Optimizing differentiated treatment models for people living with HIV in urban Zimbabwe: Findings from a mixed methods study

Abstract

Introduction Zimbabwe is scaling up HIV differentiated service delivery (DSD) to improve treatment outcomes and health system efficiencies. Shifting stable patients into less-intensive DSD models is a high priority in order to accommodate the large numbers of newly-diagnosed people living with HIV (PLHIV) needing treatment and to provide healthcare workers with the time and space needed to treat people with advanced HIV disease. DSD is also seen as a way to improve service quality and enhance retention in care. National guidelines support five differentiated antiretroviral treatment models (DART) for stable HIV-positive adults, but little is known about patient preferences, a critical element needed to guide DART scale-up and ensure person-centered care. We designed a mixed-methods study to explore treatment preferences of PLHIV in urban Zimbabwe. Methods The study was conducted in Harare, and included 35 health care worker (HCW) key informant interviews (KII); 8 focus group discussions (FGD) with 54 PLHIV; a discrete choice experiment (DCE) in which 500 adult DART-eligible PLHIV selected their preferences for health facility (HF) vs. community location, individual vs. group meetings, provider cadre and attitude, clinic operation times, visit frequency, visit duration and cost to patient; and a survey with the 500 DCE participants exploring DART knowledge and preferences.

Exploring Relative Preferences for HIV Service Features Using Discrete Choice Experiments: a Synthetic Review

“…1d); in the three choice experiments which evaluated three versus five or six monthly refills, ART refill frequency ranked lower than provider attitude in all, and additionally ranked lower than opening hours, waiting time, and group or individual counseling services in two (Fig. 1e) [24].…”

“…Data collection was similarly well described, and mode of administration of the experiment (cards, paper forms, electronic tools) was reported in all but one study [21]. Five studies either used the sample size estimation based on the formula N ≥ (500 × c)/(a × t) [28]where N is the number of participants, t is the number of choice tasks, a is the number of alternative scenarios, and c is the largest number of attribute levels for any one attribute [23][24][25][26]-or followed the proposed rule of 300 or more participants and 200 per sub-group [21]. Three studies did not present sample size calculations [18][19][20], but these studies had samples of over 1000 participants which may have been sufficient to estimate effects and accommodate sub-grouping and interactions.…”

“…Reporting of analytic methods was less clear: although the respondent characteristics were presented in the studies, these were infrequently compared with characteristics of the source population [29]; the quality of responses was also infrequently assessed, with only three studies exploring internal validity through the use of dominant and repeat questions [18,21,24]. Several studies included attributes which were modeled continuously, but tests to explore whether the data was consistent with linear, log, quadratic, or other functional forms were lacking.…”

“…Searches yielded 1226 records which after screening resulted in eight included studies and 11 choice experiments [18][19][20][21][22][23][24][25]. Study settings, populations, and methods are summarized in Table 1.…”

“…Mozambique, and Zimbabwe would be willing to pay these small amounts for services [18,24]. In these settings, PLWH were also willing to accept costs of approximately $5-6 if they could have other strongly desired service features (such as nice providers) and in South Africa, PLWH had a 50% probability of selecting a service which cost approximately $5 at the time [18,20].…”