Equity in health insurance schemes enrollment in low and middle-income countries: A systematic review and meta-analysis

Improving equity in service utilization and ensuring financial protection for all individuals regardless of their ability to pay are key objectives within the global Universal health coverage (UHC) goals. Universal health coverage is of critical societal importance both in high- and in low-income settings, where inequalities between the rich and poor seem particularly large [1, 2]. Health insurance schemes are currently receiving increased attention globally not only as a health financing mechanism but also as a strategy to achieve universal health coverage [3] and as a means to reduce inequities between population groups [4].

In the absence of clear international guidelines, many low-and middle-income countries (LMICs) have started implementing a mix of social, national and community-based/mutual health insurance schemes over the past 15 years. Traditional social health insurance, which originated in Europe, uses earmarked payroll taxes from the formal sector as part of its health financing arrangements. Despite the generally small size of the formal sector, this type of health financing scheme has been adapted in many low-income settings, particularly in sub-Saharan Africa [5]. For example, in 2018, Zambia, which has an informal sector of almost 90%, passed the National Health Insurance bill which uses payroll taxes to improve access to quality health care for all its citizens [6, 7]. To extend health insurance coverage for those self-employed and the informal sector, community-based health insurance (CBHI) or mutual health insurance (MHI) have also emerged at various scales in Rwanda, Nepal, India, Burkina Faso, Cameroon, Mali and Senegal. CBHIs and MHI are typically voluntary schemes which target the informal sector and self-employed and their funds are pooled at the community level. Some countries such as Vietnam, Mexico and Peru have established noncontributory schemes using general tax revenues aimed at those not covered by social security schemes [8]. For example, in Thailand, there are three main health financing arrangements - a social security scheme for private formal sectors, a civil servants’ medical benefit scheme for civil servants and their families and a UHC scheme for those not affiliated with the other two schemes [9].

Current evidence of the impact of health insurance schemes in LMICs suggests some positive effects of insurance rollout on UHC goals [10‐14]. Two recent reviews suggest that the reduction of financial barriers through CBHI and social health insurance improve service utilization and can protect its members from out-of-pocket expenditure [14, 15]. While these average impacts of health insurance schemes are certain, the extent to which these programs succeed in improving health and wellbeing of the most underserved population groups remains unclear [16, 17]. Knowledge and awareness of insurance programs, distance to health facilities, and payments associated with insurance schemes have been shown to be critical predictors of health insurance enrollment [18]. Meanwhile, these predicators might also undermine access of the most underserved groups to health insurance schemes.

In this manuscript, we systematically reviewed the literature on health insurance enrollment in LMICs to assess the extent to which current health insurance schemes reach poor and underserved populations.

We conducted a systematic review with the aim of assessing the extent to which health insurance schemes are currently reaching the most vulnerable population groups in LMICs. We found 48 studies, which focused on 29 health insurance schemes from 17 LMICs allowing us to compare enrollment across socioeconomic groups, most of which were published after 2013. Overall, the results of our review are clear: current health insurance schemes reach only a relatively small proportion of the most vulnerable population groups.

The only scheme in which the enrolment rate was far lower for the wealthiest populations was the Colombian subsidized regime which exclusively targeted the poor and other vulnerable groups as those in the formal sector and self-employed workers with a steady income are required to obtain the contributory regime. Two features of the scheme which seem important are, first, that the scheme is mandatory for all those who are eligible to enroll. Second, during the period of analysis by Ruiz-Gomez et al, municipalities were using a mean proxy test to select beneficiaries into the scheme through the established social service beneficiaries’ identification system (Sistema de Identificación de los Beneficiarios de los Servicios Sociales, SISBEN )[63].

Even though virtually all the other insurance schemes analyzed directly target or subsidize the most vulnerable groups, better-off households have on average almost twice the odds of enrolling in health insurance compared to poorest households. For example, the Ghana health insurance scheme stipulates premium exemptions for indigents, the elderly above 70, pregnant women and children while the Rwanda Mutelle de Santé exempts the poorest 16% of households from premium payments. Other schemes such as those implemented in Nepal, Bangladesh and Burkina Faso have subsidized rates for the poorest households.

Despite these efforts, enrollment rates of the wealthiest subpopulations are higher than those of the most vulnerable population groups in all of these countries. These results are consistent with previous work on health insurance programs showing that enrollment and willingness to purchase health insurance in LMICs is pro-rich, which are explained by factors such as greater exposure of the rich to the media and their higher income levels to pay for health insurance premiums [14, 51, 64].

The current enrollment gaps should not necessarily be interpreted as evidence that current targeting efforts do not make enrollment easier for poor households. Rather, it demonstrates that these current measures appear insufficient to equitably include vulnerable populations in health insurance schemes. Given that most health insurance schemes in LMICs are heavily financed by central government revenues, the currently observed enrollment patterns essentially make health insurance a regressive policy, primarily subsidizing health care for better-off households. Further reductions in premiums and improving geographical access to health facilities could potentially increase uptake among poor and underserved populations [12]; other policy options include automatic (free) insurance enrollment of these groups or the direct provision of free health services for these groups.

Despite our best effort to review all of the recent evidence available, the findings presented in this manuscript have limitations. First, the included studies were retrospective and cross-sectional, and primarily focused on CBHI and national health insurance schemes. Second, due to the language restriction for publications in English, there was a limitation by the exclusion of articles published in other languages. In the past two decades, many LMICs in Latin America have implemented health insurance schemes such as non-contributory schemes for vulnerable populations [65]. Therefore, restricting the literature search to English may have underrepresented the inclusion of studies from this region which in turn may have underestimated health insurance enrollment of vulnerable populations. Thirdly, it was also quite striking that nearly half (23/48) of all studies identified focused on Ghana, while no studies were found on several other countries where similar insurance programs have been launched in the recent past. In addition, studies used highly heterogeneous ways of measuring wealth or income that may not be directly comparable. Our analysis also pooled data across different designs of insurance schemes and socioeconomic group definitions and therefore, represents an average across highly heterogeneous systems. In addition, nearly all the studies relied on self-reported data about wealth or income and educational status which could lead to misclassification due to recall bias. Lastly, another limitation of our study is the lack of longitudinal data that would have allowed evaluating whether there are countries that are successfully reducing inequalities in health insurance enrollment. Large longitudinal trend studies are needed to determine the contribution of health insurance schemes in reducing inequalities between the rich and poor over time

Despite these limitations, our findings are consistent with a larger analysis of the World Health Surveys conducted in the early 2000s which suggested that health insurance schemes continue to primarily benefit the better-off populations [66]. In their current form, health insurance schemes are thus unlikely to be viable mechanisms to promote universal health coverage. Challenges faced by current schemes include difficulties associated with identifying the most poor or vulnerable populations [67‐70] as well as management of rollout and implementation at sub-national levels [71]. Increased financial, political, and institutional resources are likely needed to identify and reach underserved populations. In addition, simplified administrative processes for enrollment such as automatic enrollment after their identification could also facilitate the inclusion of underserved populations [72‐74].

Although all recently introduced health insurance schemes LMICs aim at providing access to health services as well as financial protection to the most vulnerable populations, current coverage is low among the poor, and highly regressive in most countries. Experiences from countries suggest that current strategies to improve coverage of vulnerable populations in health insurance schemes have not achieved their aim of equity. Further investigation is needed to understand why these strategies are not reaching vulnerable groups. The evidence also suggests countries that are planning to establish health insurance schemes with the aim of equity for vulnerable populations might need to reevaluate their approach given the findings of this review.