Comparison of Actigraph, Fitbit and Manual Step Count during a Two-Minute Walk Test in People with Multiple Sclerosis: A Pilot Study

Thursday, June 2, 2016
Exhibit Hall
Valerie A J Block, PT DPTSc(c) , Physical Therapy, San Francisco State University, San Francisco, CA
Antoine Lize, MSc , UCSF, San Francisco, CA
Elizabeth Crabtree-Hartman, MD, MAS , UCSF, San Francisco, CA
Carolyn J Bevan, MD , UCSF, San Francisco, CA
Jennifer Graves, MD, PhD , UCSF Pediatric MS Center, San Francisco, CA
Ari J Green, MD , Neurology, University of California, San Francisco, San Francisco, CA
Bardia Nourbakhsh, MD , UCSF, San Francisco, CA
Matthew Tremblay, MD, PhD , UCSF, San Francisco, CA
Gregory M Marcus, MD, MAS, FACC, FAHA, FHRS , Director of Clinical Research, Division of Cardiology, UCSF, San Francisco, CA
Jeffrey E Olgin, MD , 505 Parnassus Avenue Room M-1182A,, UCSF, San Francisco, CA
Mark Pletcher, MD, MPH , University of California San Francisco Medical Center, San Francisco, CA
Bruce AC Cree, MD, PhD , University of California San Francisco, San Francisco, CA
Diane D Allen, PT, PhD , Graduate Program in Physical Therapy, University of California San Francisco/San Francisco State University, San Francisco, CA
Jeffrey M Gelfand, MD, MAS , UCSF, San Francisco, CA

Background: Multiple sclerosis (MS) commonly leads to impairments in gait and balance. Gait in MS is usually assessed by history and direct observation in clinic. Remote physical activity monitoring of ambulation in the patient's natural environment has the potential to augment measurement of MS-related disability and disease progression and might have prognostic value. 

Objectives: To compare remote physical activity monitoring using an affordable, convenient to wear, and commercially available device in people with MS with clinic based assessments of ambulation in a dataset of MS subjects with a broad range of ambulatory impairments.

Methods: We evaluated 49 adult participants with MS, recruited from the UCSF MS Center. Participants were excluded if they had a clinical relapse in the last 30 days or had physical comorbidities that could contribute to gait impairment and confound results. Participants completed a 2-minute walk (2MW) test wearing both the commercial-grade Fitbit FlexTM(on their wrist) and the research-grade ActiGraph, GT3X (on an elastic belt around their waist over their non-dominant hip) while a researcher manually counted steps. Fitbit data were collected and stored using the UCSF Health e-Heart electronic study platform.

Results: Interclass correlation coefficients (ICC: 95% Confidence Intervals) comparing total step counts obtained by different methods during the 2MWdemonstrated modest correlations: Fitbit vs. manual (0.69: 0.50 - 0.81), Fitbit vs. ActiGraph (0.61: 0.40 – 0.76), and ActiGraph vs. manual (0.75: 0.60 - 0.85). Bland- Altman plots showed no systematic difference between the number of steps recorded during the 2MW test for Fitbit vs. manual results across the range of step counts. ActiGraph measurements were biased towards lower step counts (slower gait speeds), because ActiGraph tended to under-record actual steps taken. 

Conclusions: Wrist-worn commercial-grade accelerometers (i.e. Fitbit Flex) appear to provide a valid alternative to more expensive, research-grade accelerometers for determining step count in people with MS, particularly at slower gait speeds. Further study is ongoing to validate the Fitbit against ActiGraph in the home setting over several days and evaluate associations with longer-term patterns of neurological impairment and gait abnormalities.  Confirmation of these findings in a larger sample is planned.


Acknowledgement: The FITriMS study is a sub-study of the UCSF Health e-Heart Study.