Background
Universal health coverage (UHC) is a key concern in global health policy, especially in low- and middle-income countries (LMIC) [
1,
2]. UHC is grounded in the sustainable development goals (SDGs), which aim to ensure healthy lives and promote well-being for all at all ages by 2030, including financial risk protection and equal access to quality essential health-care services [
3,
4]. The World Health Organization (WHO) and other international actors consider the implementation and expansion of health insurance central to achieving UHC. In LMICs, different types of insurance have been implemented, which vary in their scale, providers (government vs private sector), and, often, types of beneficiaries [
5]. Across most scheme types, existing evidence underscores the importance of health insurance as a tool to enhance UHC [
6‐
11]. For example, in a systematic review on the impact of health insurance in Asia and Africa, enrolment had a positive impact on reducing out-of-pocket spending, while also increasing utilization of health services [
7].
However, while evidence suggests health insurance schemes can improve health care utilization and financial protection for its members, they can also risk compromising equity by excluding high-risk and/or vulnerable individuals in society [
12]. For example, people living in poverty may not be covered in health insurance if they cannot afford contributions or are not exempted to pay, leading to inequity in enrollment among the most vulnerable in society [
13]. Similarly, certain groups, such as older adults, people with chronic illness and people with disabilities are less likely to participate in social protection programs or may have health service needs that are not covered in standard benefit packages [
14‐
19]. While studies on social inclusion and exclusion in LMIC and health are available, there are hardly any systematic studies of how social exclusion may affect access to equitable health financing arrangements [
20]. Conversely, studies on health insurance schemes are available, however those do not evaluate social inclusion of specific vulnerable groups as such [
7,
21]. In most cases, neither schemes nor governments have rigorously analyzed and aligned enrolment patterns, needed services and benefit packages based on needs of their population [
22‐
24]. If health insurance schemes are truly to be used as a tool to UHC and thus “leave no one behind”, there is a need to evaluate insurance enrolment and the impacts of participation amongst groups that are most vulnerable to exclusion [
6,
25].
The WHO Social Exclusion Knowledge Network (SEKN) developed a Social, Political, Economic and Cultural (SPEC) conceptual model, explaining social exclusion not as a ´state´ but as a process, operating along several dimensions and at different levels from the individual to regional and global levels ‘state’ [
26]. These exclusionary processes create a continuum of inclusion/exclusion characterized by an unjust distribution of resources and unequal access and return to the capabilities and rights required to enable participatory and cohesive social systems [
26]. Moreover, having a particular disadvantage does not indicate that an individual is socially excluded. Rather, it indicates that the individual is vulnerable to social exclusion [
17]. Due to inconsistencies in how social inclusion and exclusion are defined and measured, there are no single sets of indicators which for assessing social exclusion as a process [
26]; therefore, we searched for a model to identify social determinants for inclusive health systems that provides an approach towards individuals who might be vulnerable to social exclusion. The EquiFrame offers a social determinants approach to assess the extent to which a given policy is consistent with promoting social inclusion, service coverage and reduce barriers to access – all key components of inclusive health systems [
27,
28]. In accordance with the WHO, the EquiFrame has given priority to 12 vulnerable groups [
29,
30]. This review selected eight priority vulnerable groups that seem most severely affected by exclusionary processes and for whom strategies to expand or improve health insurance plans have been described in general terms only [
31]. Those eight groups are female-headed households, children with special needs, older adults, ethnic minorities, displaced populations, chronically ill and individuals with disabilities. The remaining four groups (those with limited resources, increased relative risk for morbidity, mother-child mortality and those living away from services) were based on parameters that have been frequently used in health insurance evaluations (such as income level, urbanization level), or on maternal health and specific disease programs for illnesses with increase relative risk, which have shown to be covered by health insurance initiatives in previous studies [
7,
10,
11,
25,
32‐
38].
Our systematic review evaluates three types of health insurances, on various outcome indicators that evaluate the impact of the particular health insurance scheme. Various types of health insurance are available in LMIC and have different impact on the population they serve. Social health insurance (SHI) are schemes based on mandatory enrolment, often scaled to the national level and provided by governments; SHI involves the formal sector, with payroll taxes for mobilizing funds and pooling risks. Private health insurance (PHI) and community-based health insurance (CBHI), including micro health insurance, involve voluntary enrolment [
5,
7]. CBHI differs from PHI by targeting specific population groups, including vulnerable low-income groups [
38]. Following the framework of Preker and Carrin [
39] and the SPEC model [
40], outcome indicators of health insurance schemes included in this review are enrollment rates in schemes and its impact on health care utilization, healthcare expenditures and financial protection, health outcomes and quality of care. As each outcome indicator represents a step followed by the enrollee through the scheme, social exclusion can occur at each of those outcome indicators. Therefore, we assess the way vulnerable groups behave with regard to each reported outcome at group level, in comparison with other (vulnerable) groups or the general population.
To the best of our knowledge, a robust and up to date overview of empirical evidence substantiating and assessing enrollment in and impact of health insurance schemes on health equity among the selected vulnerable groups is lacking [
7,
24,
41]. This systematic review aims to address this gap by assessing health insurance enrolment patterns and the impact of health insurance (health care utilization, financial protection, health outcomes and quality of care) in LMIC for the most vulnerable groups – namely, female-headed households, children with special needs, older adults, youth, ethnic minorities, migrants, and those with a disability or chronic illness.
Discussion
This paper provides a comprehensive review of studies that have assessed enrolment and impact of health insurance for specific vulnerable groups. The most notable finding is the dearth of high-quality evidence which evaluates access to and impact of health insurance schemes for vulnerable groups in LMIC, such as individuals with disabilities, female-headed households, displaced populations, ethnic minorities, displaced populations, youth and children with special needs. Despite all attention paid to UHC, this calls for more attention to the issue of equity for specific vulnerable populations in health insurance, in line with the recommendations of the WHO [
92]. The existing evidence gathered from this systematic review and its policy implications are discussed below.
The reported vulnerable groups
This literature review revealed that available knowledge about health insurance inclusiveness is sparse for each of the vulnerable groups and of variable quality. Out of the eight included vulnerable subpopulations, the chronically ill were the most commonly reported, followed by older adults and individuals with disabilities. These subpopulations might have a clear set of ‘conditions’ or ‘non-communicable diseases’ which are relatively easy to define and therefore, easy to tackle within health insurance impact evaluations. This health-related ‘state’ of vulnerability might not be the case for the other, less addressed groups, such as gender, ethnic minority or citizenship (for displaced populations), which are especially influenced by political or cultural dimensions. One exception here is that children with special needs are unreported as subgroup based on our review findings, probably because children are often included in household enrolment evaluations, where the insurance status of the head of the household is taken as an assessment of enrollment. Other reviews on the impact of health insurance schemes for children also did not report findings on children with special needs due to disabilities [
33,
37].
Health insurance indicators addressed: enrollment and impact
Henceforth, we discuss our results by the SPEC-by-step tool, based on the SPEC conceptual model [
40]. The first step, “awareness”, ideally targets all people to become aware of health insurance schemes. This systematic review aimed to assess actual enrolment in health insurance, rather than levels of awareness. However, vulnerable groups are more likely to face social exclusion/discrimination and less likely to be included in initiatives that may promote awareness of health insurance (e.g. in identification process, risk pooling, awareness raising activities) [
10,
77,
93]. Though we did not specifically assess this step, we believe that with regard to inclusionary processes, equity starts by being reached and empowered in order to decide about the involvement in health insurance schemes.
The next step is “enrolment”, defined as registration to a health insurance scheme. For most of the eight vulnerable subpopulations in this review, enrolment numbers are scarcely reported. For the chronically ill, and to a lesser extent for older adults, there is some limited evidence that they are more likely to enroll than the general population [
44,
94]. This phenomenon of adverse selection can be defined as strategic behavior by the more informed people in a contract against the interest of the less informed people such as those in the informal sector [
95]. This review shows that older adults and chronically ill, those who seem mostly in need of frequent health care services, tend to enroll more. Also, it is possible that this review included studies where enrolled individuals in households had the worst health status and non-enrolled individuals in household had the best health status, meaning inequity in enrolment among vulnerable groups within households still exists [
96,
97]. Another possibility is that adverse selection manifests itself through healthy people choosing managed, well-organized care and less healthy people choosing more generous plans, and adverse selection is therefore more likely to happen in voluntary insurance schemes [
98]
. Ways to tackle adverse selection, and therefore ensure better inclusion of other groups, are mandatory scheme enrolment, enrolment at the household level or introduction of a waiting period [
7,
36]. Notwithstanding a possible positive effect on enrolment found in this review through adverse selection, it was reported in previous reviews on individuals with disabilities that access to social protection programs still appears to be far below need [
18,
20]. Possibly, enrolment remains low due to high premium rates, hidden costs to enroll, e.g. travelling with assistance, and intensive information interventions or reforms using voluntary contributory mechanisms have no effect on enrolment of the less affluent [
13,
99,
100]. Despite the aim of risk pooling [
11], the health care delivery system should be equitable and thus favor vulnerable population groups in order to increase their utilization of health care services and reduce as much as possible their exclusion to affordable health care [
101].
Being enrolled in a health insurance scheme with a valid membership card should in principle ensure access to the health care benefit package for them. The positive results for the “accessing care” step in this review concur with previous reviews that for the reported groups, health insurance membership can increase utilization of health care services [
7,
95].
In terms of the “benefits” step, due to inconsistency in findings, we cannot draw clear conclusions on the impact of health insurance on financial risk protection (e.g. use of savings, CHE), health outcomes (e.g. access to medicines) and quality of care for the eight vulnerable subpopulations. Further, it remains unclear who gets what services and in what proportion health services and related costs are really covered and reimbursed [
102].
On financial protection, for instance, our review suggests that for chronically ill, older adults and individuals with disabilities, formal insurance schemes do not guarantee protection from CHE, despite greater needs for health services [
50,
81]. Our findings extend those of others, confirming that households with older adults, young children or chronically ill borrow money or sell assets to finance illness, underlining the bilateral link between health and poverty [
43,
103]. Looking at a wider spectrum of health financing arrangements, a Nigerian study found that households with older adults participating in informal health financing arrangements (other than health insurance schemes) were less likely to incur CHE than those with formal schemes [
104]. Unsurprisingly, there is a discrepancy between what is formally covered and what is in reality paid for, and therefore, it is likely to underestimate the true extent of economic poverty among vulnerable groups [
43]. Despite the fact that some of the results do show that health insurance, to a certain extent, serves as a mechanism to protect from CHE, the impact of health insurance on poverty remains insufficiently clear for vulnerable subpopulations [
7,
38,
95,
105]. Hence, the notion of equity in utilization and financing of care may not be enough to judge whether a health system protects the income of vulnerable groups against expensive health care use [
106]. What is often overlooked is that social protection programs such as health insurance not only exist for poverty reduction but also for poverty prevention, by helping to prevent people to move into the group of the extreme poor [
107].
As we turn to the delivery of care, limited reports are available on the received quality of care by the included vulnerable groups, whilst higher quality of care could enhance member renewal decisions [
44]. A way to assess quality of care could be through vertical equity, using indicators of access and use of health care according to needs [
108]. Lastly, due to methodological challenges, there are inconclusive results on the impact of health insurance on health status. Evidence in our review is insufficient to understand the complex causal chain behind the impacts of UHC programs on health status [
11].
The complexity of health insurance research
Generally, social exclusion is preferably viewed as a process rather than a ‘state’, in the latter case risking to neglect the relational nature of these ‘states’ and the exclusionary processes generating them [
26]. Health insurance evaluations often use parameters such as income and household size, rather than parameters that refer to societal risk factors to exclusion. In our review, we found this measurement issue particularly in the low-quality studies, using unclear or inconsistent measures to define the vulnerable group, or in the vulnerable groups that were based on socio-demographic characteristics only (such as older adults, youth, female-headed households), in which no measurements are taken that include the socio-cultural context of the person ‘at risk’ of vulnerability. Therefore, the findings of the review should be interpreted with caution since we recognize that exclusionary or inclusionary processes will impact in different ways to differing degrees on different groups and/or societies at different times.
Few studies in this review qualified as high-quality impact evaluations based on the study design, risk of bias or insufficient information provided. No RCTs or interrupted timeline series studies were identified. High-quality impact evaluations appear difficult to apply to health systems, both for economic reasons (costly and labor intensive) and for ethical and political reasons. Specifically defining what is meant by ‘the poor’ (e.g. the very poor, indigent and vulnerable) and who qualifies to be categorized as such is challenging, costly and politically sensitive [
77,
109,
110]. Furthermore, there are ethical issues around withholding of services in order to create control groups [
110]. In conclusion, our review shows that there is a need to develop more specific social determinants and equity indicators for specifically defined vulnerable population groups to use for health insurance scheme impact evaluations. We also suggest the usefulness of including the social, political, economic and cultural dimensions for appropriately measuring the population coverage, possibly by using tools such as the SPEC-by-step tool [
40].
Health insurance impact on equity and universal health coverage
Returning to the question of UHC, this can be promoted through actions to improve efficiency, equity in the distribution of resources as well as transparency and accountability [
111]. Equity needs to be differentiated from distribution. Financial reforms that improve equity in the distribution of resources can also lead to improvements in equity in the use of services. However, equal access may not be sufficient to improve the situation of vulnerable individuals. The overall aim of UHC is to match and optimize the distribution of resources to the relative health service needs of different individuals and groups in the population [
112]. Therefore, on one hand, vulnerable subpopulations may need additional assets to participate in general (redistribution, advancement to at-risk groups) and if this happens, it is important to understand how this ´targeting´ has been realized. On the other hand, for the sustainability of the system, too much favor can be detrimental in economic, social and political terms, so a right balance will have to be found. A new health insurance scheme is either designed for the purpose of making its members better off in terms of health or designed and intended to serve as an agent of change to improve equity in the use of quality health services and its financial protection for the entire population, positioned in a broader context of “leaving no one behind” [
111]. For both cases, and for voluntary as well as for mandatory schemes, good coverage for some people comes at the expense of the rest of the population. Therefore, the interests of the schemes can be in conflict with UHC objectives at the level of the entire system.
Because health systems aim to promote universal protection against financial risk, health insurance schemes in low- and middle-income countries undertake various initiatives to reach the vulnerable members of the populations such as discount cards, conditional cash transfer (CCT) programs, exemption fees or free enrolment for vulnerable populations [
7,
46]. Exemption fees, such as one study in this review reported in Ghana [
76] and cash transfers for vulnerable children and families seem ways to promote universal protection and health coverage by social protection programs [
113,
114]. Thus, to do so in social protection programs, how do we get everyone around the table? Study findings show that the limited effectiveness of other programs for cooperation is primarily linked to political factors -such as power relations, interests and incentives of the various actors [
115]. In other words, power analyses should be central to inclusive development research and social protection programs such as health insurance [
116].
Reflecting on the results of our study and the universality of those results, we compare in brief how social inclusion and equity towards health insurance has been captured in high-income countries. One review evaluating equity within UHC in high-income as well as LMICs found that studies from high-income countries tended to focus on access to specialized services or for specific diseases, and those studies focused on distinct populations such as children, elderly or psychiatric patients, rather than the population as a whole [
117]. In our study we found a few studies focused on distinct populations or specific diseases, however most studies focus on the impact of health insurance for the whole population. This could possibly be the case due to the relative new nature of health insurance schemes in LMIC and the limited coverage of services, often emergency care and basic in- and outpatient services versus specialized care in high-income countries. Nevertheless, this same study revealed findings in the research describing inequities in receipt of specialized health care in Canada and Australia in spite of insurance systems [
117]. Studies targeting one or more specific vulnerable groups with regard to health insurance describe comparable results to our review also. For example, among chronically ill, two systematic reviews showed that being covered by health insurance improved outcomes on health care utilization and health outcomes [
118,
119]. Two studies covering several European countries, found that vulnerable populations, such as poor citizens, elderly citizens or elderly with chronic conditions, had (catastrophic) medical expenses and thus health insurance did not provide adequate financial protection [
120,
121]. Furthermore, a review of systematic reviews analyzed to what extent hosting advanced countries provide equal access to health insurance for migrants. As in our systematic review, this study also showed the lack of empirical evidence on enrolment of migrants in health insurance and the need for strategies such as information and application support to expand health insurance coverage in vulnerable populations [
122].
On the measurement issue of social inclusion in health systems and health insurance, a study published by WHO on financial protection in Europe, found that the association between gaps in population coverage and financial hardship is weak because people lacking coverage usually only account for a small share of the population, and European countries generally provide all residents with access to emergency services, which is often not the case in LMIC [
123]. However, the incidence of catastrophic health spending and financial protection still varies hugely among households in Europe, especially among countries that joined the EU after 30 April 2004 [
123,
124]. Similar strategies, such as exemptions for poor people and regular users of health services – e.g. people with chronic conditions, are used in those countries to protect against financial hardship (WHO Europe) [
123].
In conclusion, we observe similar achievements and challenges towards equity of health insurance schemes and its impact for particular vulnerable groups between LMIC and high-income countries, however the level of financial protection and access to emergency care of or non-specialized care might be covered more sufficiently in high-income countries.
Strengths and limitations
To our knowledge, this is the first systematic review of social inclusion of the eight defined vulnerable groups regarding the access and impact of health insurance. A strength of this study was the extensive search in 11 databases covering literature from 1995. As our harvest of papers indicate, the studies on health insurance impact evaluations is on the rise, reflecting a rise in health insurance schemes more recently [
7,
125]. With the design of this systematic review, we could not incorporate changes per country per scheme. Also, this review did not integrate the specificity of each country and therefore, no conclusions can be drawn on country-specific levels of equity and its implications for the particular scheme and related policies. Nevertheless, by focusing on specific vulnerable groups related to several types of health insurance instead of a broad categorization of parameters, we were able to collect a relevant data set that could inform policy and research in a way that it provides a picture of the current extent of evidence in the literature regarding inclusion of health insurance schemes for the targeted vulnerable groups. However, since we chose not to include studies about all other possible social determinants or individuals at-risk of vulnerability, such as sex workers, individuals who are homeless or living in institutions, the data about inclusionary processes are incomplete. Also, the definition of the included vulnerable groups was not consistent in each study, for example varying age minimums for older adults, or defining the presence of a chronic illness. Furthermore, we choose a rather generic definition of chronic illness in the search strategy, not defining each chronic condition or specific disease programs, in order to prevent an unmanageable amount of hits that seemed not related to health insurance schemes. As a result, only one study on the impact of health insurance on HIV/AIDS management is included.
Another strength was the comprehensive approach to evaluate health insurance impact allowing a wide variety of study designs and outcome indicators ranging from the inclusiveness of enrolment to the quality of delivered care and degree of financial protection once being insured. However, its assessment of a positive effect, no effect or a negative effect was not always straightforward because control groups were often not clearly described.
Lastly, only English written articles and peer-reviewed articles were included while in many LMIC other official languages or formats such as reports might be used to publish results. The conclusions of this review may be considered general or preliminary by nature, due to the diversity of the included studies and the low levels of evidence found. Nevertheless, in the absence of more precise research findings for most of the vulnerable groups considered here, our results provide a good first indication of the extent to which inclusionary processes towards equity in health insurance schemes in LMIC is represented in the literature.
Policy implications and future research
The review findings point to a major gap in knowledge regarding the inclusiveness of health insurance schemes and its impact on health outcomes, quality of care and financial protection. What is the magnitude of the gap and what additional measures can bridge this gap? Regarding the first part of this question, it is clear that more data and disaggregated analysis is needed. Social protection programs, including health insurance schemes, can help build human capacity and enhance the stability of economic growth [
116]. As we found, there is a need to assess the impact on the utilization of care, its quality, and the effects on invisible CHE and health outcomes for those who are prone to increased utilization of health services. Results of this review point to the need of country specific policies and impact evaluations for vulnerable groups to be used to improve the inclusion of and benefits for those groups in health systems in particular health insurance schemes. This could be assessed by evaluating the cost of health insurance and medical benefits, the level of financial protection, e.g. reimbursement rates, and whether health insurance schemes satisfy the needs of specific subpopulations. For other groups, we have seen that community-based mechanisms are highly useful [
13]. Community-based mechanisms seem to recognize the diverse needs of the vulnerable subpopulations though how it contributes to access to health care is not shown on a large scale. This is not a short-term remedy and long-term commitment of governments is required in order to lead to equal health outcomes. A strategy to achieve UHC is through a risk-adjusted equalization of budgets to health care providers or purchasing agencies; this may improve equity in the distribution of resources and services and the reduction of fragmentation in pooling to enable greater financial protection and equity in the distribution of resources and services [
111].
We then address the issue of additional measures to bridge the gap in knowledge in the inclusionary process of health insurance. Systematic health systems research needs to be strengthened by assessing how criteria for prioritizing groups that are disproportionately affected change, as well as equity impact of cost sharing, especially for the most vulnerable [
11]. Accordingly, the issue of targeting needs must be critically examined. Possible solutions to improve quality of research are to combine quantitative analysis of effect with qualitative information describing context and implementation issues and to seek for measures that assess equity and the effect of changes in policies [
95]. Future evaluations should consider mixed or qualitative approaches such as realist evaluation that seek to answer the questions of what works for whom in which circumstances in order to bring about various mechanisms in the contexts in which they are delivered [
126], such as one realist review by Robert et al. (2017) resulting in key mechanism to seek free public healthcare in sub-Saharan Africa [
127]. In this approach, both qualitative and quantitative methods can be used to trace these mechanisms.
Conclusion
This review provides an assessment of the available evidence on the impact heterogeneity of health insurance schemes in LMIC by means of a systematic review of the literature. Despite a sizeable amount of literature published on health insurance there is still a dearth of good quality evidence, assessing equity and inclusion of specific vulnerable groups. People with disabilities and individuals who belong to female-headed households, youth, children with special needs, displaced populations and ethnic minorities were minimally reported within health insurance impact evaluations. Based on the evidence available from the studies that assess equity, social exclusion is visible in both the access to health insurance schemes and in lower financial protection. From a social inclusion viewpoint, health insurance has not yet shown to serve as an optimal tool to UHC, in a way that vulnerable groups are covered, from being aware and enrolled in health insurance schemes to proven impact on financial protection and improved health outcomes once carrying a health insurance card. This review also clearly demonstrates that current literature - by using common parameters such as income - is insufficiently clear on the impact of HI on specific vulnerable groups in terms of social inclusion. We therefore propose to move beyond a focus on the overall group of the poor, to develop specific measures for those at risk of exclusion and to assess inclusionary processes. With such measures, more sounder conclusions can be drawn on the social inclusion impact of health insurance in LMICs.
Furthermore, the review shows that while some groups are underreported, there is moderate evidence that mostly SHI schemes do enroll the chronically ill, implying a positive effect on social inclusion. CBHI schemes show some studies addressing various vulnerable groups. For chronically ill, older adults and individuals with disabilities we found that once being enrolled in a scheme, utilization of health care services seems to increase. Regarding financial protection among several schemes dealt with here, the picture shows negative or insufficient effects for chronically ill, older adults and individuals with disabilities, and incomplete results for other underreported groups. Minimal evidence was found that being a member of a health insurance scheme could prevent from CHE and reimbursement rates and ‘real’ (including indirect) costs could be easily underreported. No conclusions can be drawn for the included vulnerable groups on quality of care and health outcomes.
Evidence should be strengthened within health care restructuring systems to achieve UHC by redefining the dimensions to identify vulnerability and the investigation of in- or exclusionary mechanisms that are more context specific. We recognize that for health insurance schemes, different strategies are required to include individuals who are ‘at risk’ of vulnerability and to assess exclusionary processes in order to “leave no one behind”.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.