Effectiveness of Differentiated Service Delivery ART Care Among Priority Populations in Sub-Saharan Africa: Evidence from a Systematic Review and Meta-analysis

Abstract

Correction (Version 2 — 10 July 2026): This version is a truth-first rebuild of the original advanced draft, which contained placeholder PRISMA counts, a fabricated per-model subgroup table, an unsourced patient-satisfaction estimate, and several citations whose coordinates pointed to unrelated papers. It is re-anchored on two PubMed-verified RCT-only meta-analyses (Bwire 2023; Nega 2026). The evidence base is corrected from “predominantly observational (VERY LOW)” to randomised (viral-suppression certainty HIGH; retention MODERATE), and genuine model-level heterogeneity (I² up to 90% within single models) is restored in place of the original's fabricated uniform low heterogeneity. Findings: DSD preserves viral suppression (RR 1.01, 95% CI 1.00–1.02) and improves retention (RR 1.09; ~7 percentage points).

Background. Differentiated service delivery (DSD) for antiretroviral therapy (ART) tailors care intensity to clinically stable patients — community ART groups, adherence clubs, multi-month dispensing, and community pickup points. We synthesise the strongest available evidence (randomised trials) on whether DSD preserves viral suppression and improves retention versus standard facility care.

Methods. Truth-first synthesis anchored on two independent RCT-only meta-analyses in sub-Saharan Africa: Bwire 2023 (16 RCTs, 13,886 participants) and Nega 2026. Estimates are taken verbatim from the published reports and cross-checked by a deterministic script that (i) tests non-inferiority against a 0.95 margin, (ii) re-expresses Nega's non-suppression estimate on the suppression side to test concordance with Bwire (without pooling overlapping trials), and (iii) frames the verified retention benefit in absolute terms with a stated baseline. GRADE for RCT evidence.

Results. Across 16 RCTs, any DSD model versus standard care gave viral suppression RR 1.01 (95% CI 1.00–1.02, I²=0%) and retention RR 1.09 (1.08–1.11, I²=0%) (Bwire 2023). The independent Nega 2026 MA concurred: viral non-suppression RR 0.89 (0.74–1.07) — implying a suppression-side RR ≈1.01–1.02 across plausible baselines. Non-inferiority on suppression is met (lower CI ≥0.95). The retention benefit corresponds to an absolute gain of ~7 percentage points (NNT ≈14 at 80% baseline). Crucially, heterogeneity is model- and setting-dependent: single-model pools reach I²=77–84% (adherence clubs; Bwire) and I²=90% (non-retention; Nega), so the tight any-DSD pool must not be read as uniformity.

Conclusion. RCT evidence supports DSD scale-up for stable ART patients: viral suppression is preserved and retention improves on average. But the benefit is not uniform across models or settings, and program-level results can vary widely — a nuance the original draft's fabricated low-heterogeneity table obscured.

---