A statistical approach to discriminate between non-fallers, rare fallers and frequent fallers in older adults based on posturographic data

•Principal component analysis is used to select features from postural parameters•Association between fall history and values of new parameters is analyzed•130 elderly non-fallers, infrequent fallers and frequent fallers are analyzed•Eyes open compliant surface results optimal task to identify non/infrequent fallers

BackgroundIdentification of future non-fallers, infrequent and frequent fallers among older people would permit focusing the delivery of prevention programs on selected individuals. Posturographic parameters have been proven to differentiate between non-fallers and frequent fallers, but not between the first group and infrequent fallers.MethodsIn this study, postural stability with eyes open and closed on both a firm and a compliant surface and while performing a cognitive task was assessed in a consecutive sample of 130 cognitively able elderly, mean age 77(7)years, categorized as non-fallers (N = 67), infrequent fallers (one/two falls, N = 45) and frequent fallers (more than two falls, N = 18) according to their last year fall history. Principal Component Analysis was used to select the most significant features from a set of 17posturographic parameters. Next, variables derived from principal component analysis were used to test, in each task, group differences between the three groups.FindingsOne parameter based on a combination of a set of Centre of Pressure anterior-posterior variables obtained from the eyes-open on a compliant surface task was statistically different among all groups, thus distinguishing infrequent fallers from both non-fallers (P < 0.05) and frequent fallers (P < 0.05).InterpretationFor the first time, a method based on posturographic data to retrospectively discriminate infrequent fallers was obtained. The joint use of both the eyes-open on a compliant surface condition and this new parameter could be used, in a future study, to improve the performance of protocols and to verify the ability of this method to identify new-fallers in elderly without cognitive impairment.