Investigating membership attrition dynamics in community-based health insurance: a survival analysis of socioeconomic and program-specific determinants in the Amhara Region, Northwest Ethiopia

Abstract

Background: Despite substantial growth and increased enrollment in Ethiopia's CBHI program, achieving universal coverage and retaining members remain challenging. This study, however, focuses on the persistent issue of dropout rates, which threaten the program's sustainability, while previous research has often focused on enrollment.

Method: The dependent variable is"time to membership attrition,"defined as an event "Dropout,"with independent variables including socioeconomic and program factors. Using Cochran's formula, data were collected from 772 (208 failure) respondents across five administrative zones. Analysis was performed using Kaplan-Meier estimation, Cox Regression, and the Weibull AFT Model.

Result: Dropout rates peaked at an average membership duration of 4.09 years, with increasing hazard rates (Weibull shape parameter = 2.077, p < 0.001). The Kaplan-Meier analysis indicates safety net beneficiaries had a lower dropout rate (67.8%) than non-beneficiaries (76.6%) and longer survival (Chi-square = 4.083, p = 0.043). Respondents with 4-6 hectares had the shortest survival (5.88 years) and a higher dropout risk (B = 0.417, p = 0.042), while non-landowners had a higher attrition risk (HR = 1.266, p = 0.814), Farm owners had a lower dropout rate (70.7%) (Chi-square = 1.569, p = 0.021). Lower-middle-income members had a higher attrition risk (HR = 1.999, p = 0.042) with a mean survival of 6.28 years, compared to 6.47 years for upper-middle-income. Perceived healthcare quality influenced dropout risk, increasing it by 5.9% for fair quality than poor (HR = 3.368, p < 0.000), and significantly for good quality as well (HR = 2.284, p < 0.000). Moderate financial protection (not catastrophic spending) increased dropout risk by 7.3% compared to high protection (HR = 1.125, p = 0.040).

Conclusion: Dropout rates peak in the early years of membership and increase over time. Enrolling in safety nets and having smaller land sizes enhanced retention rates, while extensive landholdings and being classified as Lower-Middle Income led to higher dropout rates; however, ownership of modern amenities had minimal impact on retention. Perceived healthcare quality and financial protection significantly influence CBHI retention, while program service quality has little effect, underscoring the need for policies that prioritize improving service quality, accessibility, and affordability.

Keywords: Cox regression; Kaplan–Meier method; Survival analysis; Weibull AFT model.

Fig. 1

Cox regression and Kaplan–Meier proportional hazards for respondent age

Fig. 2

Cox regression and Kaplan–Meier proportional hazards for SafetyNet membership

Fig. 3

Cox regression and Kaplan–Meier proportional hazards for hectare amount

Fig. 4

Cox regression and Kaplan–Meier proportional hazards for total annual income

Fig. 5

Cox regression and Kaplan–Meier proportional hazards for perceived healthcare quality

Fig. 6

Cox regression and Kaplan–Meier proportional hazards for program service quality

Fig. 7

Cox regression and Kaplan–Meier proportional hazards for financial protection from uncertainty

Fig. 8

Weibull regression for respondents' ages

Fig. 9

Weibull regression for safety net membership

Fig. 10

Weibull regression for land size in hectare

Fig. 11

Weibull regression for total annual income

Fig. 12

Weibull regression for perceived healthcare quality

Fig. 13

Weibull regression for program service quality

Fig. 14

Weibull regression for financial protection from uncertainty