Ethics statement

This study is based on the analysis of secondary data from an original study (RSS trial protocol registered on ClinicalTrials.gov (Identifier: NCT03237585)), that obtained ethical approval from the Institutional Review Board of the Noguchi Memorial Institute for Medical Research of the University of Ghana (Protocol ID # 006/15-16) and the South African Medical Research Council’s Ethics Committee (Protocol ID # EC031-9/2015) [15]. Hence, the secondary data analyses did not require additional ethical approval from an Ethical Review Board. However, permission was obtained for the use of data for this study from the original study investigators, who are co-authors on this manuscript. The original study obtained written informed consent from all participants. Prior to participation in the survey, researchers discussed the study’s consent form with respondents, which included explanation of study objectives, potential risks, benefits, confidentiality of information and voluntariness of participation. Respondents were assured of anonymity with the use of unique identification codes that could not be traced to them. Thus, the data used for this study was fully anonymized before being accessed for this study and identity of the respondents is not known to the authors of this study. The secondary data obtained was strictly used to address the objectives of this study for which it was intended and nothing else.

Study design and population

A cross-sectional study using secondary data from a mixed-method unmatched cluster-randomized controlled trial carried out in four districts (two inland and two coastal) in the Central Region of Ghana was conducted. The four districts were the Upper Denkyira, Agona, Abura-Asebu-Kwamankese and Komenda-Edina-Egyafo-Abirem municipal districts. Upper Denkyira and Agona districts were considered “inland” districts, whilst Abura-Asebu-Kwamankese and Komenda-Edina-Eguafo-Abirem (KEEA) municipal were considered as “coastal” areas of the region. The trial assessed the Rural Response System (RRS), which used Community-Based Action Teams ‘COMBAT’ in preventing violence against women (VAW) in Ghana. The study involved 2000 randomly selected women aged 18–49 years living in 40 localities within the four districts selected for the study. Although the prevalence of HIV testing was assessed for the whole ‘COMBAT’ study population of 2000 women at baseline, only data for 1008 of the women in the study, who experienced any form of IPV were used for this study on the predictors of HIV testing among IPV-exposed women. Respondents were interviewed in English and Twi/Fanti, the local dialects of the study area.

Study variables

Data analysis

Descriptive statistics were used to summarize the characteristics of women in the study, including whether they had ever tested for HIV or not. Pearson’s chi-square tests were done to determine associations of the socio-demographic and sexual behaviour characteristics with “HIV testing” as an outcome variable. Logistic regression was then used to estimate adjusted odds ratios and 95% confidence intervals for variables associated with HIV testing among women experiencing IPV. The Least Absolute Shrinkage and Selection Operator (LASSO) with a 10-fold cross validation method was used to select the most predictive variables to include in the multivariable logistic regression model. The LASSO approach selects the minimum number of variables that best predicts the outcome and, in the process, avoids over-selection and under-selection of variables. The crude and adjusted odds ratios and respective confidence intervals and p-values of the variables used in the final model are presented. The goodness of fit test of the final model had a non-significant p-value of 0.4093 indicating an appropriately fitted model. The mean variance inflation factor (VIF) was 2.76 (range: 1.32 to 7.08), which is less than 10 indicating an acceptable level of multicollinearity among the variables used in the multivariable logistic regression model. All statistical analyses were performed using STATA IC Lasso Reference Manual version 16.

Characteristics of study participants

The mean age of the respondents for this study was 32.1 ± 8.59 years. About 55.5% of the respondents lived in inland areas and 44.5% in coastal areas. The study population was quite stable with 18.9% having lived in their communities for more than 29 years. The mean number of years that respondents had lived in their community was 17.4 ±12.36 years. Most of the respondents (64%) had more educated male partners. However, 14.6% of the respondents were more educated than their male partners. Also, 83.0% were employed in the past 12 months and 96.6% had partners who were unemployed. About 66.7% of respondents reported not being satisfied with life. Regarding the sexual-behaviour characteristics, a few respondents (11.1%) had their initial sexual experience with their husbands, while most respondents (77.1%) had their first sexual experience with their boyfriends. Most respondents (87.3%) had one sex partner and 17.9% had been involved in two or more transactional sexual relations. Only 30% of respondents had ever used condom. Regarding the prevalence of HIV testing, the study found that less than half (43.3%) of all study participants involved in the baseline assessment for the cRCT had ever tested for HIV in their lifetime. Prevalence of HIV testing among respondents who had experienced IPV was 42.4% compared to 44.2% among those who had not experienced any form of IPV. The sociodemographic and sexual-behaviour characteristics of study participants are presented in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of study participants (N = 1008).

https://doi.org/10.1371/journal.pgph.0000376.t001

Sociodemographic and sexual behaviour factors associated with HIV testing

The socio-demographic characteristics of women experiencing IPV that were significantly associated with HIV testing were age of respondents (χ² = 35.7 p-value<0.001); area of residence (χ² = 23.5, p-value<0.001); years lived in their community (χ² = 10.6, p-value = 0.014); highest level of education (χ² = 42.1, p-value<0.001); ever moved or travelled outside the community for work (χ² = 11.8, p-value = 0.001); and having employment within 12 months preceding the survey (χ² = 6.1, p-value = 0.013). The sexual behaviour factors that showed significant association with HIV testing among respondents from the Pearson’s chi-square test were age at first sex (χ² = 7.5, p-value = 0.024) and ever use of condoms (χ² = 16.5, p-value<0.001). HIV testing was significantly different across various levels of the sexual behaviour characteristics. The socio-demographic and sexual behaviour factors associated with HIV testing among respondents are shown in Table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Bivariate analysis of factors associated with HIV testing among respondents.

https://doi.org/10.1371/journal.pgph.0000376.t002

Predictors of HIV testing among women experiencing IPV

Age, education, employment, residence and condom use and reported satisfaction with life were found to be significant predictors of HIV testing among women experiencing IPV in the study area, after adjusting for confounding factors. Younger women aged 25–39 years were more than twice as likely to test for HIV (AOR:>2.0; 95% CI:1.49–3.52) compared with those above 45 years (Table 3). With regards to education, the likelihood of HIV testing increased with increasing levels of education, although the likelihood of testing among those with primary education was not significantly different from those without formal education. In comparison to respondents without formal education, the odds of HIV testing among women who attained various levels of education were as follows: Primary School (AOR:1.27; 95% CI:0.84–1.93); Junior High School (AOR:2.10; 95% CI:1.42–3.12); Senior High School (AOR:3.87; 95% CI:2.07–7.26); and all 10 respondents with Post Senior High School education (100%) had tested for HIV. Women who were employed during the last 12 months were less likely to test for HIV than those unemployed (AOR:0.66; 95% CI; 0.45–0.96), and those resident in inland areas were also less likely to have an HIV test compared with those living in coastal areas (AOR:1.97; 95% CI: 1.45–2.67). Women who had ever used condoms had a higher odds of testing for HIV than those who had never used condoms (AOR:1.42; 95% CI:1.05–1.93). HIV testing was significantly lower among women whose first sexual encounter was with a boyfriend compared to a husband (AOR: 0.46, 95% CI: 0.29–0.72, p = 0.001). HIV testing was also higher among women who reported satisfaction with their lives (AOR: 1.44, 95% CI: 1.08–1.92, p = 0.014) (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Logistic regression model of predictors of HIV testing among respondents.

https://doi.org/10.1371/journal.pgph.0000376.t003

Prevalence of IPV among women in the study population

Data from the four districts of the Central region of Ghana, which informed the current study revealed that more than half (50.8%) of the women had experienced some form of IPV in their lifetime [16]. This estimate, which compares well with previous estimates of lifetime experience from the GDHS (2008) of 38.7% and the 12-month experience of 27.7% from the GFLHS (2015) [10, 11] suggest continuing perpetuation of IPV against women in Ghana. The reported prevalence of IPV in Ghana, though higher than in other countries in the west Africa sub-region, is within the 15% to 71% range estimated for low- and middle-income countries, obtained from the WHO multi-country study on women’s health and domestic violence [17]. Prevalence of the various types of IPV experienced by study participants include sexual (36.7%), physical (63.8%), economic (20.0%) and emotional (72.8%) with overlapping occurrence. The overlapping occurrence of sexual, physical and other forms of IPV perpetuated against women by their intimate male partners have been well documented globally [6].

Prevalence of HIV testing among women experiencing IPV

Studies on HIV testing among women experiencing IPV in Ghana are non-existent. However, the prevalence of HIV testing among various population sub-groups in Ghana have been reported to be varied. Tenkorang & Owusu (2010) found a 10% prevalence of HIV testing among women in the general population of Ghana [18]. Secondary analysis of data from the nationally representative, 2014 Ghana Demographic and Health Survey found a low HIV testing prevalence rate of 13% among women and 6% among men aged 15 to 49 years in the 12 months period preceding the survey [19]. Wide variations in HIV testing rates were observed among sub-groups of the population, within regions and across the 10 regions of the country. In that study, the following subgroups had higher HIV testing rates compared with the national average of 13%: Women aged 25–34 years (19%); those married (16%); those with previous STIs (16%), those with only one sexual partner (16%); from richest households (18%); and those with higher than secondary education (26%). Clearly, age, education, marital status, income and sexual behaviour influenced HIV testing among men and women aged 15 to 49 years in the general population of Ghana [19]. The estimates obtained compared well with the previous estimates by Tenkorang & Owusu (2010) [18]. Gyasi & Abass (2018) reported an HIV testing prevalence of 22% among youth in Metropolitan Kumasi of the Ashanti region of Ghana [20]. Although differences in HIV testing rates reported from various studies can be attributed to several factors including study designs, testing settings, study population as well as geographic locations, in general, HIV testing rates in Ghana are considered low, but have been observed to be increasing [21].

The prevalence of HIV testing among women exposed to IPV in the study setting (42.4%) was similar to the prevalence observed among women who had not experienced any form of IPV (44.2%). This was a particularly interesting observation given the documented increased risk of HIV infection among women exposed to IPV. We did not find any documented studies in Ghana or elsewhere reporting similar observations with any plausible explanations. However, potential explanations could include a lack of awareness of the increased risk to HIV infection among women exposed to IPV and/or poor access to HIV testing services, resulting in a low uptake of HIV testing services among these women. Although the observed HIV testing rates among women exposed to HIV in the Central Region are higher than those observed in other areas and sub-groups in Ghana [18–20], the prevalence of HIV testing among women experiencing IPV in the Central region (42.4%) is lower than the reported rate of 82% among women exposed to IPV in Uganda [22]. Uganda has a dual HIV and IPV epidemic with yearly IPV prevalence of 14–33% and lifetime prevalence of 45–54%; and a relatively mature HIV control program that effectively targets at-risk populations and ensures access to HIV testing services through Provider Initiated Testing and Counselling (PITC), Voluntary Counselling and Testing (VCT) and awareness creation through education and information services. Hence, the higher uptake of HIV testing in that population [22]. The relatively low prevalence of HIV testing among women who had experienced IPV in the current study population in Ghana, demonstrates unmet need and gaps in testing services, as a greater proportion of women with IPV, who are at an increased risk of HIV infection, had not had an HIV test at the time of the study. The gaps may exist as a result of inadequate information provided on testing options available to these and other vulnerable groups such as young people and adolescents. It further aligns with the assertion that some women experiencing IPV in their relationships report HIV risk-related behaviours but are often not tested for HIV.

In Ghana, HIV testing services are routinely offered to women attending antenatal care (ANC) clinics through PITC at no cost, towards the prevention of mother-to-child-transmission of HIV, and commercially accessible to the general population on demand, through VCT in public health facilities and private sector pharmacies [19]. Apart from the information and testing services provided to pregnant women during regular ANC services, most vulnerable women, including those exposed to IPV, are not aware of their risk to HIV infection and the need or benefit of testing for HIV. Similarly, knowledge about ART and access to ART among this sub-group of the population and the general population at large, is also poor and hampered by stigma [4, 19]. Thus, it remains essential to provide avenues for women with IPV exposure to obtain HIV test through alternative means such as self-testing and rapid voluntary counselling and testing services within HIV programs. Of critical importance is the need for appropriate health educational intervention messages to increase awareness about the increased risk of HIV infection among women exposed to IPV and the need for regular testing among this vulnerable population. Similarly, knowledge about access to ART and its benefits are essential. The Health Sector must deem it necessary to play a protagonist role in making comprehensive health care available to women imperiled by IPV and serve as an entry point for referring victims for further social support services.

Predictors of HIV testing among women experiencing IPV

The results of this study indicate that HIV testing among women who experience IPV is influenced by age, education, employment, place of residence, condom use and reported satisfaction with life. Although, there are no comparable studies on HIV testing among women exposed to IPV in Ghana, the findings are consistent with observations from analysis of the nationally representative 2014 GDHS data which identified age, education, marital status, number of sexual partners, household wealth index and region of residence as predictors of HIV testing among women aged 15 to 49 years in the general population of Ghana [19]. Young age, as a predictor of HIV testing, could be related to a lack of knowledge and awareness regarding the vulnerability of IPV exposure to HIV infection as well as poor access to services, which is well documented for young people. Education tends to empower women with literary and analytical skills that enable them to make appropriate health decisions and adopt promotive and protective health behaviours such as condom use and HIV testing. This empirical observation also explains the relationship between condom use and HIV testing. Educated women are also more likely to be economically empowered to access health services. Residence as a predictor of service utilization tends to reflect disparities in access to HIV, IPV and other health and social services in most developing countries, where urban communities tend to have better access to various services, while rural poor communities remain deprived. Other factors found to predict HIV testing among women who experienced IPV in Uganda include knowledge, access to information and decision making to access HIV services [22]. Inadequate testing levels among IPV victims are driven by the assertions that IPV is a risk factor for HIV infection and the fear of violence intimidates women and prevents them from accessing HIV testing services [23]. The relationship between IPV with HIV test result disclosure, for instance, has been shown to be negative as partners who disclosed their HIV test results suffered further and varied forms of IPV [24–26]. Women’s reported satisfaction with life as a predictor of HIV testing might reflect their socioeconomic status and improved accessibility to health services, including HIV test. The high rate of unemployment among partners of women who experienced IPV is noteworthy, given a previous observation from Ghana indicating that women whose husbands were unemployed had 2.41 to 2.58 times the odds of experiencing psychological and physical violence [27]. These observations should strengthen concerns about IPV and its relationship with HIV and the need to address this growing public health problem concurrently.

The findings of this study and others, highlight the importance of sociodemographic characteristics such as women’s education and employment, and behavioural factors such as number of sexual partners, condom use and engagement in transactional sex, among other risky behaviours, which need to be explored to target interventions to improve HIV testing options [28]. To achieve high testing rates in Africa, increasing the availability of testing sites, as well as provider-initiated testing within routine healthcare remain significant interventions. These should be coupled with promoting home-based self-testing to minimize the effect of stigma and related violence, increasing testing centers in communities, as well as decentralized systems for women to know their HIV status, as such systems have been found to be acceptable and accessible to vulnerable groups [29]. These actions are inherent in the efforts to empower women and other vulnerable groups as interventional attempts to reduce the increasing trends of HIV infection and IPV among women [30].

The findings also highlight the need for sensitization campaigns to increase awareness of the importance of HIV testing among women exposed to IPV. This observation should inform programmatic efforts targeted at health education and promotion, women’s empowerment and access to information through media. The diverse interplay of factors influencing HIV testing among women who experience IPV should direct the structuring of interventions to improve access to testing services for this at-risk population and inform a multisectoral approach to reducing the incidence of IPV and improving access to HIV testing among women exposed to IPV.

Importance of the study and global health implications

The relationship between intimate partner violence and HIV infection is well documented. However, data on the extent of HIV testing among women exposed to IPV and the predictors of HIV testing among these women, which are essential to inform policy and programming to address the dual menace of IPV and HIV is scant, globally, and non-existent in Ghana. This study is the first to document the prevalence and predictors of HIV testing among women exposed to IPV in Ghana and second only to Uganda in the Africa region, where the burden of HIV infection is highest, globally. Both HIV and IPV are global health challenges with gender dimensions. Thus, understanding the factors associated with the uptake of HIV testing services among women exposed to IPV is important. The outcome of this study is expected to influence policy and programming and elicit the design of further studies to unravel the critical issues surrounding the interactions between IPV and HIV testing among women, globally.

Study limitations

The current analysis used data from the baseline assessment of a cRCT, which assessed the Rural Response System using Community-Based Action Teams in preventing violence against women in four districts of the Central Region of Ghana. Hence, the findings may not be generalizable to women outside the Central Region of Ghana. However, the results offer useful clues about the interactions between IPV and HIV testing among women.