Background
Differentiated service delivery (DSD) models have reduced strain on health services and improved client experience, retention and viral suppression, but little is known about the impact of HIV DSD models on quality of life (QoL), which is essential for optimizing person-centered care. This study assessed the impact of DSD models on QoL, loss to follow-up (LTFU), and mortality among persons living with HIV (PLHIV) on ART over time at a large urban HIV clinic in Uganda.
Methods
We retrospectively reviewed records of all the 1,000 PLHIV enrolled in a conveniently sampled 10-year cohort at the PEPFAR-funded Infectious Diseases Institute (IDI) clinic in Kampala, Uganda. Participants were on ART for 10 years at enrollment in 2014 or 2015. We scored QoL with an adapted Medical Outcomes Study (MOS-HIV) tool. We determined and compared QoL scores (summed scores from 10 questions expressed as a percentage), sustained annual viral suppression (<200 copies/mL), all-cause mortality and LTFU (missing clinic visits for ≥3 months) for PLHIV in any of three DSD models for ≥6 consecutive months—fast-track drug refill (FTDR), facility-based groups (FBG), and composite model combining these two—versus facility-based individual management (FBIM), the standard of care (SOC). Using inverse probability treatment weighting mixed effects models, we estimated QoL scores as weighted mean ratios and checked robustness of results using G-computation analysis. Sustained viral suppression was compared using odds ratios; all-cause mortality and LTFU were compared using hazard ratios from the Cox proportional hazard regression model.
Results
Of 1,000 PLHIV, 980 had ≥1 follow-up and were included in the analysis. Median age was 45 years (IQR: 40-51), 62% were female, and 95% had a suppressed viral load at baseline. Baseline QoL was 90.1% in any DSD vs 89.2% in SOC. After eight years of follow-up, weighted mean QoL was higher in participants enrolled in DSD models than the SOC (90.4% vs 89.1%; weighted mean ratio 3.66, 95% CI 2.10-6.37, p-value <0.001); there were no statistical differences across DSD models. Participants in DSD models were more likely to have sustained viral suppression (weighted odds ratio 1.69, 95% CI 1.24-2.31), lower mortality (weighted hazard ratio 0.08, 95% CI 0.03-0.20) and lower LTFU rates (weighted hazard ratio 0.08, 95% CI 0.02-0.31).
Conclusions
These findings support the broader adoption of DSD models in HIV care programs to enhance the QoL, well-being, and clinical outcomes of PLHIV.