Estimating long-term health effects of housing instability in the US using a lifecourse framework

The United States (US) is facing a housing affordability crisis. Over 30% of adults experience a high housing cost burden by formal economic definitions, and 69% report being “very concerned” about the cost of housing. As a result, many people in the US experience unstable housing, which includes frequent or involuntary residential moves and homelessness. Current evidence suggests that housing instability is harmful to health: it is associated with adverse health outcomes across multiple lifecourse periods, and select, short-term interventions targeting housing instability appear to improve health outcomes. To date, limited work has examined its long-term health effects using a lifecourse framework. In this dissertation, I estimated long-term health effects of housing instability in the US using a lifecourse framework and used causal inference methods to evaluate common sources of bias in lifecourse epidemiology, including selection bias and time-varying confounding. Further, I examined potential sources of heterogeneity and specified policy-relevant estimands, strengthening the real-world relevance of the findings. In Aim 1, I estimated the effect of childhood housing instability, measured by experiencing a residential move due to financial difficulties before age 16 years, on mortality among Health and Retirement Study participants aged (i) 50-60 years and (ii) >60 years at baseline using inverse probability of treatment-weighted Kaplan Meier curves. Findings indicated small, long-term effects of childhood housing instability on mortality among participants aged 50-60 years at baseline, but not among participants aged >60 years at baseline. In Aim 2, I examined the relationship between childhood housing instability and functional impairment trajectories among Health and Retirement Study participants aged 50, 60, and 70 years at baseline under a hypothetical scenario in which everyone in the sample remained in the study through the end of follow-up. Effect estimates were obtained using generalized estimating equations with weights to account for early-life confounding, censoring due to loss to follow-up, and censoring due to death. Findings from this aim suggest small effects of childhood housing instability on functional impairment in mid-life that diminished in late-life after accounting for early-life confounding and selection due to loss to follow-up and death. In Aim 3, I estimated the effect of a hypothetical intervention eliminating involuntary residential moves compared with the natural course of involuntary residential moves on mortality over 50 years (ages 0-5 years through mid-life) in a national US cohort study using the parametric g-formula. Results from this aim suggested that the hypothetical intervention eliminating involuntary residential moves had a small, long-term protective effect on mortality, resulting in a 2% reduction in deaths in mid-life (50 years of follow-up) had everyone remained in the study throughout follow-up. Across all 3 aims, I found evidence of small, long-term effects of housing instability on morbidity and mortality in mid- and late-life in national US cohort studies.