2012 Annual Conference

Coby D. Nirider, PT, DPT

Touchstone Neurorecovery Center, Conroe, Texas

Activity monitors provide objective assessment of physical activity in individuals. While the validity of activity monitors has been tested in normal populations, research is lacking in individuals with Brain Injury (BI). The purpose of this study was to examine the accuracy and agreement of these sensors in persons with BI.

A convenience sample of twenty-two individuals with BI (20 traumatic, 2 acquired) participating in post-acute BI rehabilitation who could ambulate without assistance was recruited for the study.

Participants performed a 2-minute walk test (2MWT) at their self-selected pace while wearing a pedometer and two different accelerometer based sensors (Fitbit and Nike+Fuelband). Participants were videotaped during the 2MWT and actual steps taken (aSteps) were counted from the video. All participants were also tested with two different Fitbit devices. Accuracy and agreement of the different sensors were determined using two different analyses techniques: Bland Altman and Interclass Correlation (2,1) (ICC).

Participants had a mean age of 40.9 (11.4) years, who were 74.1 (114.3 ) months post injury. Mean Berg Balance Scale score was 49.6 (7.3), mean gait speed 1.04 m/s (0.24), and mean lower extremity Fugl Meyer motor score of the more affected lower extremity was 27.91 (3.4). The Bland Altman plot for the 2MWT revealed a mean difference of       -21.0 steps (95% limits of agreement (LOA) 80.7 to -122.7) between aSteps and pedometer, -1.4 steps (95% LOA 10.0 to – 12.8) between aSteps and Fitbit, and -63.6 steps (95% LOA 58.7 to -185.9) between aSteps and Nike+Fuelband. The Bland Altman plots to ascertain agreement between the two Fitbit monitors revealed a mean difference of 1.4 (95%LOA 8.5 to -5.7) on the 2MWT. 

ICC_2,1 between aSteps and Fitbit was 0.96 (95% CI .908 to .983; p=.000), between aSteps and pedometer was0.32 (95% CI -.068 to .636; p=.054), and between aSteps and Nike+Fuelband was0.21 (95% CI -.108 to .530; p=.057).  ICC between the two Fitbit devices was 0.99 (95% CI .963 to .994; p=.000).  

The Fitbit activity monitor was found to be the most accurate device in detecting walking activity in individuals with BI who were able to walk at relatively fast speeds.  Good reliability was also found between the two Fitbit devices tested.  The pedometer and Nike+Fuelband were not accurate.  The results of this study support the consideration of the Fitbit as a simple, inexpensive and accurate means of monitoring walking activity in clients with BI.  Further research of activity monitoring devices is necessary to determine their accuracy with a broader spectrum of neurological diagnoses and also under different walking conditions.