THE ECONOMICS OF NON-COMMUNICABLE DISEASES IN RURAL BANGLADESH
Mirelman, Andrew J.
MetadataShow full item record
Background: In Matlab, Bangladesh, a rural sub-district with ongoing demographic surveillance, an epidemiological transition is well under way with an emerging burden of disease attributed to non-communicable diseases (NCD). In this setting there is a need to understand NCDs in terms of socioeconomic determinants and economic impacts to individuals and households, which helps inform the decision to develop NCD-related public policies. This work addresses these issues by characterizing the education gradient in mortality over a period of 24 years and by evaluating the household-level economic impacts after an adult NCD death and subsequent coping strategies. Methods: Paper #1 uses data from the routine Matlab surveillance system for the populations in 1982, 1996 and 2005, looking prospectively at both NCDs and infectious disease mortality over a five year follow-up period. Cox proportional hazard models are used for multivariate analysis to assess the education gradient in mortality for each broad cause of death category and to what extent components of wealth, occupation and marital status contribute to this gradient. In papers #2 and #3, all of the adult NCD deaths in 2010 in Matlab were identified and directly matched to a comparison group of households with no deaths. A regression standardization approach is used in Paper #2 to obtain a marginal estimate of the relative risk of a household being poor after an NCD death in terms of an asset-based wealth index, self-rated economic condition and land ownership. Paper #3 examines the coping strategies that households use after an NCD death. Logistic regression is then used to look at household and individual-level characteristics related to coping and an econometric difference-in-difference (DiD) approach is used to examine changes in household composition. Results: Paper #1 finds a larger education gradient for females for both NCD and infectious disease mortality when the data is pooled for all three time periods. For both males and females, a larger gradient is also found for infectious diseases. Marital status of an individual explains more of the education gradient in mortality than occupational status or household wealth. Paper #2 shows that there is a 14-19% increased risk of a household being poor two years after an adult NCD death, depending on which measure of economic status is used. Individual characteristics of a male death, prime age death or death of a married household member leads to a higher risk of a household being poor. In Paper #3, the results for coping enriches this picture further. The most common coping strategy among households after the death was the reduction of spending on basic household items. A prime age death is positively associated with the most number of coping strategies, four, and there is evidence that poorer households have more limited coping options. The DiD results for household composition show that households moderately replace human capital in terms of recruiting new adults to the household and that households are more likely to recruit adult females after a prime age death. Conclusion: The rising NCD burden in low income countries means that more understanding of the economic impacts of these diseases is needed. Using census data in the demographic surveillance system in rural Matlab, Bangladesh, this dissertation explores the individuals and household economic impacts associated with NCD deaths. An NCD death has the potential to impose severe economic consequences for households, impacting household wealth and propagating a poverty trap where poor household are not able to make gains in terms of economic mobility. Health shocks from NCDs lead to coping strategies that may have long term negative consequences for households. This appears to be especially true when the death is to a prime, working-age household member. Policy Recommendation: This work emphasizes the need for more intense prevention efforts for NCDs in rural Bangladesh. In Matlab, where this study is based, there have been intensive efforts to reduce maternal and childhood mortality that has been documented as a global success story. This work shows that there are important distributional and efficiency concerns related to NCD health that should motivate more public intervention. In terms of equity, there are higher rates of mortality among the least educated and health gains can be made with continuing progress in rural education and access to social psychological resources. As well, there may be longer term costs to household members, in terms of a poverty trap, when there is an NCD death. Better access to financial protection resources and preventive care is needed. This is especially true for households that are at risk of having a premature adult death. The review of the economic impacts from NCDs that are provided from this work provide an argument for developing more NCD-related policies in rural Bangladesh.