To address the external validity of a trial of multifactorial fall prevention through an analysis of differences between participants and nonparticipants regarding socioeconomic and morbidity variables.
Analysis of nonresponse in a randomized clinical trial.
Geriatric outpatient department.
One thousand one hundred five community-dwelling adults aged 65 and older who had sustained at least one injurious fall.
Marital status, housing tenure, income, comorbidity, hospitalization, fractures, and drug use before invitation to participate in the trial. Fractures, hospitalization and death were measured for 6 months of follow-up.
Four hundred forty-seven responding nonparticipants and 266 nonresponding nonparticipants were compared with 392 participants in the trial. Lower income (odds ratio (OR)=2.38, 95% confidence interval (CI)=1.28-4.28) and more days of hospitalization during the previous 5 years (OR=1.96, 95% CI=1.15-3.33) predicted responding nonparticipation; independent predictors of being a nonresponding nonparticipant were unmarried status (OR=2.0, 95% CI=1.36-2.94), lower income (OR=4.74, 95% CI=2.30-9.78), more days of hospitalization (OR=3.49, 95% CI=1.99-6.11), and prior fractures (OR=1.56, 95% CI=1.02-2.38). Nonresponding nonparticipants were significantly more likely to die (OR=12.99, 95% CI=1.6-105.6) or be hospitalized (OR=2.66, 95% CI=1.7-4.1) than participants during 6 months of follow-up.
Nonresponding nonparticipants of a trial of multifactorial fall prevention differed significantly from participants in terms of socioeconomic and morbidity variables and were more likely to be hospitalized or die during 6 months of follow-up. Because of the differences between the two populations, it is questionable whether results from this randomized trial can be generalized to people potentially eligible for participation.