Effect of a person-centered care intervention on return and sustained reengagement after treatment interruptions from HIV care in Zambia: A post-hoc analysis of a stepped-wedged cluster randomized trial

BACKGROUND: Poor client-provider interactions lead to reinforcing cycles of treatment interruptions (TIs) from HIV care and fear of return, undermining sustained retention. We previously demonstrated a multi-component person-centered care intervention (PCC) targeting healthcare worker (HCW) behavior improved client experience and retention in HIV care. To understand mechanisms of improved retention, we evaluated the effect of the PCC intervention on return after TIs and sustained retention after return in a post-hoc analysis, hypothesizing that friendlier HCW attitudes enable successful reengagement into care. METHODS AND FINDINGS: We implemented the PCC intervention at 24 clinics in Zambia in a stepped-wedge, cluster randomized trial from Aug 2019 to Nov 2021. Clinics were allocated into 4 groups (8 clinics in Groups 1 and 4, and 4 each in Groups 2 and 3) and randomly assigned to crossover from control to intervention every 6 months. Under control, clinics provided routine HIV care without additional support. Afterwards, clinics received the PCC intervention targeting caring aspects of HCW behavior with 1) training and coaching on PCC practices, 2) measurement and feedback of client experience (via exit interviews), and 3) small facility-level incentives. We examined individuals living with HIV who became >30 days late to a scheduled visit (i.e., TI), categorizing intervention exposure based on the last visit prior to TI. We used electronic health records to assess the effect of the PCC intervention on 1) return to HIV care after a TI, and 2) repeat TIs among those who returned. We used multistate analytic methods to estimate crude incidence of return and repeat TIs and prevalence of being in care at 1-year. We used shared frailty Cox proportional hazards and mixed-effects Poisson models to obtain formal comparisons that accounted for the stepped-wedge design. Individuals were censored at intervention crossover or database closure. During the study period, 128,910 clients (n = 69,671 control, n = 59,239 intervention) became >30 days late (64.3% female, median age 38y [IQR 31,45], median years in care 2.5y [IQR 0.6,6.7]). The PCC intervention increased incidence of return after TI at 12-months (72.3% intervention versus 67.7% control, crude risk difference [RD] +4.6% [CI 4.1,5.2], adjusted hazard ratio [aHR] 1.16 [CI 1.12,1.20]; p < 0.001). Among returners, the intervention decreased incidence of repeat TIs (44.3% versus 55.6%, RD -11.4% [CI -12.2,-10.6], aHR 0.50 [CI 0.45,0.55]; p < 0.001) and increased the proportion in care (with or without repeat TIs) (82.7% versus 73.4%, RD +9.3% [CI 8.6,10.0], adjusted risk ratio [aRR] 1.14 [CI 1.05,1.25]; p = 0.002) at 12-months after return. Overall, the PCC intervention increased the proportion in care 12-months after a TI from 51.5% to 60.1% (RD +8.7% [CI 8.0,9.3], aRR 1.19 [CI 1.05,1.35]; p = 0.008). Effect sizes were similar in direction and magnitude in multiple sensitivity analyses with varying precision. Limitations included post-hoc analysis and those associated with stepped-wedge design. CONCLUSIONS: The PCC intervention likely improved rates of return to care after TIs, sustained reengagement after return, and overall retention among individuals with TIs. Strategies targeting caring aspects of HCW behavior and client experience may help lower barriers to and address persistent challenges with reengagement in care.