Analysis of the Effect of Health Insurance on Health Care Utilization in Rwanda

In Rwanda, more than 90% of the population is insured for health care. Despite the comprehensiveness of health insurance coverage in Rwanda, some health services at partner institutions are not available, causing insured patients to pay unintended cost. We aimed to analyze the effect of health insurance on health care utilization and factors associated with the use of health care services in Rwanda. This is an analysis of secondary data from the Rwanda integrated living condition survey 2016-2017. The survey gathered data from 14580 households, and decision tree and multilevel logistic regression models were applied. Among 14580 households only (20%) used health services. Heads of households aged between [56-65] years (AOR=1.28, 95% CI:1.02-1.61), aged between [66-75] years (AOR=1.52, 95% CI: 1.193-1.947), aged over 76 years (AOR=1.48, 95% CI:1.137-1.947), households with health insurance (AOR=4.57, 95% CI: 3.97-5.27) displayed a significant increase in the use of health services. This study shows evidence of the effect of health insurance on health care utilization in Rwanda: a significant increase of 4.57 times greater adjusted odds of using health services compared to those not insured. The findings from our research will guide policymakers and provide useful insights within the Rwanda context as well as for other countries that are considering moving towards universal health coverage through similar models.

developed to meet the needs of Rwandans outside of the formal sector, where access to and utilization of healthcare services had been historically very low [15]. Beginning with pilots in 1999, and established as national policy in 2004, Rwanda quickly scaled Community-Based Health Insurance (CBHI) across the country. Membership grew to 91% of the target population by 2010-11 [16]. Enrollment decreased in recent years, with a June 2015 estimate of 75% coverage, among those eligible for the scheme [16].
Nowadays in Rwanda, health care insurance is highly successful because of the high effort put into it. The new system was implemented, where people paying their contributions according to their income and their financial capacity; this new system use data from UBUDEHE categories [17]. Category one persons have their contribution paid by government, categories two and three pay 3000frws and the fourth category pays 7000frws. This policy has increased the potential of mutual health insurance, so that the majority of the population of Rwanda could benefit from it [18].
Over the years, Rwanda has seen an important increase in total health expenditure (THE) per capita, increasing from US$ 34 in 2006 to 840 in 2014 [19]. Public sources accounted for the majority of total health expenditure, and households contributed for 26% of total health expenditure through out of pocket health expenditure (OOP) [19]. In its efforts to improve access, the country has developed a comprehensive health sector strategic plan [19,20]. A major focus of this plan is the expansion of health insurance to the informal sector through Mutual Health Insurance [20]. Building on the experience of earlier pilots, the government supported start-up initiatives and Mutual Health Insurance schemes were created between 2000 and 2003 [19,20]. Moreover, population coverage increased continuously during this period and was estimated to reach 51.5% in 2004 [20]. Mutual Health Insurance (MHI) was further scaled up in 2005 with the support of external funding [21]. The aim of this expansion was to rapidly increase membership of vulnerable groups through premium subsidies and strengthen administrative capacities and pooling mechanisms [22].
By 2007, around 74% of the population had some form of health insurance cover. Furthermore, in 2008, a formal legal framework for Mutual Health Insurance was created with the adoption of a law on mutual health insurance [23] and this set a new milestone towards universal coverage by making health insurance compulsory. Even though there is facilitation for getting health care utilization for those who have health insurance in Rwanda, there are still problems for some people who did not get all contribution at the same time. This is a serious problem; consider for example a family of nine persons in the second ubudehe category [23]. This family would need to pay a premium for each of the nine members for any of the members to be covered [23]. This is problematic because once you get ill or one of your family gets ill, you cannot receive health care as an insured patient and you have to get medical treatment by paying 100% of the cost as a private patient, unless all family members have paid a premium and this cause mal-treatment in hospitals [23] The findings of other researchers from developing countries assessed the impact of health insurance on health services utilization and health outcomes in Vietnam (G.Ewmmanuel Gwindon, (2014)) [26]; their findings showed that for both poor and students, health insurance increased the use of inpatients services but it was not the case for outpatients services. The same study found that for young children, health insurance increased the use of outpatients services but it was not the case for inpatients services [26]. A study with some similarities was conducted by Agatuba et al., (2012) on the impact of health insurance on health care utilization and out of pocket payments in South Africa and found that the use of health insurance coverage increases the use of private of health services as was expected and there was no significant effect on the use of public services. The same study showed that the use of health insurance did not contribute to the reduction of out of pocket payments compared to those without health insurance coverage [27]. Furthermore, Thuong NTT.et.al (2020) [25], conducted a study on the impact of health insurance on healthcare utilization patterns in Vietnam. Their results indicated that the use of health insurance increased the number of outpatients visits; the enrollment was varying from 0.87 to 1.29 and their findings suggested that participation in health insurance had the most effect on the use of healthcare services at district hospitals. [25]. Finally, referring also to the previous studies in developed countries, on the impact of massive expansion of health insurance program on health care utilization and health outcomes in Japan, Manning et al. (1987) [17] showed a substantial increase in health care utilization, which were much larger than what would be implied by the individual-level effect estimated [17]. Existing research in Rwanda, conducted by Ruhara et al. (2016) [22] on the role of economic factors in the choice of medical providers indicated that health insurance was an important factor in the choice of health facilities and the user fees were major financial barriers to health care access in Rwanda. The same study suggests that as household income increases, patients shift from public to private health facilities where quality was assumed to be higher than for public health providers [28]. Previous research in Rwanda has talked about various aspects of the effect of health insurance on health care utilization in Rwanda focusing much on mutual health insurance or Community Based Health Insurance(CBHI), including helping to understand the determinants of enrolment for mutual health insurance [28]. However, some studies also reveal that out-of-pocket health payment is still powerful and penetratingly in Rwanda despite the presence of Mutual health insurance (Lu C [22,29,30,34,35,36]. To our knowledge no study on the effect of health insurance on health care utilization in Rwanda has used decision tree models to classify districts of Rwanda on the use of health services by households based on the same characteristics of the households.
This paper contributes to the literature in analyzing the effect of health insurance on health care utilization in Rwanda at the household level and complements existing research by analyzing the covariates that influence most the use of health care services in Rwanda.

Study setting
Rwanda has a gross national per capita income of 2155 USD [31] and is a landlocked sub-Saharan African (SSA) country. Rwanda is ranked 160 th out of 189 countries on the 2019 Human development index [31]. Council (NHIC) that has the mandate of advising, supervising health insurance activities and setting the prices for health services as is stipulated in the health insurance law issued in 2016 [33].

Data Source
This study used the data from the Rwanda household living standard survey conducted from 2016-1017 (EICV 5). The survey was conducted by the National Institute of Statistics of Rwanda (NISR) [34]. This nationally representative survey gathered data from 14580 households and 64314 individuals, including head of households who received health services with or without health insurance. [33,34]. Moreover, in this paper data related to consumption expenditure on food and out-of-pocket health expenditures including: consultation; laboratory tests; hospitalization; medication, socio-economic indicators and insurance status, self-reported health need and utilization of health services at household level were extracted. Finally, we have explored the relationship between health insurance and utilization of health care among insured and non-insured heads of households. A logistic regression model will be constructed as well as a decision tree model to assess the interaction between covariates [33,34] 2.3 Variables and their measurements

Dependent variable
In line with the objective of this study to analyze the effect of health insurance on health care utilization in Rwanda, we defined whether the head of household received health services in the last four weeks as the outcome. Whether the head of household received medical care or not will be used as dependent variable and it will take two values such as Y=(0,1), where the head of households received medical care for Y=1, and for Y=0. In this research paper a binomial logistic regression model has been used to estimate the dependents' health services. We also include household expenditure quintiles [35], sex of the head of household, households size, region, household insurance status, district, occupation status of the head of household, marital status of the head of household, head of household insurance status, out of pocket health expenditures of the household, household per capta expenditures. Finally, in this study we performed first the decision tree model for the purpose of classifying the districts based on the use of health services at the household level. Those classes were then used in a logistic regression analysis to assess the interactions between districts and other covariates on the use of health services.

Decision tree model
Decision Trees are a non-parametric supervised learning method used for both classification and regression tasks [31].   Table 2 indicates the logistic regression model where the odds ratio with p values less than 0.05 were used to determine whether a change in a predictor variable makes the use of health care services more likely ceteris paribus.

Fig2: Main health insurance carried out by the head of the household
The high rate of health insurance carried out by the head of the household was found in mutual health insurance and covering 70% of all respondents.

Discussion
This research aimed at examining the effect of the health insurance on health care utilization in Rwanda.
The findings from our study showed a significant age effect. In the present study, the covariates affecting the health service utilization were identified. The households with health insurance significantly increased their use of health services. This study also found that the use of health services in groups of district 1, 2,3 with health insurance used health services compared to those located in Rusizi District.
There is a need for the government of Rwanda to verify the triggers of these decreases of the use of health services in latter districts. The findings of this research contributes to the existing knowledge on the effect of health insurance on health care utilization in Rwanda at household level and complements existing research by analyzing the covariates that influence most the use of health care services in Rwanda.