Factors affecting viral load suppression amongst pregnant women on second line ART in South Africa 2022

Abstract

Introduction: This study was a secondary analysis of data collected by The National Institute for Communicable Diseases (NICD) during the 2022 antenatal surveillance. It described the characteristics and distribution of pregnant women who were on second-line regimen, determine viral suppression levels among women on second-line antiretroviral therapy (ART), and factors associated with viral suppression in pregnant women. Methods: The study included pregnant women aged 15-49 attending antenatal care who consented to the survey and had viral load and ART status data available. Analysis was done using STATA Standard Edition version 18.0. Descriptive statistics were employed to determine the distribution of sample characteristics using proportions and percentages. Included and excluded participants were compared and the Pearson’s chi-square test was used to determine any statistical differences between the two groups. Univariable analysis was employed which was inclusive of all the variables and stratified by viral suppression, exploring two viral load thresholds (Suspected virologic failure (≥1000 copies/ml) and virally unsuppressed (≥50 copies/ml)). In the same univariable table, univariable logistic regression is presented with odds ratios (OR), 95% CI (Confidence Interval), and p-value (probability value). For the multivariable analysis ART regimen type was used as the primary exposure and viral suppression as the outcome variable. Six models were constructed. Model I and Model II were fitted using the DAGs minimal sufficient set of exposure variables and all variables considered as confounders were included in the models. Model I used VL ≥1000 copies/ml as the threshold for unsuppressed VL while Model II used VL ≥50 copies/ml as the threshold for unsuppressed VL. Model III and Model IV were multivariable logistic regression models fitted on factors that were significant at univariable logistic regression, with Model III using ≥1000 copies/ml and Model IV using ≥50 copies/ml thresholds. In both models III and IV, a category coded as missing for variables that had more than 5% of missing observations was included. Model V and Model VI were multivariable logistic regression models built on factors that were significant at univariable logistic regression, with Model V focussed on the a Suspected virologic failure (≥1000 copies/ml) and Model VI using the virally unsuppressed (≥50 copies/ml), and both these models use full case analysis of the logistic regression. Model VII and Model VIII were constructed as backwards stepwise logistic regression models stratified by ART regimen type (first-line and second-line, respectively), and focused on factors 6 associated with suspected virologic failure (≥1000 copies/ml). All the analyses conducted in this study included survey weights and we used α =0.05. Results: The study assessed 42 875 participants for eligibility for enrolment. Of those, 36 677 were eligible for enrolment with and final analysis included 8 606 participants aged 15-49 who had complete data on ART regimen and viral load. The majority of participants included in the secondary analysis were virally suppressed, on first-line ART, aged 25 to 34 years, of African race, attending follow-up visit. Moreover, most were in their second pregnancy, had one living baby, were in their third trimester not living together but in a relationship with the baby’s father, partner was HIV positive, had one sexual partner, had negative syphilis test results, and started ART before pregnancy. The rate of virally unsuppressed pregnant women was 32.5% among those on first line and 45.0% among those on second-line regimen using the ≥50 copies/ml threshold. When comparing all the models fitted, the factors that were significant through all fitted models were regimen type, age groups, first Antenatal visit type, and HIV status of current sexual partner. However, some factors, such as gestational age group, were significant from Model I to Model IV, while the number of live births was significant in all these models except for Model III. Syphilis test results were only significantly associated with viral load status in Model IV. Conclusion: The findings reveal that those on second-line ART, younger women, and those unaware of their partner's HIV status face higher odds of being virally unsuppressed. Additionally, women with multiple live births and those initiating ART during pregnancy also show increased odds of being virally suppressed. These results emphasise the need for a targeted intervention that strengthens adherence counselling and sexual reproductive health education for pregnant women on ART to improve viral suppression outcomes, particularly among youth aged 15 to 24 The study's contributions highlight the complex interplay of demographic, clinical, and social factors in achieving viral suppression, underscoring the importance of a multifaceted approach to HIV management in pregnant women