Operational and structural factors influencing enrolment in community-based health insurance schemes: An observational study using 12 waves of nationwide panel data from Senegal

Community-based health insurance (CBHI) has been implemented in many low-and middle-income countries to increase financial risk protection in populations without access to formal health insurance. While the design of such social programmes is fundamental to ensuring equitable access to care, little is known about the operational and structural factors influencing enrolment in CBHI schemes. In this study, we took advantage of newly established data monitoring requirements in Senegal to explore the association between the operational capacity and structure of CBHI schemes-also termed ‘mutual health organizations’ (MHO) in francophone countries-and their enrolment levels. The dataset comprised 12 waves of quarterly data over 2017-2019 and covered all 676 MHOs registered in the country. Primary analyses were conducted using dynamic panel data regression analysis. We found that higher operational capacity significantly predicted higher performance: enrolment was positively associated with the presence of a salaried manager at the MHO level (12% more total enrolees, 23% more poor members) and with stronger cooperation between MHOs and local health posts (for each additional contract signed, total enrolees and poor members increased by 7% and 5%, respectively). However, higher operational capacity was only modestly associated with higher sustainability proxied by the proportion of enrolees up to date with premium payment. We also found that structural factors were influential, with MHOs located within a health facility enrolling fewer poor members (-16%). Sensitivity analyses showed that these associations were robust. Our findings suggest that policies aimed at professionalizing and reinforcing the operational capacity of MHOs could accelerate the expansion of CBHI coverage, including in the most impoverished populations. However, they also suggest that increasing operational capacity alone may be insufficient to make CBHI schemes sustainable over time. In SSA, CBHI has become popular in policy circles where mandatory enrolment in health insurance schemes is perceived to conflict with the prerogatives of the State or the values prevailing in the population, e.g. where this is deemed ‘a too intrusive state interference in the individual sphere’ (Fonteneau et al., 2017). This is typically the case in Senegal, which has a long-standing tradition of CBHI (starting as early as the 1980s in the region of Thiès) and where expanding CBHI was put on top of the political agenda by President Macky Sall during his electoral campaigns in 2012 and 2019. Since his election in 2012, CBHI has been propounded as one of the core strategies to achieve UHC (Deville et al., 2018). This has translated into the creation of a National Agency in 2015, whose missions include structuring and accelerating the development of CBHI through sustainable MHOs (mutuelles de santé; Appendix, p. 2). MHOs in Senegal target all individuals not covered through formal-sector protection schemes, free healthcare initiatives (mainly for children, elders and the disabled) or private health insurance, including workers in rural and informal sectors (95% of the active population) and unemployed individuals (Deville et al., 2018). This corresponds to about 80% of the total country population (Fonteneau et al., 2017), i.e. 12 205 077 individuals in 2017 (Agence Nationale de la Statistique et de la Démographie. Rapport projection de la population du Sénégal, 2016). In line with conventional CBHI schemes, insurance coverage is not automatic, and enrolment is voluntary. However, in contrast to traditional CBHI schemes, the model established by the National Agency has three specific features. First, the benefits package and insurance premium are standardized at the national level. The scheme’s benefits package covers a range of essential health services and drugs delivered or prescribed by contracted providers, including health posts (primary care), health centres (primary and secondary care), referral hospitals (tertiary care) and private pharmacies. These include outpatient care (consultations, surgery and diagnostic tests), maternal care (ante-/post-natal care and family planning), hospitalization and transport. Limitations of the benefits package include long-term treatments for chronic diseases (e.g. cancer, diabetes, chronic asthma and chronic mental health), which are not covered as a prudential rule to limit the financial risks associated with overconsumption and, more generally, due to limited financial capacity (Ouattara and Ndiaye, 2017). Second, the State fully subsidizes user fees and insurance premiums for the poor. Specifically, this involves two categories of individuals: (a) the beneficiaries of the ‘Programme National des Bourses de Sécurité Familale’ (National Family Security Grant programme—PNBSF), a programme through which the poorest households of the country receive quarterly cash transfers (25 000 Francs CFA—about USD 45); (b) the holders of the ‘Carte d’Egalité des Chances’ (Equal Opportunity Card—CEC), which is delivered to people living with disabilities and gives them access to various subsidized social services. The State subsidizes only 50% of the insurance premium for all other enrolees. In addition, the State fully covers the co-payments of PNBSF and CEC enrolees for health services and drugs included in the benefits package. For regular enrolees, the scheme only covers 80% of health services and generic drugs and 50% of specialty drugs. Third, the scheme is organized so that MHOs only handle the insurance premiums for health services and drugs delivered or prescribed by health posts and health centres (i.e. for the first two levels of care). The rest of the fund, dedicated to health services and drugs delivered or prescribed by referral hospitals, is pooled at the departmental level and managed by the MHO departmental unions. As such, larger financial risks associated with (more costly) tertiary care can be pooled by larger groups (Deville et al., 2018; Bossyns et al., 2018; Daff et al., 2020). By the end of 2016, the national programme had led to the creation of 676 MHOs in all 552 municipalities of Senegal (Daff et al., 2020). However, despite these efforts, the share of the population enrolled in MHOs remains below the planned objectives. According to recent estimates, only 26% of eligible individuals were covered in 2017 (Agence de la CMU, 2020), leaving many households at risk of catastrophic health expenditures. To address this issue, the government recently launched a new ‘Strategic Development Plan towards UHC’ (2017–2021), one of the objectives of which is to accelerate the professionalization of MHOs. This plan includes, among other initiatives, subsidies supporting the recruitment of salaried managers to reinforce their operational capacity. Prior to 2017, MHOs were largely operated voluntarily. Community members in charge of the day-to-day running of the scheme (i.e. enrolling members, collecting premiums, managing finances and reimbursing providers) would only receive modest compensatory allowances as financial conditions permit. However, transition to a higher degree of professionalization has since been fostered by the National Agency. In 2017, the Agency committed to fully covering the salary of one manager per MHO for 6 months to support the professionalization of management activities (75 000 Francs CFA per month—about USD 135, standardized at the national level). In return, MHOs committed to sustaining the transition efforts towards professionalization by continuing to pay the wages past these 6 months. A major implication of the government’s development plan has been the setup of rigorous data monitoring requirements for MHOs. Since the first quarter of 2017, each MHO in Senegal has been required to deliver a quarterly report of their activities. This is done using standardized forms that summarize the progress of several indicators, including operational (e.g. number of enrolees), financial (e.g. amount of membership fees received) and administrative variables (e.g. number of meetings held by the Executive Board). The National Agency in charge of structuring the development of CBHI collects these reports through each of its 14 regional branches, where their content is checked and transcribed electronically. Such a detailed data collection process at the MHO level, nationwide and over several years is time- and resource-consuming and requires a strong political commitment. We are not aware of the existence of similar microdata elsewhere in SSA. We collated all reports from the start of 2017 to the end of 2019, which amounted to 12 waves of quarterly panel data covering all functioning MHOs in the country. Of the 676 MHOs registered in Senegal, 22 were identified as having discontinued their operations before the studied period and were therefore not included in the analysis. In addition, we excluded the 26 MHOs located in the departments of Foundiougne and Koungheul. MHOs in these two departments are subject to a special regime of governance that translates into a different implementation of the CBHI scheme through so-called departmental health insurance units (Unités Départementales d’Assurance Maladie; UDAM) (Bossyns et al., 2018). The UDAM system is a model of larger-scale insurance units at the department level (rather than the community level, on which this study focuses), with specific features: (a) centralized scheme management at the department level; (b) advanced technical assistance to MHOs; (c) advanced degree of professionalization and (d) longer-term financial and technical partnerships with external development partners. These specificities create two challenges for our analysis. First, due to major structural differences, the performance of MHOs based in UDAM departments cannot be assessed through the same lens as regular MHOs. Second, due to centralized management, some variables of interest are unavailable at the MHO level in the UDAM departments. Finally, we excluded 10 MHOs for which enrolment was reserved for members of professional guilds. The performance of these MHOs cannot be directly compared to those of regular MHOs due to restricted enrolment conditions. As a result, the sample for analysis included 618 MHOs located in 43 departments in all 14 regions of Senegal. Further description of the dataset is provided in Appendix, p. 3. Other variables used for data analysis, i.e. control variables at the community, health district, departmental or regional level, were collected from publicly available databases and reports published by the National Agency of Statistics and Demography and the Ministry of Health and Social Action of Senegal (Appendix, p. 4). We selected three performance and sustainability indicators at the MHO level. First, we focused on the total number of enrolees to indicate the MHO’s capacity to reach its target population. This number includes all individuals enrolled in the CBHI scheme in the community, i.e. household heads who registered as MHO members and (either all or some of) their household members. Second, we selected the enrolment of poor MHO members as an indicator of the scheme’s capacity to provide equitable access to financial risk protection. ‘Poor members’ refer to financially disadvantaged individuals for whom the State fully subsidizes user fees and insurance premiums, i.e. the beneficiaries of the PNBSF and CEC social programmes. For all other members (‘regular members’), the State only subsidizes 50% of insurance premiums (Daff et al., 2020). Third, as an indicator of the scheme’s capacity to be sustainable over time, we examined the proportion of regular enrolees up to date with their premium payment (‘up-to-date enrolees’). Financial viability is key to envisioning the transition from a CBHI scheme that heavily relies on public subsidies to a successfully self-financed one. On the operational side, we focused on three variables aimed to be proxies for the operational capacity of MHOs: (a) the presence of a salaried manager in charge of the day-to-day operations of the scheme (including registering new members, reviewing invoices and reimbursing health posts) and the dissemination of information about the scheme in the community; while, before 2017, MHOs used to be mainly run voluntarily, significant efforts have been made to transition towards a higher degree of professionalization through the recruitment of salaried managers; (b) the amount of operating expenditure per member; these expenses notably include the transportation of Board members attending meetings and awareness-raising activities, thereby capturing a certain degree of managerial dynamism and (c) the number of contracts signed with health posts, reflecting managerial competency and trust in MHO staff by healthcare providers. In the Senegalese system, each MHO is responsible for negotiating their contracts with local providers, i.e. health posts, which deliver primary care, health centres, which provide primary and secondary care (health centres can be regarded as district hospitals; Mané, 2012), and pharmacies. Contracts with referral hospitals (tertiary care) are negotiated at an upper level by the departmental unions. However, the number of health centres is fairly low compared to the number of health posts (in 2016, these numbers were 100 and 1458, respectively; Agence Nationale de la Statistique et de la Démographie, 2017). This is especially true in most rural regions; in 2016, for example, there were only three health centres in the regions of Kédougou and Sédhiou (Agence Nationale de la Statistique et de la Démographie, 2017). To capture a greater variation between MHOs, we only focused on the number of contracts signed with health posts. Health posts, located in urban and rural communities at the municipality or village level, constitute the country’s backbone network of care provision. On the structural side, we selected two variables relating to the location or history of the MHO: (a) whether the MHO’s head office was located within a health facility (i.e. a health post or centre) or not to explore whether closer proximity between care providers, eligible individuals and MHO staff could predict higher levels of enrolment and (b) whether the MHO started its activities before or after the wave of MHO creation initiated by the government in 2016 to examine the role of a pre-existing tradition of CBHI in the community. We estimated linear dynamic panel data (DPD) models to assess the association between the three indicators of performance or sustainability and the selected operational and structural factors between 2017 and 2019 using quarterly timepoints and MHOs as cross-sectional observation units. Unlike ‘static’ panel data models, such as fixed-effects (FE) or random-effects models, DPD models allow both current and past information to be accounted for, i.e. they model current performance indicators as a function of past performance and both past and current values of selected operational variables. Such models typically write as follows: where denotes performance indicator for MHO in quarter ; denotes its value in the previous quarter; represents selected time-varying variables for MHO in quarter (including past values); represents selected time-invariant variables for MHO ; denotes quarter effects; represents MHO (unobserved) time-invariant heterogeneity (i.e. the permanent component of the error term, commonly referred to as unit-specific effects) and denotes the transitory component of the error term. Coefficients , and are regression parameters to be estimated. We employed an approach consisting of a two-stage sequential estimation procedure where the coefficients of time-varying regressors were estimated using Arellano–Bond estimators (first stage), and the coefficients of time-invariant regressors were subsequently recovered using linear panel data estimators (second stage) (Arellano and Bond, 1991; Kripfganz and Schwarz, 2019). Control variables were added to the models to reduce the risk of confounding effects in the second stage, including demographic and socioeconomic factors at the community level—population, urban or rural administrative status, poverty rate estimated based on household consumption, presence of another MHO (since premium prices and benefits packages are standardized at a national level, competition between MHOs is only based on location and contracts with providers)—and supply-side factors related to care provision at the departmental or regional level. These included population per health post, geographic accessibility of basic health services (measured by the average distance between health posts and the closest health centre in each department) and the availability of basic patient services. The latter reflects, in each region, the proportion of healthcare facilities (health posts, health centres and hospitals) equipped to provide six essential health services: (a) child health: curative care services; (b) child health: growth monitoring services; (c) child health: vaccination services; (d) family planning services; (e) antenatal care services and (f) services for sexually transmitted infections. Lastly, fixed effects at the departmental level were included. We also estimated standard FE models to assess the robustness of the association between performance indicators and time-varying predictors. We selected the FE specification after testing the assumption that there was a significant correlation between the unit-specific effects and the selected predictors (which, if not accounted for, would lead to inconsistent estimates when using linear estimators) using Mundlak tests (Mundlak, 1978). We explored the heterogeneity of the association between performance indicators and time-varying predictors based on the following factors: (a) urban or rural community; (b) poverty rate in the community; (c) geographic accessibility of basic health services in the department and (d) incidence of malaria in the health district, to investigate whether reinforcing the operational capacity of MHOs could be more beneficial in areas with higher demand for health services.