Muller 2009

Methods	Non‐randomised, 3‐arm trial Setting: paediatric HIV clinic (Cape Town, South Africa) Adherence measures: Treatment adherence measured by MEMS every three months, and measured as >80%, >85%, >90%, and >95% in various analyses. Virological status measured every six months according to local protocols; virological suppression was defined as HIV RNA levels ≤50 copies/mL. Outcomes were measured over a two‐year period.
Participants	Children under age 10 Mean age: 51 months (standard deviation, 24) (age range not reported) Female: 40% Median time on ART: 28 months (interquartile range, 12‐38)
Interventions	Children received ART regimens based on several factors, including a history of exposure to nevirapine, receipt of concomitant tuberculosis therapy, age and date they began therapy. The three arms of the study included three different regimens: an LPV/r‐containing triple regimen with zidovudine, lamivudine, stavudine, or abacavir as the NRTI; unboosted ritonavir plus 2 NRTIs; and 2 NRTIs plus either nevirapine or efavirenz as the NNRTI.
Outcomes	Percentage of children who achieved >80% adherence at 3 months: LPV/r‐containing regimen: 64% (25 of 39) Unboosted PI regimen: 100% (6 of 6) NNRTI‐containing regimen: 57% (12 of 21) p=0.781 comparing the LPV‐r‐containing regimen to the NNRTI‐containing regimen Percent of children who achieved viral load ≤50 copies/mL at three months: LPV/r‐containing regimen: 77% (30 of 39) Unboosted PI regimen: not reported NNRTI‐containing regimen: 33% (7 of 21) p=.002 comparing the LPV‐r‐containing regimen to the NNRTI‐containing regimen
Notes	All eligible patients attending the clinic between February and April 2006 were asked to participate. Inclusion criteria for the study were: children <10 years receiving ART with liquid drug formulations with a monthly amount of not over 600 mL for the MEMS‐monitored drug lamivudine or abacavir, with one identified caregiver responsible for medication administration. Children and caregivers were excluded from the study if the caregiver refused to participate or if the child had multiple caregivers or lived in an orphanage setting. Demographic and other differences between treatment arms at baseline were not reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	High risk	Participants were not randomised; treatment arm was determined by treatment history and age.
Blinding (performance bias and detection bias) All outcomes	High risk	Neither the investigators nor the participants were blinded.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Of 78 children enrolled in the study, one was transferred to another treatment centre, one had a dysfunctional MEMS device, three reported incorrect MEMS use, and blood samples obtained from nine children were insufficient for virological analysis. (It was not reported which treatment arm these children were in.) The authors report that this leaves 66 in the final analysis, whereas in fact it should have left only 64.
Selective reporting (reporting bias)	Low risk	All outcomes reported in the methods section were reported on in the results.
Other bias	High risk	The small sample size and short duration of follow‐up (three months) limits the findings, especially with respect to virological outcomes.