Cultivating a universal understanding of ART adherence: Comments on "A meta-analysis of effectiveness of interventions to improve adherence in pregnant women receiving antiretroviral therapy in sub-Saharan Africa".

Dear Editor-in-Chief,

We would like to congratulate the authors of “A meta-analysis of effectiveness of interventions to improve adherence in pregnant women receiving antiretroviral therapy in sub-Saharan Africa,” (Omonaiye et al., 2018) on their efforts to summarize a body of literature crucial to improving HIV care for pregnant women living with HIV (WLWH) in sub-Saharan Africa. This work makes important contributions to the field. However, there are two critical points we would like to highlight for reader and author consideration.

First, the authors do not incorporate date restrictions into meta-analysis inclusion criteria. Thus, included studies span multiple anti-retroviral treatment (ART) eras (Omonaiye et al., 2018). In 2013, the World Health Organization updated ART guidelines for the prevention of mother to child transmission (PMTCT) of HIV, recommending implementation of Option B+, or lifelong ART, for all pregnant WLWH. Prior to Option B+, single-dose Nevirapine and short-course ART treatment regimens, such as daily monotherapy from 14 weeks gestation through seven days post-partum, were utilized in PMTCT programs (Pricilla et al., 2018). Given the meaningful differences in treatment recommendations across these eras, disaggregating this meta-analysis by treatmentera, or restricting inclusion criteria to studies conducted during the implementation of Option B+, would improve the relevance of study findings for implementation under current treatment paradigms.

Differences in treatment recommendations also yielded changes in measurement of PMTCT outcomes, our second point for consideration. In the era of single-dose and short course ART treatments, PMTCT programs often captured PMTCT service or treatment uptake. In the era of lifelong ART, programs often measure ART initiation or adherence. Omonaiye and colleagues identified ART adherence as the outcome of interest in this meta-analysis (Omonaiye et al., 2018). However, they appear to have included studies capturing other care outcomes, such as ART initiation (Dillabaugh et al., 2012; ENHAT-CS, 2014; Stinson et al., 2013; Herlihy et al., 2015), or receipt of treatment or PMTCT services (Kalembo et al., 2013; Finocchario-Kessler et al., 2014; Turan et al., 2015) in the meta-analysis. That is, estimates of ART uptake or receipt of PMTCT services appear to be pooled with true estimates of ART adherence (e.g., pill counts, drug level testing) (Castillo-Mancilla and Haberer, 2018) to estimate the effect of interventions on ART adherence.

To improve reliability and utility of this meta-analysis, authors could expand their search terms to include multiple HIV care outcomes of interest (e.g., ART initiation, adherence, viral suppression) and stratify their analysis by outcome and treatment era. By disaggregating effect estimates in these ways, this meta-analysis has the potential to better identify unique intervention targets for improving PMTCT outcomes among pregnant WLWH in settings with diverse epidemic dynamics.