Establishing Causal Order of Spousal Caregiving and Depressive Symptoms

Benjamin D. Capistrant, Harvard School of Public Health
J. Robin Moon, Harvard School of Public Health
Maria Glymour, Harvard School of Public Health

We sought to explicate the temporal order in the causal association between spousal caregiving and depression. We followed U.S.-based Health and Retirement Study (HRS) respondents between 2000 and 2008 to examine: (a) predictors of spouses’ caregiving at baseline and during the follow-up period; (b) the association of caregiving demands and onset of depression; and (c) whether depression predicted spouses ending care. We defined caregiving as providing ≥14 hours of care for instrumental/activities of daily living/week to a spouse. We use discrete-time survival models with time-updated exposure and covariates, and inverse probability weights of treatment and survival in the latter two models. Spousal care is predicted by mostly socio-demographic, not health, characteristics; depression does not predict prevalence or incidence of caregiving. Spousal caregiving independently predicts a near doubling in risk of depression. There is suggestive evidence that depression is a predictor of ending spousal caregiving.