Intrinsic capacity transitions predict overall and cause-specific mortality, incident disability, and healthcare...

To develop an intrinsic capacity (IC) score and to investigate the association between IC transition with overall and cause-specific mortality, incident disability and healthcare utilization.

Retrospective cohort study SETTING AND PARTICIPANTS: Data from 1852 respondents aged ≥ 65 years who completed the 1999 and 2003 surveys of the Taiwan Longitudinal Study on Aging were analyzed.

Transitions of IC score were categorized into three groups: (1) Improved IC (IC_2003-1999 >0), (2) Stable IC (IC_2003-1999 = 0), (3) Worsened IC (IC_2003-1999 <0). Cox regression and subdistribution hazard models were used to investigate IC transitions and 4-year overall and cause-specific mortality, respectively. Logistic regression were employed to develop weighted IC score (wIC, 0-16) and assess its association with incident disability and healthcare utilization. Similar analysis were repeated using non-weighted IC (nIC, 0-8) to ensure robustness.

Comparing to decreased wIC group, stable or increased wIC participants had significantly lower 4-year all-cause mortality, and death from infection, cardiometabolic/cerebrovascular diseases, organ failure and other causes. (Hazard ratio (HR) ranged from 0.36 to 0.56, 95% CI ranged from 0.15 to 1.00, p ≤ 0.049 in the stable wIC group; HR ranged from 0.41 to 0.51, 95% CI ranged from 0.22 to 0.94, p ≤ 0.034 in the increased wIC group). Moreover, individuals with stable or increased wIC demonstrated lower risk of incident disability and hospitalization. (Odds ratio (OR) = ranged from 0.34 to 0.70, 95% CI ranged from 0.19 to 1.00, p ≤ 0.048). Participants with stable wIC also exhibited reduced risk of emergency department visits (OR = 0.58, 95% CI = 0.41 to 0.82, p = 0.002). These results were generally consistent in the nIC model.

Participants with stable or increased IC experienced significantly lower all-cause and most cause-specific mortality, incident disability, and healthcare utilization, which was independent of baseline IC and comorbidities. The findings remained consistent across weighted and non-weighted IC model.