who would not otherwise do so to disclose abuse or recognize their own experiences as abuse (O'Doherty et al., 2015).However, there is controversy regarding the clinical merits of universal screening for violence in the healthcare setting (Sprague et al., 2016).

IPV and HIV related health consequences
Mental distress is hypothesized to be a pathway connecting IPV to HIV-related consequences (Hatcher et al., 2022).Prevalent mental health issues such as depression, anxiety, and post-traumatic stress syndrome are strongly associated with women's involvement in IPV.These outcomes have been reported in several longitudinal studies on IPV and depression (Buller et al., 2014), anxiety (Lagdon et al., 2014), and PTSD (Dillon et al., 2013).Consequently, depression is related to longterm declines in ART adherence.Furthermore, there is likely an interactive relationship that exacerbates each situation (Wagner et al., 2020).Post-traumatic stress and anxiety are significantly correlated with poorer ART adherence in cross-sectional studies (Glynn et al., 2021).Other studies in South Africa focused on perinatal women with HIV found that those highly affected by the violence in their relationship with their partner made intentional efforts to self-harm by interrupting their treatment (Hatcher et al., 2022).
The primary goal of IPV screening is to improve case detection and implement appropriate interventions that reduce exposure to violence and related problems.Ideally, the screening tool should encompass different aspects of violence, including physical, psychological, and sexual dimensions, using psychometric scales (Krug et al., 2002).However, many countries have inadequate IPV screening protocols due to systemic barriers, including time constraints (Erickson et al., 2001;Sprague et al., 2012), lack of protocols, policies, and screening procedures (Waalen et al., 2000).Additional barriers include health professionals' perceptions of IPV (Matavel, 2020), personal discomfort with the subject, inadequate resources, training, or privacy for screening, the view that screening for IPV is not the clinician's role, fear of offending patients, and a lack of knowledge on how to deal with women who qualify as victims of IPV (Sprague et al., 2012).A systematic review documented that only 9 to 40% of physicians screen for IPV (Todahl and Walters, 2011).
In Mozambique, despite high rates of self-reported violence, the magnitude of IPV is likely to be underestimated, mainly because of the lack of diagnostic screening tools within the National Health Service.Given this, the aim of this work was to validate the short version of Woman Abuse Screening Tool (WAST-Short) adapted for the cultural context in Mozambique.The study used WAST since a systematic review of 33 studies in which multiple screening tools were compared found WAST to be the most reliable and valid instrument for violence screening, with a sensitivity and specificity of 47 and 96%, respectively (Rabin et al., 2009).By evaluating the reliability or consistency of the adapted tool, we sought to demonstrate an equivalence of concepts and semantics between the original and back-translated versions and the pilot testing of the of the adapted WAST in Mozambique showed good internal consistency (Matavel et al., 2023).

Study design
The study was conducted in four primary health care units in two southern Mozambique provinces, Maputo and Gaza.A prospective quasi-experimental crossover designed study design was utilised in which clusters or health care units were allocated to two intervention approaches at different times.The work used a quantitative approach.The intervention consisted of IPV screening using the WAST in individuals of both sexes, female (n = 220) and male (n = 140).Although the WAST is an instrument originally designed for IPV screening in women, WAST was here applied to both genders, as pre-test results indicated that the two scales of the instrument do not discriminate between genders (WAST-Short p = 0.204, WAST-Long p = 0.271), and are not sensitive to the individual items that make up the scales (p> 0.05) (Matavel et al., 2023).
IPV screening was also carried out on the same individuals using an adapted clinical interview to which specific questions about IPV were added in order to validate the WAST-Short.Both WAST and the adapted clinical interview were implemented twice over a period of 4 months in the four primary health care units (HC I, II, III and IV).The units were randomly selected from other primary health care settings in Maputo (HC I and IV) and Gaza (HC II and III) Provinces.Study subjects in the units included all users of the Emergency, Gynecology/Obstetrics or Maternity services, Mother and Child Health, Stomatology, Psychiatry and Mental Health, and Assistance Service for Victims of Violence.
To meet the inclusion criteria, subjects had to be aged 18 years or over and to be or have been in an intimate relationship in the last 6 months.All subjects had to voluntarily agree to participate in the study and sign a form of informed consent.Patients below 18 years of age and those who have not been in an intimate relationship in the last 6 months prior to the intervention were excluded.

Data collection and analysis
Health professionals at each health care unit were initially invited to participate in a training/familiarization session on the application of the two interventions -WAST and the adapted clinical interview.The health care units were randomly assigned to implement one of the two interventions: AB and BA, two for each approach for the first 2 months, before switching over (crossover) to the other approach during the next 2 months.According to the order of arrival, participants (patients) of both sexes were then screened for IPV using the following approaches: 1. Approach AB: Patients responding first to the WAST and then to the adapted clinical interview.This approach was first implemented in health care units I (Maputo) and III (Gaza).2. Approach BA: Patients responding first to the adapted clinical interview and later to WAST.This approach was first implemented in health care units II (Gaza) and IV (Maputo).Data from the participants were collected using a simple randomization process at the healthcare unit outpatient and mental health consultation rooms where the instruments were applied.
Data collection involved the use of paper-based forms for the adapted instruments, which were later converted into an electronic dataset.To be eligible for participation, subjects had to be aged 18 years or older, currently or previously in an intimate relationship within the last 6 months, willing to participate voluntarily, and had to provide informed consent by signing a consent form.

WAST (woman abuse screening tool)
This instrument was developed in the USA to screen women exposed to physical, psychological and sexual abuse (Brown et al., 1996) and subsequently translated and adapted to the Mozambican cultural context (Matavel et al., 2023).WAST is short and easily understood by users and 90% of users reported that were comfortable or very comfortable with the questions (Brown et al., 2000;Salahi et al., 2017).The instrument consists of 8 Likert-type questions about (a) the degree of stress in the relationship; (b) the difficulty in resolving discussions and (c) the existence of violent episodes at an emotional, psychological, physical or sexual level (Binfa et al., 2018).The validity and reliability of the original WAST version were confirmed in different studies, as shown by high internal consistency, Cronbach's alpha: 0.75, sensitivity: 92%, specificity: 100% (Brown et al., 1996).
WAST has three response categories (1 to 3) for each item related to violence or its prediction.The first 2 items, known as WAST-Short, are used to screen for the presence of abuse, assessing the level of tension a woman feels in the relationship with her husband/partner, and the difficulty in resolving the discussions, on a scale from 1 (no tension/no difficulty/no problems) to 3 (a lot of tension/great difficulty/very problematic).The following 6 items were then used to obtain a complete assessment of the frequency of abuse (WAST-Long) in the three areas of IPV (physical, sexual and psychological) in the case of the presence of tension in the relationship according to the answers to questions 1 and 2 on a scale of 1 (never) to 3 (often/ many times) (Salahi et al., 2017).
The total score for WAST ranges from 8 to 24, ranging from 2 to 6 for WAST-Short.The tool developers proposed a score of 13 for positive cases of IPV in the overall 8-item WAST score (Brown et al., 1996).In the WAST-Short dichotomous score, a score of 0 is considered as "no tension/no difficulty/no problem" (Brown et al., 2000).

Adapted clinical interview
This is an assessment procedure that involves a personal exchange between the clinician and the client, designed to gather the information necessary for diagnosis and treatment (Allen and Becker, 2019).For the purposes of this study, specific questions about IPV were added, to be used as a gold standard, such as: (a) fear of the partner; (b) past year history of physical abuse by partner; (c) past year history of psychological or emotional abuse by partner; and (d) history of abuse or threat of sexual abuse in the last year by the partner.These terms were adapted from DVIS -Domestic Violence Initiative Screening (Iskandar et al., 2014), an IPV "Yes/No" tracking tool.The answers to each question are scored separately (Basile et al., 2007).

Statistical analysis
Pilot testing of the adapted WAST in Mozambique showed good internal consistency: WAST-Short scales (α = 0.813) as compared to the WAST-Long instrument (α = 0.834) (Matavel et al., 2023).Data analysis for each step was performed using the statistical  Foundation), and the significance level adopted for all tests and confidence intervals was 0.05.A cumulative scale was used to create two scales based on the WAST questionnaires: WAST-Short (sum of questions 1 -2) and adapted clinical interview (sum of questions 1 -4).The values of the WAST-Short scale and the adapted clinical interview were dichotomized as abusive IPV (if the sum ≤ 4) and non-abusive (if the sum > 4).
Positive Percent Agreement (PPA) and Negative Percent Agreement (NPA) were used to assess performance on the WAST-Short scale to identify abusive IPV, using the adapted clinical interview as a comparison.The overall agreement observed and corrected for the possibility of random agreement was computed using Cohen's Kappa coefficient, with values < 0 indicating lack of agreement; 0 -0.20 as mild; 0.21 -0.40 as reasonable; 0.41 -0.60 as moderate; 0.61 -0.80 as substantial and 0.81 -1 as almost perfect agreement (Landis and Koch, 1977).To investigate whether there was any difference in misclassification rates between approaches AB and BA, a logistic regression was applied, taking into account the clustering effect of health facilities using generalized estimating equations, assuming an exchangeable working correlation matrix (Hardin and Hilbe, 2002).

Informed consent statement
Informed consent was obtained from all subjects involved in the study.

Agreement between WAST-Short and adapted clinical interview
Using approach AB, the misclassification rate was 36.9% (95% CI: 29.4 -44.9%) and the overall agreement measured by Cohen's Kappa was 0.258 (95% CI: 0.224 -0.292), indicative of reasonable agreement between the scales.The proportion of the adapted clinical interview with an abusive IPV score and an abusive IPV score on the WAST-Short or PPA was estimated to be 65.9% (95% CI: 57.7 -74.1%) and the NPA estimated to be 59.9% (95% CI: 50.5 -69.2%), both values indicating a weak agreement between the scales (Table 2).

Effect of approach on misclassification rate
From the tables, misclassification rates differed slightly between the two approaches used.Before any adjustment, the odds of misclassifying the IPV using approach AB or BA were similar (odds ratio, OR = 1.09; 95% CI: 0.78 -1.51) (Table 5).After adjusting for age, sex, and marital status, the chance of misclassification for IPV using approach BA was lower (OR = 0.59; 95% CI: 0.41 -0.84) than that of approach AB and was modified by age.For example, the chance of misclassifying IPV using approach BA increased by 14% for patients aged 26 to 29 years (p-value = 0.003) and was twice as high for those aged 30 and over years (p-value = 0.001) than approach AB.In addition, misclassification of IPV was less likely in women than in men (OR = 0.78, 95% CI: 0.69 -0.86), for a given method, age, and marital status.Misclassification was also less likely for patients older than 30 years compared to those aged 18 -21 years (OR = 0.49, 95% CI: 0.43 -0.57) when using approach AB, but about 3 times higher for divorced compared to single patients (OR = 2.5, 95% CI: 2.36 -2.81).

DISCUSSION
Screening for violence in general and IPV, in particular, is highly recommended, especially when referral services are available for individuals identified as victims.It is, therefore, important to have screening instruments that take into account the specific country's context.While research suggests that current instruments for screening IPV still require refinement (Iskandar et al., 2014), a good approach is to develop or select from existing instruments one that can be adapted, tested, and validated for the social and cultural context of the country where it will be applied (Fogarty and Brown, 2002).The aim of this study was to validate the WAST-Short for use in Mozambique.Analysis of the data revealed fair agreement between the WAST-Short and the adapted clinical interview scales regarding the detection of IPV.Moreover, the reliability of the WAST-Short yielded Cronbach's alpha coefficients similar to those reported by Iskandar et al. (2015): 0.667 for items 1 to 2, with all correlations significant at p < 0.001 (Iskandar et al., 2015).Strikingly, the results of this study suggest that the WAST is a more effective IPV screening tool than the adapted clinical interview.The WAST-Short, based on two questions, provides insight into the intimate relationship, confirming or denying the occurrence of a single act of violence in a relationship, information not easily accessible through the standard clinical interview.
Results from the healthcare units in both of the two Mozambican provinces proved to be highly consistent.However, it is important to be cautious about the generalizability of the findings to other provinces in Mozambique without specific testing, as different sociocultural contexts, including variations in the prevalence of IPV (Feinstein and Cicchetti, 1990), and differences in service availability may exist.
Although the results obtained in the testing indicate that the WAST is an adequate IPV screening tool in the context in which it was tested, the challenge remains to develop consistent and effective preventative actions against violence and victim assistance policies, including intersectoral coordination (Binfa et al., 2018).Reaching the ultimate goal of preventing and reducing IPV will depend strongly on screening tools to increase awareness of the problem and establish referral services for victims to receive specialized victim support.However, the controversy surrounding universal IPV screening cannot be ignored, especially among those who support universal screening in healthcare settings, claiming that screening increases IPV detection rates, and those who do not support universal screening, claiming that there is insufficient evidence to support implementation (Sprague et al., 2016).
In terms of socio-demographic characteristics, the results of the present work show greater participation by women (61.1%) than men (39.9%).This apparent imbalance most likely reflects the primary focus of some of the healthcare units included in the study.For example, gynecology/obstetrics and maternal-infant health services are primarily attended by women, and as a result, more women report and discuss their experiences of IPV with healthcare professionals in these settings.Additionally, previous studies have reported that a relatively large number of women (67%) voluntarily disclose their encounters with IPV when questioned by a physician (Yut-Lin and Othman, 2008) in a respectful and considerate manner (Brown et al., 2000).

Conclusion
The present study has successfully validated the WAST-Short in a Mozambican context as an effective instrument for IPV screening in specific healthcare facilities in two Mozambican provinces.The WAST-Short proved to be a more reliable instrument than an adapted clinical interview that included specific questions about IPV.In comparison to the standard clinical interview, the two questions in the short version (WAST-Short) provide insight into the intimate relationship, measuring the presence of violence, which is information not easily accessible through the standard clinical interview.Therefore, the WAST can be applied for IPV screening in high patient volume settings and efficiently implemented in primary healthcare units, which can then refer patients for specialized care.The results have shown that WAST-Short is a simple, easy-to-apply, and comprehensive IPV screening tool that can be used by any health professional in any healthcare service.This is especially important for providing HIV prevention counseling and antiretroviral treatment information.HIV/AIDS-focused healthcare services serve as a significant point of entry for identifying, informing, and assisting women affected by violence.Providing individuals conducting HIV counseling and healthcare providers treating patients experiencing domestic violence with instruction on the use of the WAST-Short will sensitize them to the importance of connecting patients with related services such as psychosocial therapy, prevention of mother-tochild transmission (PMTCT), and antiretroviral treatment (Basile et al., 2007;Li et al., 2014).

Limitations
The study has certain limitations that should be discussed.Firstly, it is important to acknowledge that violence in general and IPV, in particular, is often concealed, underreported, or inaccurately reported due to cultural norms that discourage victims from sharing their experiences.This limitation implies that the subjects used to validate the screening instrument described here may represent a selective population of victims.
Secondly, this study was conducted in a single geographic area (although in 2 provinces) of Mozambique, a country with a diverse and heterogeneous population with varying cultural norms.Consequently, the results of this study may not be readily generalizable to other regions of Mozambique without further locationspecific investigations.
Thirdly, while IPV is more commonly experienced by women than men, our cohorts included a significantly higher number of women than men due to the nature of the two healthcare units from which subjects were recruited.This potential bias may have been further exacerbated by the fact that we did not perform a sample size calculation.As a result, the study's findings may not be adequately powered to compare the two screening instruments.Nonetheless, according to Jones (2003), the sample size used would be expected to provide estimates of the performance of the WAST tools with a precision of at least 5%.

Table 1 .
Socio-demographic data of the study participants.

Table 2 .
Measures of agreement for approach AB.

Table 3 .
Measures of agreement for approach BA.

Table 5 .
Adjusted odds ratio estimates for incorrect classification of IPV.