Methods: Participants (N = 373) were HIV infected women on first line regimens at week 32 of pregnancy recruited from rural clinics. Women completed demographic and self-reported ARV adherence assessments [7-day Visual Analog Scale (VAS), and skipped medication in the past week]. If no missed doses were reported, responses were coded as adherent; all other responses were coded as nonadherent. DBS ARV (TDF, 3TC and EFV) adherence was assessed and dichotomously coded as detectable or undetectable. Adherence was defined as 3 drugs detected or TDF + EFV detected. Nonadherence was defined as no detectable ARV, one ARV detectable, or two ARVs that were not TDF + EFV detectable.
Results: Median age was 28 ± 6; 39% of women had monthly income less than $32, 78% were unemployed, and 52% had completed at least 10 years of education. DBS ARV detection was as follows: adherent = 74% (2% TDF + 3TC + EFV and 72% TDF + EFV); nonadherent = 26% (11% no ARV, 11% 1 ARV, and 4% 2 ARVs that were not TDF + EFV). Overall proportions of adherent participants were similar across measures: 69% (VAS), 71% (skipped medication), and 74% (DBS). However, intermeasure agreement was low between the number of skips and DBS testing VAS (κ = 0.053), and the VAS and DBS testing (κ = 0.054) such that participants who had nondetectable levels of ARVs were categorized as adherent by self-report measures.
Conclusion: Detectable levels of ARV were suboptimal in pregnant HIV infected women in rural South Africa, indicating a risk of perinatal HIV infection and development of resistance. Programs to improve ARV adherence in this population are urgently needed. Findings suggest that self-reported adherence measures do not correlate with biological markers in this population. Validation of self-reported ARV adherence among pregnant HIV infected women in South Africa may be warranted to support PMTCT outcomes.
V. Rodriguez, None
K. Peltzer, None
S. Weiss, None
D. Jones, None