Prevalence and factors associated with catastrophic health expenditure among slum and non-slum dwellers undergoing emergency surgery in a metropolitan area of South Western Nigeria

Out of Pocket (OOP) payment continues to persist as the major mode of payment for healthcare in Nigeria despite the introduction of the National Health Insurance Scheme (NHIS). Although the burden of health expenditure has been examined in some populations, the impact of OOP among slum dwellers in Nigeria when undergoing emergencies, is under-researched. This study sought to examine the prevalence, factors and predictors of catastrophic health expenditure amongst selected slum and non-slum communities undergoing emergency surgery in Southwestern Nigeria. The study utilised a descriptive cross-sectional survey design to recruit 450 households through a multistage sampling technique. Data were collected using pre-tested semi-structured questionnaires in 2017. Factors considered for analysis relating to the payer were age, sex, relationship of payer to patient, educational status, marital status, ethnicity, occupation, income and health insurance coverage. Variables factored into analysis for the patient were indication for surgery, grade of hospital, and type of hospital. Households were classified as incurring catastrophic health expenditure (CHE), if their OOP expenditure exceeded 5% of payers' household budget. Analysis of the data took into account the multistage sampling design. Overall, 65.6% (95% CI: 55.6-74.5) of the total population that were admitted for emergency surgery, experienced catastrophic expenditure. The prevalence of catastrophic expenditure at 5% threshold, among the population scheduled for emergency surgeries, was significantly higher for slum dwellers (74.1%) than for non-slum dwellers (47.7%) (F = 8.59; p = 0.019). Multiple logistic regression models revealed the significant independent factors of catastrophic expenditure at the 5% CHE threshold to include setting of the payer (whether slum or non-slum dweller) (p = 0.019), and health insurance coverage of the payer (p = 0.012). Other variables were nonetheless significant in the bivariate analysis were age of the payer (p = 0.017), income (p