Health Insurance Participation and Household Out-Of-Pocket Healthcare Expenditure in Southeastern Nigeria: A Two-Part Analysis
Abstract
Background: Health insurance is expected to reduce households’ direct healthcare payments, yet its association with whether payments occur may differ from its association with the amounts paid. This study examined both dimensions of out-of-pocket healthcare expenditure in Southeastern Nigeria. Methods: Cross-sectional survey data collected in 2026 from 750 households across Abia, Anambra, Ebonyi, Enugu and Imo States were analysed. Insurance participation was defined as coverage of at least one household member. Descriptive comparisons and a two-part econometric model were employed. Logistic regression estimated the probability of positive expenditure, while a Gamma generalised linear model with a log link estimated expenditure among 592 positive spenders. Models adjusted for household demographic, socioeconomic, health-related and geographical characteristics. Standardised predictions combined both expenditure components. Results: Of the surveyed households, 294 (39.2%) reported insurance participation and 158 (21.1%) reported zero expenditure. Mean monthly-equivalent expenditure was ₦3,689.31 among insured households and ₦7,976.24 among uninsured households. Insurance participation was not significantly associated with positive-payment occurrence (β = −0.062; p = 0.756), but was negatively associated with expenditure intensity (β = −0.745; p < 0.001). Adjusted conditional expenditure was approximately 52.5% lower among insured positive spenders. Model-standardised expected expenditure was ₦3,724.08 under insured status and ₦7,938.38 under uninsured status, an adjusted difference of −₦4,214.30. The 95% bootstrap confidence interval also excluded zero, supporting statistical significance overall. Conclusion: Insurance participation was associated primarily with lower expenditure intensity rather than a lower probability of payment. Health financing evaluations should examine both dimensions alongside healthcare utilisation and unmet need. Because the analysis is cross-sectional, these findings establish associations, not causal effects.