Associations of seated dual-task ergometer training with cognitive and mobility outcomes in older adults with disabling conditions undergoing inpatient convalescent rehabilitation: a retrospective observational study

Abstract

Objective To examine associations between a 4-week seated dual-task (DT) ergometer program and cognitive and mobility outcomes in older convalescent rehabilitation inpatients.Materials and methods This retrospective observational study included 125 rehabilitation episodes involving older adults undergoing inpatient convalescent rehabilitation , primarily for orthopaedic disorders and stroke. Participants received seated DT ergometer training (n = 79) or conventional recumbent ergometer exercise (n = 46) for 4 weeks. Primary outcomes were the Mini-Mental State Examination (MMSE), Trail Making Test–Japanese Part A (TMT-J Part A), 10-m DT walking time, and the Timed Up and Go test (TUG). Analysis of covariance models adjusted post-training outcomes for baseline values, age, and diagnosis using HC3 robust standard errors.Results Adjusted between-group differences (DT minus conventional exercise) were nonsignificant for MMSE (0.08 points; 95% confidence interval [CI], −0.78 to 0.93; p = 0.862), TMT-J Part A (−6.59 s; 95% CI, −22.06 to 8.87; p = 0.403), DT walking time (−0.89 s; 95% CI, −2.77 to 0.99; p = 0.354), and TUG (−0.97 s; 95% CI, −2.62 to 0.67; p = 0.246). The six-sessions-per-week sensitivity analysis yielded consistent results. Exploratory analyses showed greater improvements in DT walking time (p = 0.003) and TUG (p = 0.016) in the MMSE ≥27 stratum. Within standard TMT-J limits, adjusted Part A time was shorter in the DT group (p = 0.035), whereas the unadjusted change-score comparison was nonsignificant (p = 0.182).Conclusions Seated DT ergometer training did not demonstrate a clear advantage over conventional ergometer exercise in cognitive or mobility outcomes. Exploratory findings are hypothesis-generating. Prospective randomized studies with standardized training exposure are needed.

The authors' abstract, as published at the source. Annals of Medicine, 2026 · DOI ↗