Factors affecting community based health insurance utilization in Ethiopia: A systematic review CURRENT

Background: The goal of health care financing in Ethiopia has been to achieve universal health care coverage by minimizing the catastrophic out of pocket health service expenditure. Even though the performance was not as planned, the promising strategy to achieve universal health care coverage in the informal sector was community based health insurance which was expected to cover more than 83 % of the population. So, we systematically reviewed determinants of community based health insurance utilization in Ethiopia. Methods: We searched DOAJ, EconBiz, ERIC, Google Scholar, Oxford Journals, PubMed, SpringerLink, Europe PMC, Microsoft Academic Search, OAIster and AJ including various relevant websites by March 9 to 10, 2019. We included articles regardless of their publication status with both quantitative and qualitative approaches. Results: The factors determining community based health insurance utilization in Ethiopia were found to be associated with supply side, health facility, demographic and socioeconomic predictors. Among demographic and socio-economic factors, the report of the studies regarding to gender and age was not consistent. However income, education, community participation, marriage, occupation and family size were found to be significant predictors and were positively related with the scheme’s utilization. With respect to health status and health service related factors; illness experience, benefit package, awareness level, previous out of pocket expenditure for health care service and health service status (quality, adequacy, efficiency and coverage) were significantly and positively related but premium amount, self-rated health status and bureaucratic complexity were found to be negative predictors. Conclusion: To achieve universal health care coverage through community based health insurance, it is advisable that special attention should be given to income level, education, community participation, marriage, family size, benefit package, awareness level and health service quality, premium amount and bureaucratic or governance issue.


Study selection
The literatures were screened by two independent reviewers (EMB and MHK). Firstly, the articles were refined by their title and abstract; secondly, by full text revision by these Authors independently and finally together.

Eligibility criteria
Any study that reported factors affecting CBHI was included with both qualitative and quantitative approach. For this article, CBHI is redefined as a non-compulsory and non-profit making risk pooling mechanism in exchange for premium payments by members aimed to address the issue of UHC by limiting financial hardship and is based on the community's traditional risk sharing arrangements.
Iddir based health insurance is thus taken as a synonym for CBHI. The included articles were those that had reported the willingness to participate, join, pay, uptake, enroll and utilize CBHI as the priority finding; and health care utilization, attitude to CBHI, availability of health service and medical equipment, satisfaction with CBHI, dropout rate from CBHI and CBHI membership renewal as additional outcome.

Risk of bias assessment
Risk of bias for individual studies was assessed by developing a modified assessment tool in a scale of measurement from "Strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies" [34]; and "Cochrane Handbook for Systematic Reviews of Interventions: Assessing Risk of Bias in Included Studies" [35]. Differences during the extraction process were resolved via discussion by the reviewers with third reviewer involvement when necessary.

Data extraction
The data extraction was accomplished by two reviewers (EMB and MHK) independently and together from all included articles by using data extraction form prepared in advance. The data and information extracted from every included study were the following: study citation, research approach and design, study area and period, study finding, unit of study and outcome. 6

Dealing with missing data
For any ambiguous information found in the included articles, we have tried to contact the Authors even though we failed to found some of them. But, even the authors that we have contacted have not followed and given us the necessary information.

Data synthesis
Due to the heterogeneous nature of the articles in their finding and outcome variables considered, we have done a detailed discussion thematically.

Study selection
All valuable sources for the review were searched and found from different data bases such as Journals (AJ) and Addis Ababa University (AAU) data base as well as from other sources through March 9 to 10, 2019 using keywords. The key words used were determinants, factors, health insurance, community based health insurance, willingness to utilize, willingness to join, willingness to pay, willingness to uptake, willingness to enroll and Ethiopia. The search was conducted by systematically combining these key words. Totally, 272 records were identified from which 257 were from data bases and the rest 15 from other sources. 79 articles were duplicates and removed by using EndNote X7.
After the duplicates were removed, 193 records were screened for eligibility by their title and abstract and 159 were excluded as they were not relevant. Then, 34 articles were found to be eligible for full text analysis. Based on full text review, 14 articles were excluded. Finally, 20 articles were included for systematic review (refer to figure 1).

Factor analysis
The utilization of CBHI service was found to be affected by supply-side factors like service availability and coverage [51], skill deficit of CBHI officials and inadequate manpower, budget deficiency to community mobilization, low accessibility of health institutions, delay to fulfill formalities to the insures, absence of registration materials and office inconvenience (narrow); demand side factors (demographic and health perception factors) like fail to fulfill formality and providing necessary information for registration and forgetting having receipt in health sectors) [43].; socio-economic factors like participation in Iddirs and health facility factors like service quality and trust [51].
From the demand side, the major factors negatively affecting CBHI utilization were seasonality of income, geographically scattered settlements and mobility of pastoralists and negative perception towards health insurance [12]. Also, with the existence of chronic illness [52], adverse selection (inclusion of chronically ill, poor and indigents; the poorest of the poor) [12,27,44] and moral hazards (miss utilizations) during enrolment were the limitations [12,27]. On the other hand, expectation of long enrollment time more than a year with a single payment [53]; consideration of CBHI as profit making and expectation of double registration in a single payment; if not used health service, consider the payment as reserve deposit as well as fail to pay fee and have ID card were found to be challenges [54]. Members of CBHI commonly practice self-medication entails the scheme's utilization was poor even among the members [55].
From the supply side, discrimination between cash and insurance users, bureaucratic complexity in cost reimbursement, lack of trained personnel, adverse selection, fraud and corruption [12]. The health professionals' job satisfaction is negatively related with the health sector reform [56]. Service hour and efficiency of CBHI service were found to be positively associated with satisfaction of CBHI scheme [57], work load without additional incentives negatively affects CBHI utilization [19]. Most of the health institutions were not ready to CBHI scheme's requirements/criteria [58]. Inadequacy of health sector in terms of quality, organized working practice, resources and premises; forcing members to search for expensive private sectors [54]. Registration time and cost [46], delay in membership card provision had hardily affected CBHI utilization [48].
In comparing the regions that CBHI has been implementing, households in Amhara and Oromiya regions were found more likely to enroll as compared to households living in Tigray and SNNPR. CBHI members in SNNPR have limited access to tertiary health care services; insured households can use tertiary services only at the nearest public hospital (while those in Amhara may visit any public hospital within the region but those in Oromiya may use care from public hospitals within and outside the region). Insured households in SNNPR cannot claim reimbursements if they use health care services from private providers in the event that medical equipment or drugs are not available in CBHI linked facilities [59]. The determinants are broadly categorized as follows (figure 1).

Demographic and socio-economic factors
The health care effectiveness largely depends on the socio-economic aspects of the family/household [25]. Six and four studies reported that there was a positive relationship between CBHI, and being male [20, 38,42,43,46,47] and female [15,21,37,40] headed of the households respectively.
Being male [38,42] and female [40] were found to be positively related with the willingness to pay (WTP) for CBHI respectively. Being male was also positively related with enrolment to [20, 43,47] and utilization of [46] CBHI. On the other hand, being female was found to be positively associated with the willingness to join (WTJ) to CBHI [15,21,37]. Age was also found to be a positive predictor to CBHI utilization [36,43,44,46,48,57,60]. It was also found to be a negative predictor to the scheme's utilization [17,20,37,40,61]. and enrolment to [64] CBHI scheme.
Occupational status, such as farming, merchandise and housewife were found to be associated with CBHI utilization. In most studies that reported employment as a predictor to CBHI utilization, it was found to be a positive factor [17,21,37,38,40,46]. But in one study it was reported that holding occupation was a negative factor for CBHI utilization [17]. Households who are employed were found to be willing to pay for CBHI [39]. Being farmer was a positive factor to the WTP for [38,40], compliance with [17] and utilization of [46] CBHI scheme. Holding merchandise occupation was a positive factor to the WTP for CBHI [38] but negatively related with the scheme's compliance [17].
Housewife as an occupation was a positive factor for the WTJ in CBHI scheme [21,37]. Regarding to income, including monetary and nonmonetary assets, it was found to be a positive predictor of CBHI  [18,22], uptake of [45,46], enrolment to [47,50], membership to [63], knowledge, attitude and practices of [60] and health care utilization among members of [61] CBHI scheme. However, one study reported that income was negatively related with CBHI enrolment [64]. Ability to pay for health care cost (financial capability) was positively related with the WTJ in CBHI [36]. Livestock size was negatively related with the WTP for CBHI scheme. Poor households (food insecure) had positive interest to the WTP for [40] and enrolment to [59] CBHI. was not only a positive predictor of enrolment to CBHI [59] but also a negative determinant of WTJ in the scheme [36]. In all studies that reported premium as a predictor of CBHI utilization, it was found that premium amount was negatively related with the scheme's utilization [17,22,[36][37][38]46] and practice to be a member in [61] to satisfy with [53,54] and utilization of [46] CBHI. Insufficiency 13 of health service both in equipment and human power was a negative predictor to the scheme's utilization [12]. Laboratory service provision was positively related with the uptake of CBHI [53,57]. Availability of medical equipment was positively related with the WTP for [40] and enrolment to [44,59] scheme. It was reported that health service delivery system provided by CBHI scheme was not satisfactory in terms of quality, referral system, human resource and building facility [54]. Prior health insurance & health service utilization was reported as both positive [42] and negative [20] predictor for the WTP and enrolment to CBHI respectively. Health sector distance was negatively related with the participation in [18], enrolment to and WTJ [37] CBHI. Travel time was negatively associated with the enrolment to the scheme [59,62]. Waiting time was also negatively related with the WTP for [40] and utilization of [46] CBHI. Trust on CBHI scheme was positively related with the WTJ [36], WTP for [42], enrolment to [48,51] to uptake [44,46]  administrative complexity was found to be a negative predictor to enroll to CBHI [47].

Discussion
In this review, studies with quantitative, qualitative and mixed approach methods were included. Most studies included in the review used cross sectional approach. This review revealed that the factors that affect CBHI utilization in Ethiopia can be grouped into: Demographic and socio-economic factors; and health status and health service related factors.

Demographic and socio-economic factors
Gender was found to be significant predictor of CBHI utilization. However, neither maleness nor femaleness has found to be a clear predictor of CBHI. This review revealed that as the household head became male, the willingness to utilize (WTU) CBHI was found to be increased; i.e., households which were headed by male were more willing to utilize CBHI scheme as compared to those households that were being led by females [20, 38,42,43,46,47]. Oppositely, as the head of the family became female, the WTU the scheme was found to be increased. Families which were headed by female were more willing to utilize the scheme as compared to those families that were being led by men [15,21,37,40]. On the other hand, as the household head became female, the WTU the scheme was found to be decreased [47]. Regarding to age of the family members, no consistent result was found; the results of the included studies were not found to be consistent. Older individuals were more willing to utilize the scheme than younger ones [36,43,44,46,48,57,60]. On the other side, if most of the family members became younger, the WTU the scheme was found to be increased; i.e., since the scheme covers family members up to 18 years of age, families with greater younger members were more willing to utilize the scheme than those families with older members [17,20,37,40,61]. With respect to marriage, the WTU the scheme was found to be more among those who were married than those who were not which entails us that social work is mandatory to keep intact family; minimizing/avoiding divorce to protect the family is important to increase CBHI utilization [20,23,37,43,47,62]. Regarding to family size, as it increased; the WTU the scheme did so. Those households with large family size were more willing to uptake the scheme than those households with small family size [1, 18, 20, 22, 23, 37-39, 41, 42, 44-51, 62]. However, as family size increased, the payment for the premium also increased. As a result, satisfaction with CBHI utilization was decreased with increased family size [40,57]. Considering educational level of the household, as educational attainment becomes advanced, the tendency to utilize CBHI scheme was found to be appreciable; as educational attainment increased; awareness level to the scheme was also increased accordingly [1, 15, 18, 21, 23, 36, 38-44, 46-49, 57, 60-63]. But, due to negative attitude to the scheme, it was also reported that as educational attainment increased, the WTU the scheme was found to be decreased [37,64].
With respect to income including monetary and nonmonetary assets, except some studies [40,59,64], the report of all articles were consistent. As income of the family becomes increased, they could be able to afford the premium, thus, the decision to utilize the scheme was also found to be analogously increased [1, 15, 18, 21-23, 36-39, 42, 45-47, 50, 60, 61, 63]. Oppositely, as livestock size increased, the WTU CBHI was found to be decreased; that means, wealth households were not interested to utilize the scheme; i.e., livestock considered to be the reserved asset of the family [40] whereas food insecure families were found to be interested to use it in that they could not afford to pay the expensive OOP payment [40,59,64]. Regarding to occupation, the WTU the scheme was higher among those households being led by heads who were employed than those who were not [39]. The WTU the scheme was found to be increased with being housewife [21,37] , farmer [17,38,40,46] and merchant [38]. However, the WTU the scheme was also found to be less among merchants [17].
Community participation, here, which is defined as peoples interaction or involvement aimed to solve critical problems in the respected society was also an important factor. Households that were known to participate in local meetings were more willing to utilize CBHI than those households did not attend [20,48,49]. Households which were members in Iddir and Ikub [15,21,37,42,44,51,59] and PSNP [32,59,[64][65][66] were found to be more willing to uptake the scheme than those households that were not. But, participation in PSNP was also known to decrease the WTU the scheme [20]. The WTU the scheme was found to be more likely among those families depend on individual social capital and community level horizontal trust [37]. As social bond or community solidarity increased, the tendency to utilize the scheme was found to be more likely [50]. Due to the ideology difference of different religions, the interest to uptake the scheme was found to be less [36] or more [59] among households having bond to religious beliefs & values.

Health status and health service related factors
Chronic illness requires continuous follow-up and prescription refill which in turn devoted to incur huge expense. Consequently, the WTU the scheme were more among those families with morbidity/chronic illness and illness experience than those who were free of it [22, 27, 36, 39-42, 45-48, 50, 61, 62]. Oppositely, the WTU the scheme was found to be decreased with good perceived health status or self-rated healthiness of the family; i.e., if the perceived health status of the family was increased, the WTU the scheme was decreased [36,39,44,46,50,51].
As premium cost increased, its affordability decreased. Thus, the WTU the scheme was found to be decreased with higher premium cost [17,22,[36][37][38]46]. As income of the family increased, the premium became affordable, thus, the decision to utilize the scheme was also found to be analogously increased [1, 15, 18, 21-23, 36-39, 42, 45-47, 50, 60, 61, 63]. However, as the premium becomes increased, the utilization of the scheme found to be decreased [17,22,[36][37][38]46]. In reverse, with respect to experience of health care service cost expenditure, the households that were experienced previous OOP expense for health service were more willing to use the scheme than those households did not [20, 40,61,62]. But it was also reported that OOP payment was better than CBHI [38]. However, households experienced borrowing money for health care service to pay OOP payment were more interested to use the scheme than those households did not borrow for that [15,37].
Accordingly awareness level, as knowledge and information about CBHI increased, the WTU the scheme also found to be increased [12, 17, 18, 20, 36, 41-46, 48-50, 53, 61, 64, 67]. However, the tendency to utilize the scheme was found to be increased with the attitude towards sense of ownership of the households with CBHI. The WTU the scheme was found to be appreciable if unused premium was reserved for future use as a deposit for the payer for the future [49].
The WTU the scheme was increased with the increased quality, availability, accessibility, coverage, adequacy of health service and capacity and readiness of health facility [12, 36, 38, 40, 42, 44, 46-48, 50, 51, 53, 54, 61, 62, 64]. The WTU the scheme was found to be decreased with insufficiency of health service both in equipment and human power [12]. If laboratory service provision was available, the WTU the scheme was more likely [53,57]. If the health facilities were well furnished, the WTU the scheme was increased [40,44,59]. However, it was reported that health service delivery system provided by CBHI scheme was not satisfactory in terms of quality, referral system, human resource and building facility [54]. Experience of prior health insurance & health service utilization found to be either increased [42] or decreased [20] the WTU the scheme. While the health sector distance increased, the WTU the scheme decreased [18,37]. If the time to access the health institution was difficult; if it was too far to reach, then the WTU the scheme become decreased [59,62]. If the waiting time to be served in the institution was too long, the WTU the scheme was unlikely [40,46]. If the service to be provided by CBHI was traceable (amenable), the WTU the scheme was found to be  Figure 1 Flow diagram of literature screening strategy.
30 Figure 2 Thematic classification of the factors affecting CBHI in Ethiopia.
31 Figure 3 Summary of the determinants of CBHI in Ethiopia.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.
Risk of bias assessment tool.docx PRISMA checklist.doc Summary of data extraction from the included studies.docx