Financial burden of noncommunicable diseases on households in Bangladesh: a quasi-experimental analysis using national survey data

Noncommunicable diseases (NCDs) are a rapidly emerging health challenge in low- and middle-income countries, including Bangladesh, where research into their household financial burden remains limited and predominantly descriptive. Existing studies lack robust adjustment for confounders, creating uncertainty about differences in financial outcomes between NCD-affected and non-NCD households. To address this gap, this study uses a quasi-experimental analytic framework and repeated cross-sectional data to estimate adjusted differences in household financial outcomes over time. We analysed data from the Bangladesh Household Income and Expenditure Surveys (HIES 2005, 2010, and 2016), employing entropy balancing (EB) and propensity score matching (PSM) to estimate the average treatment effect on the treated (ATT) while adjusting for confounders. Households with at least one individual with an NCD (treatment group) were compared to similar households without any NCDs (control group) in terms of out-of-pocket (OOP) health expenditures, catastrophic health expenditure (CHE), impoverishment effects, and forgone care for financial constraints. ATT estimates were further stratified by economic status and geographic location to assess disparities. EB-based ATT estimates show that NCD-affected households incurred significantly higher (p ≤ 0.01) OOP expenditures than similar households without NCDs, with larger differences in later study periods (2005: USD39.98, 2010: USD62.59, 2016: USD68.78). Approximately 75–80% of these higher expenses were due to differences in medicine expenditure (2005: USD29.63, 2010: USD50.40, 2016: USD54.48; p ≤ 0.01). NCD-affected households also experienced significantly higher (p ≤ 0.01) incidences of CHE regardless of measurement methods, with the difference range becoming larger in successive study periods (2005: 0.98–3.99%-point, 2010: 0.99–4.14%-point, 2016: 1.47–8.96%-point), and consistently higher impoverishment (2005: 0.72%-point, 2010: 0.48%-point, 2016: 0.69%-point). In 2016, further impoverishment (0.53%-point, p ≤ 0.01) and risk of impoverishment (0.66%-point, p ≤ 0.05) were also significantly higher among NCD households. Subgroup analyses reveal that although adjusted differences in OOP expenditures were much smaller for the lowest than the highest quintile households (USD18.60 vs. USD151.03 in 2016) and among rural than urban households (USD53.76 vs. USD115.01 in 2016), NCD-associated financial burdens were more severe among the lowest quintile (e.g., in 2016, impoverishment: 3.82%-point p ≤ 0.01 vs. < 0.01%-point, p > 0.05) and rural households (e.g., in 2016, impoverishment: 0.84%-point p ≤ 0.01 vs. 0.25%-point, p > 0.05). The PSM-based results closely mirror the EB-based findings. From 2005 to 2016, NCD-affected households in Bangladesh, particularly low-income and rural households, consistently incurred higher financial burdens than comparable non-NCD households, with disparities generally more pronounced in the later years. These findings underscore targeted strategies and interventions to alleviate the financial burdens on NCD-affected households.