Organisierte Sitzung

Stated Preferences and Discrete Choice Experiments to improve health service delivery in low- and middle-income countries

Discrete choice experiments have become a popular study design to study the preferences of health workers and patients to improve health service delivery in low- and middle-income countries. However, limited evidence on the use of discrete experiments around population level preferences and physician preferences to improve health service delivery have been reported from low- and middle-income countries. The goal of this organized session is to bring together diverse experiences in implementing rigorous discrete choice experiments from Pakistan, India, Burkina Faso and within the global context of COVID-19 vaccine distribution. In this session, we discuss emerging data and findings from four distinct discrete choice experiments eliciting preferences from the perspective of population, physicians and policymakers. The first paper presents health insurance preferences for outpatient care using a discrete choice experiment in Pakistan and highlights the challenges and opportunities in moving towards universal health coverage. The second paper reports on the drivers of physician effort in the Indian public sector using behavioral games and discrete choice experiments to measure physician motivation and incentive preferences in India, which will inform the design of interventions to change clinician behaviors. The third paper explores the preference for index-based weather insurance among farmers in rural Burkina Faso using a discrete choice experiment and underscores the need for stakeholders and policymakers to design risk strategies that align with user preferences to improve acceptance. Lastly, the fourth paper takes a broader policy lens to examine inequality aversion and preferences for international redistribution: using the case of COVID-19 vaccine distribution. Through these four different studies adopting stated preferences and discrete choice experiments, we discuss the methodological strengths and challenges in implementing these studies in LMICs and its wider application to other health systems to improve benefit package design and health outcomes. The deliberations from this organized session will be disseminated widely with other researchers in the form of a blogpost or commentary in a peer-reviewed journal.

Vorträge

Health Insurance Preferences for Outpatient Care – A Discrete Choice Experiment in Pakistan
Alina Imping, Friedrich-Alexander-Universität Erlangen-Nürnberg
Einleitung

Introduction State-funded health insurance schemes are increasingly implemented across the Global South but scheme utilization and acceptance often remain lower than desired. Accounting for the beneficiary population’s preferences could improve this, however, empirical evidence on stated health insurance preferences of low-income populations is rather scarce. We study preferences for elements of a new outpatient department (OPD) health insurance in a low-income population in Pakistan. Methods We conducted a discrete choice experiment with 359 respondents to elicit preferences for hypothetical OPD insurance plans that contained real policy tradeoffs under consideration at scheme design. The plans comprised five attributes: providers, health conditions, and services covered, annual coverage amount, and yearly premium. Each attribute could take on three different levels. From this, we first examine which of the design elements are preferred by the future beneficiary population. Then, we explore whether inferred preferences reflect the household’s self-reported observable health risks. Finally, we test whether different population groups value insurance elements differently. We used a mixed logit model to estimate insurance preferences. Results We found that respondents preferred covering chronic over infectious diseases or pre- and postnatal care. Furthermore, they preferred to have higher-level providers covered compared to only primary care providers and valued telemedicine to be included on top. We found a positive, yet only marginally significant preference for covering medication expenditures compared to only diagnostic services and fees and a strong preference for additionally including transportation costs. We see suggestive evidence that even in a setting with low insurance literacy, choices regarding which health conditions to cover were made to maximise benefits along self-reported pre-existing health complaints and risk-factors. We did not detect substantial heterogeneity in preferences across socio-demographic strata, respondent and household health status, indicating rather homogenous preferences. Conclusion Our results provide policy-relevant insights on population preferences regarding a new OPD health insurance scheme in Pakistan. They also point to more general challenges and opportunities in moving towards Universal Health Coverage: The population’s preference for higher-level providers points out a challenge in the efforts to strengthen primary care. The strong preference to cover chronic diseases illustrates the importance of increasing protection against the rising chronic disease burden. The population’s willingness to engage in telemedicine can be an opportunity to make healthcare more accessible.

Drivers of Physician Effort in the Indian Public Sector: Using Behavioral Games and Discrete-Choice Experiments to Measure Physician Motivation and Incentive Preferences
Nikkil Sudharsanan, Technical University of Munich
Einleitung

INTRODUCTION Physicians in low- and middle-income countries (LMICs) substantially under-deliver critical preventive care. Underdelivery of preventive care is a significant principal-agent challenge as it has strong negative consequences for patient welfare and a public service delivery challenge as it reduces the effectiveness of government spending on health. Understanding physician decision-making and identifying approaches for aligning physician behavior with patient welfare is an important scientific and public policy challenge. In this study, we will: i) measure drivers of physician behavior and their preferences for different types of financial and non-financial incentives and ii) understand if underlying motivators of physician behavior influence their preference for incentives. METHODS Among a sample of 300 primary care physicians practicing in the public sector in India, we will use behavioral games and discrete-choice experiments to measure preferences across two dimensions. First, we will measure physician motivation by studying how physicians trade off profit maximization (extrinsic motivation) against altruism (intrinsic motivation) through modified dictator games. Second, we will estimate physicians' preferences for different incentive and payment models through discrete-choice experiments that vary levels of financial and non-financial incentives and whether incentives are driven by quality or volume. We will then examine the relationship between physicians’ preferences regarding profit vs. altruism and their preferred incentives for performance. RESULTS We anticipate the results of this experiment to provide new insights into the drivers of physician motivation in LMIC contexts. The results of the modified dictator games will provide estimates of physicians' preferences for altruism and profit and reveal whether these preferences vary across observable physician characteristics. Our discrete-choice experiments will reveal which incentive elements physicians prefer and the marginal rate of substitution between different incentive designs. Lastly, our results will reveal the extent to which clinicians with different underlying preferences for profit vs altruism also prefer different forms of incentives. CONCLUSION This study will be one of the first to measure what drives clinician effort in an LMIC context. Additionally, while most literature on changing physician behavior in LMICs focuses on direct financial incentives and other performance-based financing models, our study will reveal physician preference for non-financial incentives (such as public recognition and feedback), which may be more feasible to implement and scale in LMIC contexts. Our results will also inform the design of interventions to change clinician behavior by revealing which incentive structures are most preferred by physicians and are best aligned with their underlying preferences.

Exploring the preference for index-based weather insurance among farmers in rural Burkina Faso using a discrete choice experiment
Edmund Yeboah, Heidelberg Institute of Global Health (HIGH), Medical Faculty and University Hospital, Heidelberg University, Heidelberg, Germany
Einleitung

Introduction Extreme weather events affect food security and health in vulnerable settings where livelihood is closely tied to subsistence agriculture. Limited adaptation capacity results in climate shocks eroding household capital and perpetuating poverty. Conventional crop insurance fits poorly in LMIC due to the high operational costs. Index-based weather insurance (IBWI) is a low-cost option that conditions payment on an objective weather index rather than upon verifying crop losses. The uptake of IBWI in LMICs has been poor due to a limited understanding of participants' preferences. Thus, this study investigates the preferences for IBWI features that drive acceptance and uptake of IBWI in rural Burkina Faso. Methods We used a discrete-choice experiment (DCE) to identify preferred IBWI attributes and attribute levels (options for each attribute). We preceded the DCE with an extensive literature review and a stakeholder interview to identify relevant attributes and levels for designing the experiment. We found four relevant attributes (and related levels): premium (5,7.5,10,12.5 and 15% of expected harvest), payout (50 (reference), 75, 00% of expected harvest), rainfall risk (drought (reference), flood, both), and distribution channel (government agencies (reference), financial institutions, farmer-based NGOs). We generated a DCE with Twelve choice scenarios divided into three blocks administered to 200 participants. We estimated a conditional logit, mixed logit and a generalized multinomial model (GMNL). However, we will report the findings of the GMNL, which accounts for preference and scale heterogeneity. Results According to the GMNL, there was a higher preference for insuring flood and drought (β=3.45, p=0.004) followed by flood (β=1.541, p=0.010) compared to drought. Respondents showed a higher preference for a 75% payout (β=0.596, p=0.031). For the distribution channel, respondents preferred financial institutions(β=0.648, p=0.008). The premium amount, total payout, and NGOs as distribution channels did not significantly influence respondents's choice behaviour. We found no evidence of preference heterogeneity (non-statistically significant standard deviations) but evidence of scale heterogeneity (tau=0.5581, p=0.000). Conclusion Participants significantly preferred higher payouts (but not the highest), financial institutions as distribution channels, and an insurance contract that combined flood and drought risk. Our results highlight the need for stakeholders and policymakers to design risk strategies that align with users' preferences to improve acceptance and uptake.

Inequality aversion and preferences for international redistribution: The case of the COVID-19 vaccine distribution
Henrike Sternberg, Technical University of Munich
Einleitung

Introduction This article aims to investigate how individuals’ attitudes towards international (re-)distribution differ with their degree of inequality aversion. We investigate this question in the heretofore unexamined context of the global distribution of vaccine doses during the COVID-19 pandemic. Previous research has found citizens’ other-regarding preferences and particularly their degree of inequality aversion to be positively correlated with their support for different types of within-country redistributive policies. Few studies have examined this in the context of an international distribution problem and none in the case of the COVID-19 pandemic, with the object of distribution being scarce vaccine doses during a global health crisis. Methods We conducted an online survey with 2,400 German respondents in April 2021, a time where vaccines were scarce and incidence levels high. The survey included incentivized modified dictator and ultimatum games to elicit Fehr & Schmidt (1999) parameters of advantageous and disadvantageous inequality aversion. Respondents’ attitudes towards the global distribution of COVID-19 vaccines were measured through a discrete choice experiment, in which respondents had to select one out of two hypothetical candidates to receive the vaccine who varied in terms of (i) Age, (ii) Risk of COVID-19 death, (iii) Employment status, and (iv) Country of residence. Results We found that respondents’ inequality aversion matters significantly for their global COVID-19 vaccine distribution preferences. While advantageous inequality aversion (guilt parameter) significantly increased German citizens’ support for distributing vaccines to LICs, disadvantageous inequality aversion (envy parameter) reduced it. The results were robust to controlling for various behavioral, political and socioeconomic alternative drivers. Conclusion These findings improve our understanding of (the drivers behind) citizens’ international distribution preferences during a global health crisis and provide valuable information to policy makers who may be more inclined to discard vaccine nationalism if there is public support for donating vaccine doses. Moreover, the highly statistically significant relevance of both types of inequality aversion - in opposite directions -, represents an important piece of evidence for research in the field of other-regarding preferences and underlines the relevance of this distinction when assessing individuals’ aversion to unequal distributions.