Coaching clinics on kinder care brought more Zambians back to HIV treatment

Keeping people on HIV treatment is a harder problem than starting them on it. A trial across 24 clinics in Zambia has tested an unglamorous fix — training and coaching staff to treat patients better — and found it measurably pulled more people back into care after they had fallen out [s1].

The study, published in PLOS Medicine on September 1, is a post-hoc analysis of a stepped-wedge cluster randomised trial [s1]. These are the numbers, and the reasons a package built around courtesy produced them.

The problem being solved

The trial did not study people who never came back. It studied people who slipped. Investigators followed clients living with HIV who became more than 30 days late to a scheduled visit — a treatment interruption — and asked whether a "person-centred care" package changed what happened next [s1].

The scale of the problem is the first finding. During the study period, 128,910 clients became more than 30 days late: 69,671 under control conditions and 59,239 under the intervention [s1]. The group was 64.3% female, with a median age of 38 years [s1]. Treatment interruption, in other words, is not a fringe event; it is a routine part of a large programme's life.

What the intervention was

The package targeted the "caring" side of health-worker behaviour rather than the clinical protocol [s1]. It had three parts: training and coaching staff on person-centred practices; measuring and feeding back client experience through exit interviews; and small facility-level incentives [s1]. The trial ran at 24 clinics from August 2019 to November 2021, with clinics crossing over from control to intervention every six months [s1].

Nothing here is a new drug or a new test. It is an attempt to change how a clinic treats someone who walks back in after missing appointments — the moment when a curt reception can turn a lapse into a loss.

What changed

Three outcomes moved, all in the same direction [s1].

Return after an interruption rose. At 12 months, 72.3% of interrupted clients had returned under the intervention versus 67.7% under control — a risk difference of 4.6 percentage points and an adjusted hazard ratio of 1.16 [s1].

Repeat interruptions fell, and this was the largest effect. Among those who returned, 44.3% interrupted again under the intervention versus 55.6% under control — a risk difference of 11.4 percentage points and an adjusted hazard ratio of 0.50 [s1]. Halving the rate of second interruptions is what turns a single return into sustained engagement.

Sustained engagement improved accordingly. The proportion in care 12 months after returning was 82.7% under the intervention versus 73.4% under control [s1]. Combining return and retention, the overall share of interrupted clients in care at 12 months rose from 51.5% to 60.1% — a risk difference of 8.7 percentage points and an adjusted risk ratio of 1.19 [s1].

Why it plausibly works

The mechanism the authors point to is retention, not re-recruitment. The biggest single number is the drop in repeat interruptions, which suggests the intervention did most of its work by changing the experience of being in care — making the second and third visits more likely to happen — rather than merely coaxing a one-time return [s1]. That is consistent with a package aimed at how staff behave over time, not at a single outreach contact.

Why it matters beyond Zambia

The stakes are set by the treatment cascade. WHO estimates that 41.0 million people were living with HIV at the end of 2025, 64% of them in the African Region [s2]. Of people who know their status, 89% are on antiretroviral therapy, and 95% of those on treatment have suppressed viral loads [s2] — targets that only hold if people stay on therapy. Interruptions erode viral suppression and drive onward transmission, so an intervention that keeps returners in care acts on the same lever as the "95-95-95" goals [s2].

The caveats

This was a post-hoc analysis, meaning the reengagement question was examined after the trial rather than being its pre-registered primary aim — a design that can find real effects but warrants confirmation [s1]. The confidence intervals on the overall retention effect are wide, with the adjusted risk ratio of 1.19 ranging from 1.05 to 1.35 [s1]. And the package bundles three components, so the trial shows that the combination worked without isolating which part — training, feedback or incentives — did the lifting.

What to watch

Whether the effect survives outside a trial, when coaching and exit interviews compete with everyday clinic pressure and the small incentives end.

Which component carries the effect, since programmes replicating this cannot afford to fund all three blindly.

And whether the gains reach viral suppression, not just attendance — the outcome that ultimately determines whether a returned patient is protected and non-infectious.

This article is informational and is not medical advice.