Supplementary Material for: Cognitive performance in late midlife as a predictor of frailty from late midlife into old age: a longitudinal birth cohort study

Abstract
Background: Cognitive decline is cross-sectionally associated with frailty in older adults. However, the longitudinal associations between cognitive performance in late midlife and subsequent frailty are less well understood.
Methods: Participants of the Helsinki Birth Cohort Study (n=1279) were assessed for cognitive performance in late midlife and a 41-item frailty index on up to three occasions from late midlife into old age. The CogState assessment assessed reaction time, response accuracy, choice reaction time, working memory, divided attention, and associated learning. Generalized linear mixed models estimated the association between baseline cognitive performance and the rate of change in frailty and were adjusted for sex, education, and smoking.
Results: Each standard deviation slower reaction time was associated with a 1.02-1.07-fold higher level of frailty in late midlife in all domains except for simple reaction time. Each standard deviation higher response accuracy in choice reaction time was associated with a lower level of frailty (0.95-fold) in late midlife; better working memory response accuracy was associated with lower levels of frailty (0.94-fold) in women only. Longitudinally, each standard deviation higher reaction time response accuracy was associated with a slower increase in frailty from late midlife into old age; the annual rate of change was 0.97-fold. Furthermore, each standard deviation higher divided attention response accuracy was associated with an equally slower annual increase in frailty among men only.
Conclusion: Poorer cognitive performance is associated with higher levels of frailty in late midlife. The association between cognitive performance and frailty from late midlife into old age was less consistent and varied by the cognitive trait being measured.