Dual Task–Related Changes in Gait Performance in Older Adults: A New Way of Predicting Recurrent Falls?

To the Editor: Recently, Faulkner et al.1 reported the results of an interesting study in which the relationship between dual task–related changes in performance and a history of recurrent falls (>2 falls in the year preceding gait assessment) was studied. Based on cross-sectional study, walking time during usual walking and while performing a visuo-spatial reaction-time response was measured in 377 older adults (mean age 78±3). The occurrence of falls within the 12 months preceded the assessment was collected using record review. Subjects were separated into two groups: those who had never fallen or had a history of one fall, and those who reported two or more falls (recurrent fallers). The main result was that poor walking-time response due to visuo-spatial task and characterized by increased reaction time was significantly associated with high odds of recurrent falls history. The authors concluded that changes in performance while dual-tasking could be used to identify subjects at risk for recurrent falls and, as a consequence, require prospective studies to confirm the predictive value of poor dual-task performance for recurrent falls. Few studies have explored the predictive value for risk of falling of dual task–related changes in gait performance. Most have focused on prediction of the first fall,2–5 and only one has explored the predictive value of dual-task related gait changes for recurrent falls using a prospective cohort study design.6 The walking-associated task was a simple mental calculation task. The dual-task performance was measured in 311 community-dwelling subjects aged 70 and older. No significant relationship between poor dual-task performance and recurrent falls was established. Therefore, similar to the data obtained for the occurrence of one single fall, the predictive value for recurrent falls of dual task–related gait changes is doubtful. Since 1997 and the first study by Lundin-Olsson et al.,2 who showed that stopping walking when talking was a predictor of falls in elderly people, changes in gait characteristics due to a simultaneously performed attention-demanding task have frequently been reported in elderly subjects7 but only occasionally associated with risk of falling.2–6 Most of these subsequent studies used the occurrence of single fall as the main outcome.2–5 The originality of the study by Faulkner et al.1 was to focus on recurrent falls. The authors explained their choice by stating that recurrent falls seemed to be more frequently associated with subject-related risk factors for falling than isolated falls. Because dual task–related gait changes were closely correlated with subject-related risk factors for falling,8 the hypothesis that changes in gait performance while dual-tasking could be related to recurrent falls is justified. Therefore, Faulkner's suggestion to confirm the predictive value of dual task–related decreases in performance for recurrent falls requires attention. Before launching new studies, the reasons for what may be considered to be a failure of the predictive value of dual task for nonrecurrent falls should be understood. Two recent publications have indicated that changes in the performance of an attention-demanding task and gait variability under dual-tasking conditions may be helpful for the prediction of falls. All previous studies except one have focused on dual task–related gait changes. In that study, changes in counting performance score were strongly related to the occurrence of a first fall (adjusted odds ratio=53.3, P<.001) during the 12-month follow-up of 187 subjects aged 75 to 100 (mean age 84.8±5.2).9 Recently, Kressig et al.10 showed that greater stride-to-stride variability while dual-tasking was also predictive of the occurrence of a first fall in 57 geriatric inpatients (mean age 85.0±6.6). Both these results constitute a relevant new orientation of research in the detection of the risk of falling based on dual-tasking and should be used as a support of prediction of recurrent falls in further research. Conflict of Interest: Beauchet, Annweiler, Allali, Berrut, and Dubost have no relevant financial interest in this letter. Author Contributions: Beauchet, Annweiler, Allali, and Berrut: drafting manuscript. Dubost: critical revision of the letter for important intellectual content. Sponsor's Role: There was no sponsor for this letter.