A combined stepping and visual tracking task predicts cognitive decline in older adults better than gait or visual tracking tasks alone: a prospective study.

It is unclear whether motor-cognitive dual tasks predict cognitive decline better than either motor or cognitive tasks alone. To examine the utility of the novel motor-cognitive dual-task test [Stepping Trail Making Test (S-TMT)], as a predictor of cognitive decline, and compare its predictive performance with single motor or cognitive tests. This 2-year population-based prospective study included 626 adults aged ≥ 70 years from Takashimadaira, Itabashi, Tokyo. The S-TMT measured the time taken to step on 16 numbers in order. Gait speed and TMT-A were assessed with standardized methods as single motor and cognitive tasks, respectively. A decline in the Mini-Mental State Examination score by ≥ 3 points over 2 years was defined as a significant cognitive decline. Over 2 years, 97 (15.5%) experienced cognitive decline. After adjusting for confounders, binary logistic regression models showed no significant associations between gait speed, TMT-A time tertiles, and risk of cognitive decline, but participants in the longest tertile of S-TMT time were more likely to develop cognitive decline than those of the shortest tertile (odds ratio 2.14; 95% confidence interval 1.17–3.90). Only the addition of the S-TMT time to the covariates model significantly improved the reclassification indices for predicting cognitive decline (net reclassification improvement: 0.31, P < 0.01; integrated discrimination improvement: 0.01, P = 0.02). Only the S-TMT was significantly associated with cognitive decline and improved reclassification indices, indicating that it is more useful for predicting cognitive decline than individual gait speed or visual tracking tests alone.