Pharmacy Refill Data are Poor Predictors of Virologic Treatment Outcomes in Adolescents with HIV in Botswana.

In adults living with HIV, pharmacy refill data are good predictors of virologic failure (VF). The utility of pharmacy refill data for predicting VF in adolescents has not been reported. We evaluated data from 291 adolescents on antiretroviral therapy. The main outcome measure was VF, defined as two consecutive HIV viral load measurements ≥ 400 copies/mL during 24-months of follow-up. Pharmacy refill non-adherence was defined as two consecutive refill adherence measurements < 95% during the same period. Fifty-three (18%) adolescents experienced VF. One hundred twenty-eight (44%) adolescents had refill non-adherence. Refill non-adherence had poor discriminative ability for indicating VF (receiver operating characteristic AUC = 0.60). Sensitivity and specificity for predicting VF was poor (60% (95% CI 46-74%) and 60% (95% CI 53-66%), respectively). The lack of a viable surrogate for VF in adolescents highlights the urgent need for more access to virologic testing and novel methods of monitoring adolescent treatment adherence.