Feasibility and accuracy of fall reports in persons with dementia: a prospective observational study

The feasibility, test-retest reliability, and accuracy of different fall recording methods have not been studied in older persons with dementia.

This was a prospective observational study, nested within a randomized controlled trial on motor training, in which 110 participants were monitored for falls over 12 months. Seven methods of fall recording were compared: face-to-face interviews; phone interviews: weekly for three months, monthly for 9 months, a final interview after 12 months; prospective calendar method; interviews with a proxy and the general practitioner (GP). Summing the count of falls and removing duplicate reporting of the same fall was found to provide the best approximation of the actual number of falls and was chosen as the criterion-standard.

The combination of calendar method and phone interviews showed the highest accuracy (74% of falls, 93% of fallers). As a single measure, weekly phone calls were superior to calendars or proxy-report. Monthly phone calls recorded only half the falls that were picked up by weekly calls (p = 0.002) and were inferior to the calendars (p<0.001) and proxy-report (p = 0.015). GPs knew of only 14% of falls and 19% of fallers. In addition, 49% of subjects who documented a fall prospectively did not recall a fall after 12 months.

The combination of fall calendars with regular telephone interviews can be recommended for persons with mild to moderate stage dementia. If feasible, recall periods should be as short as one week; additional information by care-givers increases accuracy of reports. Retrospective recall of falling with long recall periods is not recommended.