Abstract
Falls among community-dwelling older adults can be life threatening. While an association between social isolation and falls has been described, the nature of that relationship is not well documented and could be important for fall prevention interventions. Study objectives were to describe the incidence of falls, prevalence of social isolation, and extent to which social isolation predicts falls in older adults. Secondary analysis of longitudinal data from the National Health and Aging Trends Study, involving a nationally representative sample of Medicare beneficiaries, included four rounds of annual interviews in participants’ homes (round one n=7,609). Social isolation was operationalized for the current analysis as a multiple-indicator, domain-inclusive construct based upon the Social Network Index. Falling during the previous year was self-reported. Incidence of falls ranged from 22.4–26.2% across the four rounds. Social isolation prevalence ranged from 19.8–21.9%. The probability of falling increased with each increase in social isolation construct score. Even after adjusting for age, gender, and education, social isolation significantly predicted falling (OR=1.08; CI=1.02–1.14). Adding self-reported general health, depression, and worry about falling to the model weakened the relationship between social isolation and falls (OR=1.02; CI=0.96–1.08). Adding the Short Physical Performance Battery, assistive mobility device use, and activities of daily living to the model weakened the relationship further (OR=0.99; CI=0.94 –1.04). Social isolation as a predictor of falls is partially explained by the strong relationship between social isolation and physical performance. Fall-prevention interventions targeting social isolation could have an important impact on physical performance and future falls.