Non-equivalent, but still valid: Establishing the construct validity of a consumer fitness tracker in persons with multiple sclerosis
Fig 3
Manual and device-derived step counts during scripted walking tasks.
Walking was assessed A) during a six-minute walk test, B) during walking interspersed with postural transitions, C) during a simulated cleaning task, and D) during stair climbing and descent. Measurements in perfect agreement would fall along the indicated diagonal black line. Regression lines (A, C, D) demonstrate deviations between actual and perfect agreement. Regression lines were omitted from B because the population did not exhibit sufficient variation in step count to yield linear trends. Fitbit registered no steps for two patients who used rollators (noted with +) during the 6-minute walk test (A). One participant’s dyskinesia (noted with +) caused Fitbit to overestimate step count while walking with postural transitions (B), though not on other tasks when she could use a rollator or hold on to nearby structures for balance.