2013 年 12 巻 p. 9-15
This longitudinal study examines whether a novel dual-task walking test— “5-m walking test with serial subtraction” —more accurately predicts falls than the single-task habitual walking test does. Participants at baseline were 116 community-dwelling older adults (61 men, 55 women; mean age = 73.1 years, SD = 5.1, range: 65–86). We calculated time on the 5-m single-task habitual walking test (ST-time); time on the 5-m dual-task test, which involved counting backwards (DT-time); difference between DT- and ST-time (delta); and delta/number reached counting backwards. During a 1-year follow-up period, 19 subjects had one or more falls. Cox proportional hazard analyses adjusting for age, gender, and history of falls at baseline showed that subjects with below-median scores for DT-time (hazard ratio [HR] = 2.16, P = 0.14), delta (HR = 1.51, P = 0.39), and delta/number reached counting backwards (HR = 2.16, P = 0.12) tended to have a higher risk of falls than those with above-median scores for each. These HRs were, however, lower than that of ST-time (HR = 3.29, P = 0.07). Consequently, the novel test is probably not more useful than the single-task habitual walking test in predicting falls in the study population.