Caregiver-reported antiretroviral therapy non-adherence during the first week and after a month of treatment initiation among children diagnosed with HIV in Ethiopia

To achieve optimal virologic suppression for children undergoing antiretroviral therapy (ART), adherence must be excellent. This is defined as taking more than 95% of their prescribed doses. To our knowledge, no study in Ethiopia has evaluated the level of treatment adherence at the beginning of the child's treatment. Our aim was therefore to evaluate caregiver-reported ART non-adherence among children and any predictors for this during the early course of treatment. We conducted a prospective cohort study of 306 children with HIV in eight health facilities in Ethiopia who were registered at ART clinics between 20 December 2014 and 20 April 2015. The adherence rate reported by caregivers during the first week and after a month of treatment initiation was 92.8% and 93.8%, respectively. Our findings highlight important predictors of non-adherence. Children whose caregivers were not undergoing HIV treatment and care themselves were less likely to be non-adherent during the first week of treatment (aOR = 0.17, 95% CI: 0.04, 0.71) and the children whose caregivers did not use a medication reminder after one month of treatment initiation (aOR = 5.21, 95% CI: 2.23, 12.16) were more likely to miss the prescribed dose. Moreover, after one month of the treatment initiation, those receiving protease inhibitor (LPV/r) or ABC-based treatment regimens were more likely to be non-adherent (aOR = 12.32, 95% CI: 3.25, 46.67). To promote treatment adherence during ART initiation in children, particular emphasis needs to be placed on a baseline treatment regimen and ways to issue reminders about the child's medication to both the health care system and caregivers. Further, large scale studies using a combination of adherence measuring methods upon treatment initiation are needed to better define the magnitude and predictors of ART non-adherence in resource-limited settings.