Methods: Data and samples: We pool 3 waves of data (2012, 2014, 2016) of the Health and Retirement Study, a nationally representative panel study of community-dwelling older adults and their spouses who are over the age of 50. Analyses are based on sample of 17,044 repeated observations on 6,636 unique primary respondents from these three waves.
Measures: We examined ADL limitations (Katz ADL scale), cognitive impairment (<12 on the TICS-27 scale), and depressive symptoms (8-item CES-D). We considered depression as a continuous (0-8 score) and binary (3 or more depressive symptoms) outcome. We constructed a categorical variable that represents those participants with no limitations, only cognitive impairment, only ADL limitations, and both limitations.
Analytic strategy: We estimate linear and logistic multivariate regression models and individual fixed effects models investigating the associations between ADL limitations, cognitive impairment, and depressive symptoms, controlling for a standard set of socioeconomic and health factors.
Results: We find that 66% of respondents report neither cognitive impairment nor ADL limitations, 16% report only cognitive impairment, 11% report only ADL limitations, and 7% report both types of limitations. Multivariate analyses indicate that both cognitive impairment and ADL limitations are associated with depression; however, across all models, ADL limitations have a much stronger association with both a continuous and binary measure of depression. Most importantly, when we examine the association between our constructed categorical variable of the unique type of limitation and depression, we find that although having both limitations is the strongest predictor of depression, having just ADL limitations and no cognitive impairment is nearly as strongly predictive of depression as having both limitations. These associations are robust across alternative specifications, including the addition of a measure of depression from the prior wave, individual fixed effects models, and adjusting for complex survey design weights.
Conclusions and implications: While we find that both cognitive impairment and ADL limitations have poor consequences for the mental health of older adults, these results point to the importance of ADL limitations regardless of cognitive status. In next steps, we will delve deeper into these results to understand what other factors may be driving the association between ADL limitations and depression and explore heterogeneity in associations across relevant groups.