Understanding women's engagement in HIV care after initiating antiretroviral therapy during pregnancy in South Africa

• Main Author: Phillips, Tamsin Kate
• Other Authors: Myer, Landon
• Format: Doctoral Thesis
• Language: Eng
• Published: Faculty of Health Sciences 2019
• Subjects: Epidemiology & Biostatistics,
• Online Access: http://hdl.handle.net/11427/30428

Background: Sustained engagement in HIV care, including adherence to antiretroviral therapy (ART) and retention in HIV services, is essential to optimize maternal health and prevent perinatal, postnatal and sexual HIV transmission. However, engagement in care remains a substantial challenge for pregnant and postpartum women. Women’s experience of and response to barriers to engagement in care, including ART side effects, transfer of care and mobility, may be altered by the transitions experienced in pregnancy and motherhood, and there have been few quantitative analyses of these risk factors in maternal ART cohorts. The way engagement in HIV care is measured also varies widely and no gold standard measures of ART adherence or retention exist. Composite assessments of adherence and retention, including drug concentrations, longitudinal self-reported adherence, and interlinked routine electronic health data, have not been thoroughly evaluated among African women living with HIV. To address these gaps in knowledge, this thesis investigates novel measures of ART adherence, and evaluates interlinked routine electronic health data to measure retention in a South African maternal ART cohort. It describes maternal engagement in HIV care, and examines barriers to engagement that require consideration specific to maternal ART. Methods: This research included women who initiated ART during pregnancy in a large integrated antenatal care and ART clinic in Gugulethu, South Africa (2013-2014). Until July 2013, only women who met certain clinical criteria (CD4 cell count <350 cells/µL or disease stage III or IV) were eligible for lifelong ART, thereafter guidelines changed to recommend lifelong ART for all pregnant women living with HIV. In this setting, all women receive ART and antenatal care in an integrated clinic during pregnancy and are required to transfer to a general ART clinic postpartum. Data were obtained from questionnaires (including demographics, self-reported ART adherence and self-reported side effects) and blood specimens (for HIV viral load) collected at study visits approximately every three months from pregnancy through 18 months postpartum. One additional visit took place 3-4 years postpartum where blood specimens for drug concentrations were also collected. In parallel, routine electronic data, linked across clinics and data sources including HIV clinical visits, laboratory testing and pharmacy dispensing data, were obtained through 30 months on ART. Findings: Substantial disengagement from care, both non-adherence and non-retention, was observed in all analyses. At least one ART side effect was reported by 97% of women during pregnancy and high overall side effect burden was associated with reported missed ART doses. Retention worsened over the first two years on ART and 21% of women were lost immediately after transfer from the integrated clinic. Women who linked to care spread to multiple different ART clinics after transfer; 21% moved clinics two or more times. Using combined routine medical records, only 59% of women had evidence of accessing routine HIV care in consecutive 12-month windows through 24 months on ART. Among women with viral loads available, attending ≥2 clinics was associated with viraemia. In analyses of ART adherence, TFV-DP in DBS provided a more nuanced adherence measure but plasma efavirenz and tenofovir assays had similar ability to predict viral suppression. Areas under the Receiver Operating Curve were higher for all drug concentrations (all >0.850) compared to self-reported adherence using a cross-sectional three-item scale (0.756). Longitudinal measurement of the same self-reported adherence scale showed that reporting worse adherence on any of three items over consecutive visits could predict viremia (>50 and >1000 copies/mL), particularly among women who were suppressed at the initial visit. Measuring retention using routine interlinked electronic data facilitated tracing of women beyond transfer from the integrated clinic to any clinic where they accessed HIV care postpartum. Estimates of retention varied widely using different retention definitions and data sources. Overall, electronic primary health care data, linked across clinics, performed better than laboratory data alone and was a robust measure for monitoring retention in HIV care. Conclusions: Taken together, these findings underscore a concerning level of disengagement from HIV care during and after pregnancy. Potential ART side effects, required transfer of care, the potential challenges of mobility and the importance of sustained engagement in care beyond pregnancy and breastfeeding, should be emphasised in ART counselling. Drug concentrations in DBS and plasma strongly predict viral suppression, but these data on longitudinal self-reported adherence provide a proof of concept for a low resource interim adherence measure that warrants further investigation in routine care settings with limited resources for viral load or drug concentration testing. Transfer of care and postpartum mobility mean that interlinked data sources are essential to obtain accurate estimates of retention postpartum. Further evaluation of the optimal approaches to transferring maternal ART care and the development of interventions to support engagement both in and beyond the clinic of ART initiation will be critical to sustain maternal engagement in HIV care in the long term.