Answers are grounded in this paper and cite their sources.
A discrete choice experiment (DCE) is a stated-preference method used to examine how individuals make trade-offs among policies, service features, or interventions. DCE have been widely used in high-income countries and increasingly in low- and middle-income countries to assess preferences across healthcare, transport, agriculture, and marketing. In health, DCEs have commonly assessed preferences for services, interventions and jobs, but have been applied less often to health-financing arrangements, including provider payment mechanisms. We describe the process of developing a DCE to elicit healthcare providers’ preferences for the design of capitation under the proposed Universal Health Insurance (UHI) scheme in Tanzania. We used a four-step mixed methods approach to systematically develop attributes and levels for the DCE: (i) a literature review to map and characterize existing provider payment methods in Tanzania and identify key attributes for each payment method; (ii) 35 qualitative encounters; 21 individual in-depth interviews and 14 group interviews, each involving two to four participants, who were healthcare workers, managers and purchasers to understand their experiences with current payment methods across different levels of care; (iii) internal expert review and consensus; and (iv) experimental design and pilot testing. The draft DCE was piloted with 28 healthcare workers from the selected primary healthcare facilities. Pilot data were analysed using conditional- and mixed-logit models to assess the choice tasks and generate priors for the final DCE design. Step 1 and 2 identified ten candidate capitation attributes with varied levels. Internal expert review retained five feasible, measurable and context-relevant attributes: adequacy of payment amount, services to be covered, payment schedule, additional payment for high performing facilities and timeliness of payment. One attribute (payment rate) overlapped with adequacy of payment amount and was merged. Four attributes: predictability of payment amount, bundling of services, autonomy in the use of funds, and accountability mechanisms were excluded because of conceptual overlap, limited familiarity, multidimensional interpretation, and difficult in operationalization. In the pilot, participants understood the attributes and levels, completed all choice sets, and made observable trade-offs. Relative to their reference levels, conditional-logit estimates were positive for inadequate payment with co-payment (β = 0.78; p < 0.01), full-service coverage (β = 0.72; p < 0.05), monthly payment (β = 0.72; p < 0.01), and timely payment (β = 0.85; p < 0.01). Adequate payment unexpectedly had a negative coefficient (β=−0.45; p < 0.05). These estimates were used to test the instrument and inform priors, and not to draw substantive conclusions about providers’ preferences. Mixed-methods approach was essential for identifying feasible, relevant, and context-specific attributes and levels, that reflects design and implementation issues healthcare providers consider important for effective service delivery. Findings from the pilot supported the feasibility of the choice tasks and informed the design of the final DCE which will provide evidence to inform the capitation design for the proposed UHI in Tanzania.
No takes yet. Share an insight, caveat, or question.
Ngadaya et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: