Analyzing data from a 1998 household survey in Burundi, this dissertation measures the population-wide effects of protracted conflict on child mortality, physical illness and mental health status, as well as assesses the roles of household economic status and health service availability. Individual reports on the effect of the crisis on life are used as an indicator for the impact of the conflict. Bivariate associations and multivariate models are estimated, including two- and three-level models that account for random variation in the outcomes at the household, cluster and province levels.

Mothers who report that the crisis greatly affected life are 1.20 times more likely to have experienced an additional child death (p = 0.01), controlling for other observable factors. Individuals greatly affected by the crisis experience 1.43 times greater odds of physical illness (p < 0.001), and are 1.55 times more likely to suffer poorer mental health status (p < 0.001). One quintile increase in economic status reduces risk of an additional child death by 0.92 times in urban areas (p = 0.04), with no effect in rural areas. Economic status is not protective with regard to the risk of physical illness. In urban areas, a one quintile increase in economic status reduces risk of a poorer score on a mental health indicator by 0.91 times (p < 0.001), while this effect is less clear in rural areas. In urban areas, one hour additional walking time from the nearest health facility is associated with 1.20 times greater risk of an additional child death (p = 0.02), 1.65 times greater odds of physical illness (p < 0.001), and 1.13 times greater risk of a poorer mental health score (p = 0.01). No such effects are observed in rural areas.

Given that 57% of the population consider themselves to be greatly affected by the crisis, the analysis emphasizes how the physical and mental health of entire populations can be affected by protracted conflict.