Seventy-nine studies were included in the review: 64 RCTs and 15 cohort studies. There were 23,067 patients in 143 treatment groups. Mean follow-up was 14.3 months.
Efficacy: Treatment-related variables that were independently associated with higher treatment success were dosing relative to food (p=0.001), dual nucleoside backbone (favoured didanosine or tenofovir with emtricitabine or lamivudine) (p=0.002) and nonnucleoside analogue or ritonavir-boosted protease inhibitor as the third drug (p<0.0001).
Three factors related to population characteristics (nonwhites, exclusion for low haemoglobin, lower CD4 cell count) and shorter duration of follow-up were independently associated with higher treatment success. The model accounted for 79% of variability in rates of treatment success. Results of sensitivity analyses were similar to those of the primary analysis.
Tolerability: Adverse events were the most common reason for treatment cessation (9%), but were reported in only half of the included studies. The most common adverse events were diarrhoea (29%), nausea (25%), headache (18%), rash (15%) and fatigue (13%). Grade 2 or higher adverse events occurred in 21% of participants. The only adverse event significantly associated with lower efficacy was nausea (p=0.028).
There was no evidence of publication bias.