Selection Bias and Psychosocial Mediators Do Not Fully Explain Caregivers’ Decreased Mortality Risk

Benjamin D. Capistrant, University of Minnesota

A few recent studies have shown spousal caregiving to be associated with a decreased mortality risk. However, the main hypothesized mechanisms to explain this association (healthy worker selection bias into caregiving roles and psychosocial benefits of altruism) have not been well tested empirically. I assessed the association between spousal caregiving (>=14 care hours/week) and all-cause mortality using data from the Health and Retirement Study between 2000-2012. I used a stabilized inverse-probability-weighted marginal structural model, specifically a logistic model, to estimate the odds of caregiving on morality. After accounting for both healthy worker selection bias and mediation by altruism, the controlled direct effect of spousal caregiving was a significant, reduced odds of mortality; this was true after accounting for psychosocial mediators, for long-term caregivers, and robust to a sensitivity analysis. Neither health worker selection bias nor altruism mediation can fully account for spousal caregiving being associated with lower odds of mortality.