To determine the relationship between falls and deficits in specific cognitive domains in older adults.

An analysis of the English Longitudinal Study of Ageing (ELSA) cohort.

United Kingdom community-based.

5197 community-dwelling older adults recruited to a prospective longitudinal cohort study.

Data on the occurrence of falls and number of falls, which occurred during a 12-month follow-up period, were assessed against the specific cognitive domains of memory, numeracy skills, and executive function. Binomial logistic regression was performed to evaluate the association between each cognitive domain and the dichotomous outcome of falls in the preceding 12 months using unadjusted and adjusted models.

Of the 5197 participants included in the analysis, 1308 (25%) reported a fall in the preceding 12 months. There was no significant association between the occurrence of a fall and specific forms of cognitive dysfunction after adjusting for self-reported hearing, self-reported eyesight, and functional performance. After adjustment, only orientation (odds ratio [OR]: 0.80; 95% confidence intervals [CI]: 0.65–0.98, p = 0.03) and verbal fluency (adjusted OR: 0.98; 95% CI: 0.96–1.00; p = 0.05) remained significant for predicting recurrent falls.

The cognitive phenotype rather than cognitive impairment per se may predict future falls in those presenting with more than one fall.