Evaluation of the accuracy of the Fitbit Zip in clinical populations

Author: Craig Farmer

Farmer, Craig, 2019 Evaluation of the accuracy of the Fitbit Zip in clinical populations, Flinders University, College of Nursing and Health Sciences

Terms of Use: This electronic version is (or will be) made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. You may use this material for uses permitted under the Copyright Act 1968. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact copyright@flinders.edu.au with the details.


Demand for rehabilitation services is increasing with the ageing population. Financial pressures on the health system and allocation of scarce resources require efficient and innovative rehabilitation therapy to maximise gait and mobility function and achieve the patient goal of returning home. In recent years, many commercially available activity monitors have become available, and are increasingly popular in rehabilitation for monitoring patients’ activity levels. These activity monitoring devices use accelerometer technology, and commonly attach to the patient’s clothing, counting steps taken during walking. Accurate step count measurement in the rehabilitation setting is important to understand the association between physical activity and better health outcomes. Having accurate measurements of daily walking mobility and physical activity levels can then be utilised for motivational goal setting and exercise progression. Accurate therapist feedback on walking activity may assist in increasing physical activity dosage and in maximising walking potential. However, accuracy of commercially available activity monitors has been poor in people with altered gait parameters including slow gait speed, slower cadence and short step length. Recent research suggests that activity monitors are potentially more accurate when worn at the distal leg in rehabilitation patients, who walk more slowly than the general population.

There are two main parts to this thesis. The first part investigated the step count accuracy of a commonly used commercially available activity monitor, the Fitbit Zip, worn on the shoe in controlled conditions, compared to direct observation. The second part investigated the Fitbit Zip, worn in free-living conditions, compared to the ActivPAL activity monitor (commonly used in field research). Ninety sub-acute rehabilitation patients with a gait speed of 0.50 – 1.0m/s were recruited from the day rehabilitation service of a major rehabilitation hospital in South Australia. Diagnosis groups included neurological, orthopaedic, and other medical and surgical conditions. Overall accuracy, and the influence of environment was examined in indoor and outdoor conditions. High Fitbit accuracy was observed in controlled conditions, with the Fitbit undercounting steps. In contrast, the Fitbit counted more steps than the ActivPAL when worn in the free-living environment. The studies presented in this thesis identified slower gait speed as the main influencing variable on Fitbit accuracy in continuous walking in controlled conditions. Literature suggests that in free-living conditions, the interrupted patterns of step taking in activities of daily living appear to influence Fitbit accuracy. When combining outcomes from the studies, it appears that slower gait speed and other gait parameters that are more prominent in the community conditions are associated with lower Fitbit accuracy, and the most distal Fitbit location on the forefoot of the shoe maximises step count detection.

Keywords: activity monitor, accelerometer, Fitbit, ActivPAL, step count, rehabilitation, gait speed, walking, indoors, outdoors

Subject: Public Health thesis

Thesis type: Masters
Completed: 2019
School: College of Nursing and Health Sciences
Supervisor: Chris Barr