Adherence remains the modifiable step between therapy and suppression

Design

systematic review and random-effects meta-analysis of observational studies

Population

39 studies of people living with HIV on antiretroviral therapy in Ethiopia

Primary outcome

pooled prevalence of good adherence and of viral load suppression, and their association

Effect

adherence 79.4% (95% CI 74.8–83.4), suppression 77.5% (72.5–81.8), odds ratio 6.30 (4.84–8.19), I² >90%

Thirty-nine Ethiopian studies were pooled to estimate adherence to antiretroviral therapy and viral load suppression among people living with HIV. Pooled good adherence was 79.4% (95% CI 74.8 to 83.4) and pooled viral suppression 77.5% (72.5 to 81.8). Good adherence was strongly associated with suppression, odds ratio 6.30 (4.84 to 8.19).

Both figures sit below the international 95% targets, and the gap between them is small — which suggests that in this setting suppression is tracking adherence closely rather than being limited by regimen failure or resistance. Heterogeneity was substantial, with I² above 90% for both pooled prevalence estimates, so the summary numbers describe a range of very different programmes rather than a single national figure.

The association is observational and the direction is not established by the data: people who are well and engaged are both more likely to take tablets and more likely to be suppressed. But the size of the odds ratio, and the consistency of the direction, make adherence support the highest-yield place to intervene in a programme that is already delivering therapy. That reasoning transfers directly to Indian programmes, where the constraint is more often engagement and follow-up than drug availability.

• Measure adherence with something better than a global impression — pill counts, pharmacy refill dates, or self-report against specific weeks
• Investigate an unsuppressed viral load as an adherence question first, before assuming resistance
• Identify the patients who miss appointments rather than the ones who attend and report well
• Record the barrier when adherence is poor: cost, distance, side effects, disclosure

Why it matters

It locates the gap to global targets in adherence support rather than in the drugs.

Don't overread it

Observational and pooled across very heterogeneous studies — it identifies an association, not a proven causal lever.

The statistics, in plain English

An odds ratio of 6.30 with an interval of 4.84 to 8.19 is a strong and precise association, but this is pooled observational data — it cannot show that improving adherence causes suppression, only that the two travel together. I² above 90% means the included studies disagree substantially, so the pooled prevalence figures should be read as a central tendency across heterogeneous programmes rather than a national rate.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.