A double-blind study was undertaken to determine the relationship (if one exits) between the extent of postural sway in standing of individual elderly subjects and their frequency of falling. A total of 205 subjects were studied; their average age was 81.8 years. Thirty per cent of the men, and 46% of the women, had one or more falls; the proportion of all subjects with one or more falls was 42%. The average speed of sway was significantly greater (P less than 0.05) for those who fell one or more times in a year than for those who did not fall. In this group of institutionalized elderly, there was no sex-related difference in the mean speed of sway; moreover, no age-related trend was demonstrated by the regression of mean speed upon age for all subjects. The mean speed of sway even for the non-fallers was found to be greater than that measured in a sample of non-institutional elderly subjects studied in the past. Thus, mean speed of postural sway was found to be only of statistical value for determining the risk of falling among these institutionalized elderly. We found postural sway to be an indicator of a tendency to fall, but the difference was less than might have been expected. No trend of increasing postural sway correlating with the increased frequency of falls was found.

Falls among the elderly, although associated with substantial morbidity and mortality, have not been previously the subject of a longitudinal study. This five-year prospective study of an active ambulatory institutionalized population over 65 years of age, revealed an annual fall rate of 668 incidents per 1000, with a rising frequency in successive age groups after age 75. Forty-five per cent of all subjects suffered at least one fall during the study period. Women had a higher fall rate in all age groups and the severity of their injuries appeared to increase with age. Injuries severe enough to be brought ordinarily to the immediate attention of a physician occurred in 17.5% of falls. Among subjects who ultimately died after suffering many falls, there appeared in some to be a 'clustering' of falls prior to death.

Patients using antidepressants should be followed closely because the risk of falls is increased. Previously reported relationships between benzodiazepines and diuretics and falls are not supported by the present findings. Clinical detection of orthostatic hypotension is unlikely to be useful in predicting future risk of falling.