The study was conducted in Kellem Wallaga, Oromia regional state, Ethiopia, among insured rural households from March 30, 2022, to April 30, 2022, using a community-based cross-sectional study design. The zone was one of the 20 zones of the Oromia regional state, which was established in 2007, and it had 12 districts, one of which was urban and the others rural. Dembi Dollo is the capital of the zone, which is located about 652 km west of Addis Ababa, the capital of the country.

The zonal health department’s reports of 2019/2020 showed that six districts were implementing the CBHI scheme, and all of these districts have begun offering health services as part of this program. Seyo and Gidami districts began offering services in 2016, while the four remaining districts namely Dalle Sadi, Dale Wabera, Hawa Galan and Lalo Kile began doing so in 2020. They launched the CBHI in different year; 53.5% of the zone’s population were enrolled in the CBHI. The two pilot districts were selected for this study because the remaining four districts have not yet started the CBHI membership renewal process.

The total population of the Gidami was estimated to be 112,190 and from a total of 7,010 households, only 3,225 (46%) were enrolled in the CBHI program. The community had five government health centers and 22 health posts that provide health services for the community. The total population of the Seyo district was 76,207, and from a total of 4,760 households, only 2,428 (51%) were enrolled in the CBHI program. The district had 6 government health centers and 24 health posts that provide health services for the community, according to the 2019/2020 zonal health department report (16).

All households who were ever enrolled in the CBHI scheme in rural districts of Kellem Wollega Zone were considered as the source population. All households that were registered to the CBHI scheme and found in selected villages of the zone were considered as the study population, whereas the household head who ever enrolled in the CBHI scheme was considered as the study unit.

All households enrolled in the CBHI scheme in selected kebeles of the Kellem Wollega zone were included in the study, whereas all households who were members of the scheme but came from other areas, whose residence was less than 1 year, and those whose enrollment duration less than 1 year were excluded from the study.

Sample size was calculated by using the single population proportion formula, assuming 68.1% of the households made the decision to renew their CBHI membership (17), 95% CI, and a 0.05 margin of error. Since two-stage sampling was used, the 1.5 design effect was considered, n=Zα/22p1−pd2, where, n = calculated sample size, z = 95% CI, d = margin of error at 0.05, and p = proportion of households’ membership renewal decision = 68.1%.

So, n = (1.96)² × 0.681(1–0.681)/ (0.05)² = 334, multiplying by 1.5 design effect and adding 10% non-response rate, the final sample size determined was 551.

Since the zone consisted of 12 districts and 256 rural kebeles, and six districts were implementing the CBHI scheme while the other four districts started to implement the scheme in 2020, presumably two districts that started implementing the CBHI scheme before 5 years ago when the study was conducted were selected. Then the required sample size was proportionally allocated to both districts, and by the simple random sampling method, kebeles (villages) were selected from both districts, and 30% of the total kebeles were selected from both districts. Accordingly, 6 kebeles out of 22 kebeles from Gidami district and 7 kebeles out of 24 kebeles from Seyo district were selected (Figure 2).

The lists of households having households or spouses with informal work were obtained from the district administration office (village register), and after getting the lists of households from each selected kebele, samples were selected by systematic random sampling to get the data required. Finally, the study participants were selected using systematic random sampling methods from the CBHI insurance scheme registration book list at every k^th household in the frame from each kebele until the sample size of that kebele reached the value shown in this formula: k = N/n. Accordingly, N = study households (5653), and n = calculated sample size (551); then k = 5653/551 = 7, every 7^th household was selected from each kebele. The lottery method was used to select the first households from 1 to 7 serial numbers.

Community-based health insurance scheme membership renewal decision was considered as a dependent variable, whereas socio economic and demographic factors (age, sex, occupation, gender, religion, educational status, marital status, household size, wealth index, presence of under five children in households, and presence of the older adult in households), health and health services related factors (accessibility of health facility, availability of health service, utilization of health service, perceived quality of service, chronic disease, and recent illness episode), CBHI-related factors (benefit package, scheme experience, premium collection convenience time, duration of CBHI enrolment, affordability of premium collection, premium collection convenience time and duration of CBHI) (Figure 1).

An interviewer-administered structured questionnaire that was adapted from related articles (12–14, 17) to collect relevant information. The questionnaire had five parts: socio-economic and demographic characteristics; CBHI information; individual or household-related factors; CBHI-related factors; and health service use-related factors.

Data was collected by six BSc-trained health professionals using the pre-tested structured questionnaire through face-to-face interviews. Two supervisors who were qualified with a master’s degree in public health were recruited. The data collectors and supervisors were trained for 1 day on the objectives of the study, the data collection tool, approach to the interviews, details of interviewing techniques, respect, and maintaining the privacy and confidentiality of the respondents.

After the data collection from each selected household, the data was entered into EPI data version 3.1 and analyzed using SPSS 25 software. Hosmer-Lemeshow goodness-of-fit was used to check the fitness of a multi-variable logistic regression model that had been eventually fitted, the p-value was found to be 0.712. Descriptive statistics, including frequencies, cross-tabulation, averages, and percentages, were performed. Binary and multiple logistic regressions were used to identify a significant association between the CBHI membership renewal decision and independent variables. Using multiple logistic regressions, a significant association between the CBHI membership renewal decision and independent variables was identified, with the statistical significance level using 95% CI at p < 0.05.

The internal consistency of items for independent variables measured in Likert’s scale was checked by Cronbach’s alpha. The questionnaire was prepared in English and translated from English to Afan Oromo. Before the actual data collection, pre-testing was done by the data collectors on 28 individuals (5% of the sample size) in Dalle Sadi district, Kellem Wollega Zone, and appropriate modifications were made after analyzing the pre-test results. To ensure data quality, the principal investigators provided training all data collectors and supervisors. All collected data were checked for completeness and consistency by the supervisors every day, and onsite close supervision and technical support were given by supervisors and the principal investigators. During the data collection period, when the household head was unavailable, the data collectors asked the house member who could respond to the interview or when he/she would be available and returned to home in the next day for interview.

Ethical clearance (Ref: WU/RD/526/2014) was obtained from the Research Ethics Review Committee (RERC) at Wollega University. A formal letter that explained the objectives, significance, and expected outcomes of the study was written to Kellem Wollega zonal health department from the Wollega University, Institute of Health Sciences and requested cooperation. Based on a supportive letter from Wollega University, Kellem Wollega zonal health department issued a support letter to the districts Health Office.

All study participants were well informed about the purpose of the study, and informed written consent was secured from the study participants prior to the interview. The study participants’ confidentiality was maintained, and no personal identifiers were used in the data collection tools; instead, codes were used in their place. All paper-based and computer-based data were kept in protected and safe locations. The recorded data were not accessed by a third party except the research team, and data sharing will be enacted based on the ethical and legal rules of data sharing.

A total of 540 participants responded to the interview, making the response rate 98%. The mean age of the respondents was 45.14 ± 10.84 years. Majority of the respondents, 494 (91.5%), were males, and 506 (93.7%) of the respondents were married. One third, 174 (32.2%) of the participants were unable to read and write. More than three-fourths, 472 (87.5%) of the study participants were farmers. About 250 (46.3%) of the households had fewer than five members in their household. The wealth status of the households was ranked. As a result, 108 (20%), 107 (19.8%), 109 (20.2%), 111 (20.6%), and 105 (19.4%) were extremely poor, poor, middle level, rich, and very rich, respectively (Table 1).

The participants were interviewed about the source of information they heard about CBHI for the first time before they became members. Accordingly, about 344 (63.7%) of them heard about the scheme from officials in public meetings, followed by health professionals in health facilities, which accounted for 164 (30.4%), during CBHI house-to-house awareness creation campaigns, 22 (4.1%), mass media (7.3%), and from neighbors or friends (0.6%).

The study assessed the health status and health care seeking behavior in the last 12 months among households (Table 2). As a result, three-fourths of the participants’ members, 383 (75.2%), were free from chronic diseases. The participants also rated their family members’ health status. Accordingly, the majority of them 402(74.4%) rated it as good, whereas only 138 (25.6%) rated it poor. About 428 (79.3%) household members have been ill during the last 12 months and 405 (75%) of them have sought medical treatment from the contracted health facilities. Out of 405 household members, half of them, 203(50.1%) visited a contracted health facilities only once, whereas only 19(4.7%) visited the facility more than four times in the last 12 months.

The decision made by households to renew their CBHI membership has also been evaluated by the study. As a result, the majority of households 365 (67.6%, 95% CI = 63.7–71.5%) decided to renew their membership and did so on an annual basis as per to CBHI scheme membership renewal schedule (Figure 3).

Among the study participants, 450 households (83.3%) were payers, and 90 (16.7%) of them were the poorest, exempted from the scheme’s contributions. Almost all of them, 479 (88.7%), had CBHI identification cards. Of those who decided to renew their membership, only 19 (5%) had done so consistently over the past 5 years. Among the 175 households that decided to drop their membership, 153 (87.4%) and 22 (12.6%) of them had stayed in the scheme for one to 3 years and 4 years, respectively. About 24 (4.4%) and 125 (23.1%) of those who renewed and dropped their membership, respectively, had decided not to renew their membership in the next coming year (Table 3).

From a total of 175 (32.4%) who decided for the first time to drop their membership, more than one-third of them, 69(39.5%) made their decision after 2 years of membership, followed by 57 (32.6%), 27 (15.4%) after 1 year, 19 (10.9%) after 4 years, and 3 (1.7%) after 5 years. These households reasoned out their best single reason why they decided to drop their membership, such as the quality of health services being low 84(48%), followed by a lack of awareness about the details of how the CBHI works, 32(18.3%) (Figure 4).

More than half of the households 298(55.2%) well understood about CBHI, and the majority of them 222(74.5%) renewed their membership, whereas 242 (or 44.8%) did not understand the scheme well. Among them, 143(59.1%) decided to renew their membership. Among the 316 households (58.5%) who had a favorable attitude toward the CBHI scheme, 231 (73.1%) decided to renew their membership, while 224 (41.5%) had an unfavorable attitude toward the scheme, among whom 134 (59.82%) renewed their membership.

The majority, 344 (63.7%) of the households, did not trust in the CBHI scheme, and 150 (43.6%) of them did not renew their membership, but only 25 (7.9%) of the households that trusted in the CBHI scheme decided to drop their membership and dropped it. More than half, 297 (55%), of the households trusted in the contracted health facilities, and only 33 (11.1%) decided to drop their membership, whereas among those who had no trust in the facilities, 243 (49.5%), with the majority of them, 142 (58.4%), decided to drop their membership. More than half the households, 311 (57.6%) could not afford the premium, and 144 (46.3%) of them dropped their membership, whereas only 31 (13.5%) of the households for whom the premium was affordable decided to drop their membership.

About 251 (46.5%) of the households were members for more than 4 years, among whom only 63 (25.1%) decided to drop their membership, whereas 289 (53.5%) were members for 3 years, among whom 112 (30.8%) dropped their membership. More than half of the households, 304(56%) rated the quality of health services provided by contracted health facilities as good, and the majority of them 226 (74.3%) renewed their membership. The three-fourth of households, 324(60%) had the experience of leaving the contracted health facilities without getting treatments; among them, 138 (42.6%) decided to drop their membership (Table 4).

In this study, the factors affecting the decision-making process for membership renewal were identified (Table 5). Accordingly, households with fewer than five members were 54% less likely to renew their membership (AOR = 0.46, 95% CI = 0.25–0.86). The education level of households also had a significant influence on decision-making on renewal of membership; as a result, the households whose education level was only being able to read and write were 72% less likely to renew their membership compared to those who attended secondary school or above (AOR = 0.28, 95% CI = 0.12–0.64). On the other hand, households with a lower than middle level of wealth index had a higher likelihood than wealthy households of deciding to renew their membership. As a result, the extremely poor (AOR = 1.71, 95% CI =2.37–12.35), poor (AOR = 5.44, 95% CI = 1.9–32.53), and middle-level (AOR = 9.80, 95% CI = 2.75–34.92) households were more likely to renew their membership compared to the very rich households.

Those households who perceived the annual premium as affordable compared with the expected and promised benefits from the CBHI scheme were four times more likely to decide to renew their membership compared to those who perceived the premium as unaffordable (AOR = 4.34, 95% CI = 2.08–9.04). The households that experienced leaving the contracted health facilities without getting treatments because of the unavailability of diagnosis services and drugs in the facilities in the last 12 months were 74% less likely to renew their membership compared to their counterparts (AOR = 0.26, 95% CI = 0.12–0.55).

The trust among households in contracted health facilities was another significant factor. As a result, those households that trusted in the health facilities were almost six times more likely to renew their membership compared to those that had no trust in the facilities (AOR = 5.81, 95% =2.82–11.94). The households were asked their perceptions of the attitude of health care providers toward the scheme members during health care seeking and health service provision. Accordingly, those households that perceived the health care providers as having a favorable attitude toward the scheme members were eight times more likely to renew their membership, compared to their counterparts (AOR = 8.23, 95% CI = 3.96–19.64).

In addition, those households that perceived the quality of health services provided by the contracted health facilities as good were almost five times more likely to renew their membership compared to those who perceived the quality of health care was poor (AOR = 4.47, 95% CI = 2.28–8.85). Those households whose family members sought health care from the contracted health facilities in the past 12 months were three times more likely to decide to renew their membership compared to those households that did not (AOR = 3.25, 95% CI = 1.32–7.98).

The current study indicated that the membership dropout decision was substantial, which might significantly affect the population coverage rate of CBHI and challenge its sustainability. In addition, the study findings implied that households with a larger family size, lower than middle-level wealth status, and above the primary level education of the household heads, the affordability of the scheme premium to renew membership, the availability of the promised benefit packages of health services with good quality in contracted health care facilities, trusting in the contracted health care facilities, good attitude of health care provides towards the scheme members, and health care-seeking behavior when the households’ members experience health problems were the significant responsible factors that positively influenced the membership renewal decision among rural households to continue their CBHI scheme membership. As a result, the study findings also assist health planners and decision-makers in developing an appropriate plan and strategy to sustain the CBHI scheme in order to achieve UHC, with a focus on how to maintain membership after enrollment taking into account the identified determinant factors.

Despite the fact that the study well addressed the determinants of CBHI scheme membership renewal decision among rural households, it could have a recall and social desirability bias because the respondents were interviewed about the events that occurred in the past 12 months. Furthermore, the current study did not address the potential adverse selection bias that might impact the decision to renew scheme membership.