Psychocognitive vulnerability and risk of incident ADL disability and mortality across six prospective cohort studies

Depressive symptoms and lower word-recall performance often co-occur with chronic disease in middle-aged and older adults. We examined their separate and joint associations with subsequent activities of daily living (ADL) disability and mortality across multimorbidity strata.

We pooled harmonised data from CHARLS, ELSA, HRS, KLoSA, MHAS and SHARE. Adults ≥45 years contributed observed person-intervals. Study-wave standardised scores defined none, single-domain or dual-domain vulnerability. Outcomes were incident ADL disability (primary) and all-cause mortality. Modified Poisson models estimated risk ratios (RRs); cohort-stratified Cox models estimated hazard ratios (HRs).

Among 257,922 participants (mean age 62.4 years; 55.5% female), 181,551 contributed 512,522 eligible ADL intervals (45,430 events). Complete-case adjusted analyses included 179,248 participants (500,911 intervals; 44,178 ADL events) and 213,268 participants (605,704 mortality windows; 27,782 deaths). ADL RRs versus none were 1.68 (95% CI, 1.65-1.71) for single-domain and 2.19 (2.12-2.27) for dual-domain vulnerability, with 48.7 and 85.3 excess events per 1000 standardised interval observations, respectively. Corresponding summary mortality HRs were 1.56 (95% CI, 1.51-1.60) and 1.79 (1.71-1.86). Interaction-model absolute ADL risk was 27.2% with dual-domain vulnerability and ≥ 3 chronic conditions at the next observed assessment. Four categories distinguished endpoint-specific single-domain associations; joint ADL associations were below multiplicative and additive expectations (interaction ratio 0.756; relative excess risk due to interaction [RERI] -0.240). Mortality hazards were nonproportional.

A simple psychocognitive vulnerability classification was associated with graded risks of ADL disability and mortality across six cohorts. The largest absolute ADL risk increases accompanied multimorbidity.