Griffin J, Lall R, Bruce J, Withers E, Finnegan S, Lamb SE. J. Clin. Epidemiol. 2019; 106: 32-40.
Affiliation
Warwick Clinical Trials Unit, Warwick Medical School, University of Warwick, Coventry, CV4 7AL, UK; Kadoorie Centre for Critical Care Research and Education, John Radcliffe Hospital, NDORMS, University of Oxford.
BACKGROUND: Prospective, monthly diaries are recommended for collecting falls data but are burdensome and expensive.
OBJECTIVE: To compare characteristics of fallers and estimates of fall rates by method of data collection.
DESIGN: and Setting: A methodology study nested within a large cluster RCT. We randomised 9803 older adults from 63 general practices across England to receive one of three fall prevention interventions. Participants provided a retrospective report of falls in postal questionnaires mailed every four months. A separate randomisation allocated participants to receive prospective monthly falls diaries for one simultaneous four month period.
RESULTS: Falls diaries were returned by 7762/9375 (83%); of which 6306/9375 (67%) participants reported the same number of falls on both data sources. Diary non-responders were older and had poorer levels of physical and mental health. Analysis of time-points where both data sources were available showed the falls rate on diaries was consistently higher than on the questionnaire (mean rate: 0.16 v 0.12 falls per person month observation). Diary allocation was associated with a higher rate of withdrawal from the main trial.
CONCLUSIONS: Diary completion was associated with sample attrition. We found on average a 32% difference in falls rates between the two data sources. Retrospective and prospective falls data are not consistently reported when collected simultaneously.