Intimate partner violence is a barrier to antiretroviral therapy adherence among HIV-positive women: Evidence from government facilities in Kenya

listen audio

Study Justification:
– Intimate Partner Violence (IPV) is linked to low engagement with HIV management services and poor antiretroviral therapy (ART) adherence.
– Previous studies in sub-Saharan Africa have produced conflicting evidence regarding the association between IPV and ART adherence among different groups of women.
– This study aims to investigate the association between IPV and ART adherence among a broad group of HIV-positive women in Kenya.
Study Highlights:
– The study sampled 408 HIV-positive women receiving free ART from government health facilities in Kenya.
– The majority of participants had attended school, were in monogamous marriages, and had disclosed their HIV status to partners.
– 60% of participants reported optimal ART adherence (≥95%) in the previous 30 days.
– The prevalence of IPV by the current partner was 76%.
– Experiencing physical IPV, sexual IPV, or controlling behavior reduced the odds of achieving optimal adherence, while a higher education level and having an HIV-positive partner increased the odds.
Study Recommendations:
– ART programs should consider incorporating basic IPV interventions into regular clinic services to identify, monitor, and support HIV-positive women who are exposed to IPV.
– Further research is needed to better understand how IPV affects ART adherence.
Key Role Players:
– Government health facilities in Kenya
– Ministry of Health
– AMPATH partnership
– Consortium of universities
– Clinical officers and nurses providing and supervising ART
– Local government authorities
– Non-governmental organizations involved in gender-based violence prevention/women empowerment work
Cost Items for Planning Recommendations:
– Training for healthcare providers on IPV interventions
– Development and implementation of IPV screening tools
– Resources for identifying and referring women to social and legal assistance services
– Monitoring and evaluation of IPV interventions
– Research funding for further studies on the impact of IPV on ART adherence
Please note that the cost items provided are general suggestions and may vary depending on the specific context and resources available.

The strength of evidence for this abstract is 8 out of 10.
The evidence in the abstract is strong, but there are a few areas for improvement. The study design is a cross-sectional survey, which limits the ability to establish causality. Additionally, the sample size of 408 women may not be representative of the entire population of HIV-positive women in Kenya. To improve the evidence, future research could consider using a longitudinal design to establish temporal relationships and increase the sample size to improve generalizability.

Introduction Intimate Partner Violence (IPV) is linked to low engagement with HIV management services and adverse clinical outcomes, including poor ART adherence. In sub-Saharan Africa, studies on pregnant/postpartum women and transactional sex workers have produced divergent evidence regarding IPV’s association with poor ART adherence. We investigate this association among a broad group of women. Methods We sampled 408 HIV-positive women receiving free ART from different types of HIV clinics at government health facilities, assessing for IPV exposure by a current partner, ART adherence rate, and other factors that affect ART adherence (e.g. education, disclosure). ART adherence rates were measured using the Visual Analogue Scale (VAS); responses were dichotomised at a ≥95% cut-off. Multiple logistic regression models assessed the association between the independent variables and ART adherence. Results The participants’ mean age was 38.6 (range: 18–69 years). The majority had ever attended school (94%, n = 382), were in monogamous marriages (70%, n = 282), and had disclosed status to partners (94%, n = 380). Overall, 60% (n = 242) reported optimal ART adherence (≥ 95%) in the previous 30 days. The prevalence of IPV by the current partner was 76% (CI95 = 72–80%). Experiencing physical IPV (AOR 0.57, CI95: 0.34–0.94, p = .028), sexual IPV (AOR 0.50, CI95: 0.31–0.82, p = .005), or controlling behaviour (AOR 0.56, CI95: 0.34–0.94, p = .027) reduced the odds of achieving optimal adherence, while a higher education level and having an HIV-positive partner increased the odds. Conclusion IPV is common and is associated with suboptimal ART adherence rates among a broad group of HIV-positive women. ART programs could consider incorporating basic IPV interventions into regular clinic services to identify, monitor and support exposed women, as they might be at risk of poor ART adherence. Still, there is need for more research on how IPV affects ART adherence.

This cross-sectional survey was conducted in March and April 2018 at twelve HIV clinics in government health facilities in Kenya. The HIV clinics are part of the AMPATH (Academic Model Providing Access to Healthcare) partnership that cooperates with the Ministry of Health and a consortium of universities to offer free HIV care and treatment services [43]. At the time of the survey, there were 79,728 HIV-positive women (aged ≥15 years) enrolled in the ART program; 40,370 were actively receiving ART [44]. We selected six urban and six rural clinics for the survey based on their number of active HIV-positive women receiving ART; the clinics were chosen to account for the socioeconomic diversity of the counties served by the AMPATH partnership. We included large clinics with many women active on ART, as well as smaller health facilities that we purposively selected to represent less populous communities. The study population consisted entirely of HIV-positive women who were actively receiving free ART at the HIV clinics. Eligible respondents were at least 18 years old, currently in an intimate partner relationship, and had begun ART at least six months before the survey. We set a minimum duration of six months since beginning ART in an effort to reduce factors affecting adherence that are related to the ART initiation. Additionally, according to the Kenya national ART guidelines, it is expected that at six months the clients should be compliant since they have sufficient understanding of HIV, medication dosage, their clinical progress, and the importance and benefits of ART adherence [45]. The sample included 408 women. The sample size was calculated through a power proportion test [46] in R (stats package, R Core Team 2013), using the national IPV prevalence rate of 47% [47] as the proportion, at a power of 0.8 and a significance level of 0.05, and with the aim of conducting a logistic regression. Due to feasibility considerations, we decided to use the prevalence rate of physical violence (45%) as the proportion, in order to get a workable sample size and because it was the most reported form of violence among women [47]. Proportional stratified sampling was applied to determine the number of women to sample from each of the selected clinics. That is, the number of women active on ART at each clinic was weighed against the total number of women on ART at AMPATH to determine each clinic’s share of the total sample. Exposure to physical IPV, sexual IPV, emotional IPV, and controlling behaviour by a current intimate partner was measured using the Demographic Health Survey (DHS) module on Domestic Violence. This module is a modified version of the Conflict Tactics Scale by Murray A. Straus and is widely used to measure spousal violence within the household context [47–49]. An intimate partner is defined as someone to whom the woman was currently married (whether in a monogamous or polygamous marriage) or with whom she was in a romantic relationship. The DHS questions on IPV depict specific forms of physical IPV (e.g. slapping, kicking), sexual IPV (i.e. use of physical force or threats to have sexual intercourse), emotional IPV (e.g. humiliation, insults), and controlling behaviour (hindrance of social contact). Each can be answered with ‘No’ or ‘Yes’. A ‘Yes’ response to any question was considered to constitute exposure to IPV. The DHS module on domestic violence measures both lifetime IPV and IPV within the previous 12 months by a current or former partner. By recruiting only women who were currently in a relationship, and slightly altering the DHS questions from ‘Did your (last) husband/partner ever. . .’ to ‘Did your husband/partner ever…’, we focused on exposure to IPV by a current partner within the lifetime of the ongoing relationship (relationship-specific IPV). Our interest was investigating whether being in an environment where IPV occurs affects the ability to adhere to ART. To measure ART adherence, we used the AIDS Clinical Trials Group (ACTG) Adherence Follow-up Questionnaire and the Brief Adherence Self-report. The latter contains a 30-day Visual Analogue Scale (VAS) commonly used in resource-limited settings because it is practical, easy to administer, inexpensive, and does not require high literacy levels [7,50–53]. The participants were asked to best estimate the percentage of ARV dosage they took in the last 30 days by marking an X or O on the VAS line measuring from 0% to 100%. Selecting 0 indicated that they had taken none of the prescribed drugs; 100% meant they had not missed a single dose. The DHS module on Domestic Violence and the ACTG Adherence Follow-up Questionnaire and Brief Adherence Self-report are validated measurement tools that have been used in similar populations and settings [7,31,32,54–56]. Moreover, the ACTG Follow up and Self-report measures are one of several ways that the HIV clinics monitor client ART adherence rates; therefore, they were already familiar to both the recruiters and the participants. For the other covariates, we used a broad set of socio-economic drivers which existing literature identifies as potentially affecting ART adherence: age, marital status, length of time on ART, education, income, HIV status disclosure, partner’s HIV status, social support from the partner (if the partner is involved in the woman’s ART), the partner’s alcohol consumption, and the area of living. The IPV and ART subscales and the questions concerning potentially relevant variables were combined into a four-page questionnaire administered in paper form. Participants were recruited from mixed HIV clinics (non-specialised clinics where female and male HIV-positive adults are reviewed and given medication refills), PMTCT clinics, maternal and child health clinics, and express clinics (for clients who are categorised as stable/virally suppressed, who therefore receive drug refills without rigorous review by a healthcare provider). The clinical officers and nurses who provide and supervise ART during regular clinical care visits recruited the participants. They were trained on the aim of the study, the tools, and the recruitment process (random selection, eligibility criteria). Randomised lists were created using Microsoft Excel, considering the number of clinical officers/nurses administering the questionnaires per clinic and the estimated number of female clients the healthcare provider attends to per day (retrieved from the daily clinic patient lists). Each healthcare provider received a randomised list which they used to recruit at least five women per day from their daily patient lists. On completion of the routine clinical check-up in the regular clinic examination rooms, women whose session coincided with the random number from the list and who fit the study inclusion criteria were informed about the survey and asked if they were willing to participate. Prior to this, if the woman was accompanied by another person or a child who was old enough to understand the conversation and old enough be left alone, the person/child was politely asked to leave the examination room. The women were assured that participation or non-participation would not affect access to treatment. After they consented to participation, an informed consent form was provided, which they signed before the questionnaire was administered. Assistance was provided for those who could not read, needed clarification, or preferred that the questionnaire be administered as an interview. All of the consent forms and questionnaires were available in English and Kiswahili. Unfortunately, some of the rooms at the clinics were shared; healthcare providers were responsible for ensuring maximum possible privacy by, for example, asking colleagues to temporarily leave the room or by moving the desk or drawing a curtain. On completion of the survey, the women were given financial compensation (KSh 100) for their participation; this was referred to as ‘transport money’. Women who reported exposure to IPV were offered a list of places where they could receive free social and legal assistance within the health facility. The list also contained local government authorities, or non-governmental organisations in the area which were involved in gender-based violence prevention/women empowerment work. Data from the questionnaires were entered and imported into R (Version 1.2.1335) for analysis. The responses regarding exposure to IPV were combined into five new variables: physical IPV, sexual IPV, emotional IPV, controlling behaviour, and overall exposure to IPV. If a woman answered ‘Yes’ to any of the questions on exposure to physical IPV, she was coded as ‘1’; if not, she was coded as ‘0’. The same was done for the other forms of IPV; overall exposure to IPV meant that the woman answered ‘Yes’ to any form of IPV. Since our hypothesis was on relationship-specific IPV, our main analysis was based on exposure of IPV at any time within the relationship. However, to identify the possible impact of this decision on our results, we repeated the modelling procedure, with only women who were exposed to IPV in the last 12 months. The responses to the VAS were used as the ART adherence outcome. Participants’ answers were dichotomised using the conservative optimal ART adherence level of ≥95% and suboptimal adherence of <95%, which was suggested by Paterson et al. [57] as necessary for achieving HIV viral suppression. We also performed a second analysis with a lower cut-off of ≥85% for comparison. Because both analyses yielded comparable results, we present analyses based on the more conservative cut off. We checked data quality and uni- and bivariate distributions using descriptive data analysis techniques. Since the questionnaire contained intimate questions that the participants could potentially skip, a missing value analysis was performed. None of the variables had more than 5% missing values. Additionally, no systematic relationships between the missing values were detected. We used simple logistic regression models to examine the relationships between each of the independent variables and the dependent variable (ART adherence). Next, ART adherence was regressed stepwise for each of the independent variables. A possible limitation of regressing ART adherence on all independent variables together is that it prevents the identification of suppressor/moderator effects and runs the risk of overfitting the model. Assumptions were checked before conducting all logistic regression analyses; log likelihood-based Pseudo R2 measures and AIC scores were used to evaluate goodness of fit. Due to multicollinearity between the IPV variables, we decided to report four multiple logistic regression models, each of which includes only one IPV variable alongside the other independent variables. Despite the nested structure of our data (multiple women clustered in each of the 12 clinics), we decided against a multilevel modelling procedure. The low intraclass correlation coefficient of ICC = .15 that we derived from an unconditional multilevel logistic regression implies that only 15% of the individual variation in the underlying propensity of low ART uptake is due to systematic differences between the clinics [58,59]. Additionally, with as few as 12 clusters, fixed-effect estimates associated with level-2 predictors could have been severely biased [60,61]. We also did not include clinics as a fixed effect in our main analysis because adding 12 additional dummy coded variables to the analysis would have led to predictor combinations with very few to zero observations. However, we performed a sensitivity analysis in which clinic was added as a predictor variable. We identified significant differences among the clinics but there were no significant changes in the respective model parameter estimates from previous models (S3 Table). Ethical approval was granted by the Moi University/Moi Teaching and Referral Hospital Institutional Research and Ethics Committee (IREC). Consent to participate was established through a written Informed Consent Form that the women signed.

Based on the information provided, it is not clear what specific innovations are being sought to improve access to maternal health. However, based on the study’s findings and recommendations, here are some potential innovations that could be considered:

1. Incorporating IPV interventions into regular clinic services: The study suggests that ART programs could consider incorporating basic IPV interventions into regular clinic services to identify, monitor, and support women who are exposed to IPV. This could involve training healthcare providers to recognize signs of IPV and provide appropriate support and referrals.

2. Strengthening social and legal assistance: The study mentions that women who reported exposure to IPV were offered a list of places where they could receive free social and legal assistance within the health facility. Innovations could focus on improving the availability and accessibility of these services, such as establishing dedicated support centers within health facilities or implementing mobile outreach programs to reach women in remote areas.

3. Technology-based solutions: Technology can play a role in improving access to maternal health services. Innovations could include the use of mobile health (mHealth) applications to provide information and support to pregnant women, remote monitoring systems to track ART adherence, or telemedicine platforms to enable virtual consultations with healthcare providers.

4. Community-based interventions: Engaging the community in promoting maternal health can be an effective approach. Innovations could involve community health workers or peer educators who provide education, support, and referrals to pregnant women, particularly those at risk of IPV. This could include training community members to recognize signs of IPV and provide appropriate support.

5. Integration of services: Integrating maternal health services with other existing programs, such as family planning or HIV care, can improve access and efficiency. Innovations could focus on creating integrated service delivery models that address the multiple needs of women, including IPV screening and support.

It is important to note that these recommendations are based on the specific context and findings of the study mentioned. Further research and context-specific assessments would be needed to determine the feasibility and effectiveness of these innovations in improving access to maternal health in other settings.
AI Innovations Description
The recommendation to improve access to maternal health based on the findings of the study is to incorporate basic interventions for intimate partner violence (IPV) into regular clinic services. This would involve identifying, monitoring, and supporting women who are exposed to IPV, as they may be at risk of poor adherence to antiretroviral therapy (ART). By addressing IPV within the context of maternal health services, healthcare providers can help improve ART adherence rates among HIV-positive women. However, further research is needed to better understand how IPV affects ART adherence and to develop effective interventions.
AI Innovations Methodology
To improve access to maternal health, here are some potential recommendations:

1. Mobile health (mHealth) interventions: Develop mobile applications or text messaging services that provide pregnant women with information on prenatal care, nutrition, and appointment reminders. These interventions can help overcome barriers such as distance, transportation, and lack of information.

2. Telemedicine: Implement telemedicine programs that allow pregnant women in remote areas to consult with healthcare providers through video calls. This can improve access to prenatal care and enable early detection of complications.

3. Community health workers: Train and deploy community health workers to provide maternal health services in underserved areas. These workers can conduct prenatal visits, provide education, and refer women to healthcare facilities when necessary.

4. Maternal waiting homes: Establish maternal waiting homes near healthcare facilities to accommodate pregnant women who live far away. These homes provide a safe place for women to stay before delivery, ensuring timely access to skilled birth attendants.

5. Financial incentives: Introduce financial incentives, such as cash transfers or vouchers, to encourage pregnant women to seek antenatal care and deliver in healthcare facilities. This can help overcome financial barriers and increase utilization of maternal health services.

To simulate the impact of these recommendations on improving access to maternal health, a methodology could include the following steps:

1. Define indicators: Identify key indicators to measure access to maternal health, such as the number of prenatal visits, facility-based deliveries, and maternal mortality rates.

2. Collect baseline data: Gather data on the current status of maternal health access in the target population. This can be done through surveys, interviews, or existing health records.

3. Develop a simulation model: Create a mathematical or statistical model that simulates the impact of the recommendations on the selected indicators. The model should consider factors such as population size, geographical distribution, healthcare infrastructure, and socio-economic characteristics.

4. Input intervention parameters: Specify the parameters of each recommendation, such as the number of mobile health users, the coverage of telemedicine services, the number of community health workers deployed, the capacity of maternal waiting homes, and the extent of financial incentives.

5. Run simulations: Use the simulation model to project the potential impact of the recommendations over a specific time period. Run multiple scenarios with different intervention parameters to assess their comparative effectiveness.

6. Analyze results: Analyze the simulation results to determine the expected changes in the selected indicators. Compare the outcomes of different scenarios to identify the most effective combination of interventions.

7. Validate and refine the model: Validate the simulation model by comparing its predictions with real-world data. Refine the model based on feedback from experts and stakeholders to improve its accuracy and reliability.

8. Communicate findings: Present the simulation findings in a clear and concise manner to policymakers, healthcare providers, and other relevant stakeholders. Use the results to advocate for the implementation of the recommended interventions and inform resource allocation decisions.

By following this methodology, policymakers and healthcare providers can gain insights into the potential impact of different innovations on improving access to maternal health and make informed decisions to prioritize and implement the most effective interventions.

Partilhar isto:
Facebook
Twitter
LinkedIn
WhatsApp
Email