Determine the direction and magnitude of the associations between Fitbit data and self-report data for sleep and exercise data collected from active users with type 2 diabetes (T2D). METHODS: STEP-D is a longitudinal, pilot study composed of individuals (n=86) diagnosed with T2D. Participants wore a Fitbit for 14 consecutive days and completed four Internet surveys taken at three time points: Day 1 (baseline), Day 7 (interim) and Day 14 (closing). The Fitbit tracked minutes asleep and number of steps taken. The questionnaire included items gauging the number of days exercised in a typical week, gym membership, number of nights in a typical week having trouble sleeping and the number of nights having sleep problems. Means and standard deviation were used to report all data and Pearson correlations were used to test the association between the Fitbit and self-report data. RESULTS:

Participants, on average took 4,955.0 steps/day and slept 6.7 hours/day. They also self-reported an average of 2.0 days of exercise and 2.3 nights having trouble falling asleep in a typical week. The association between self-reported days exercised in a typical week and the correlation for mean steps was strong for Fitbit (r=0.60; p<0.01). Self-reports of sleep issues were moderately correlated with sleep variability. Self-reported nights having trouble falling asleep in a typical week was associated with more time spent in bed based on the Fitbit (r=0.28, p=<0.05). CONCLUSIONS:

Findings indicate that Fitbit and self-report data are positively associated for sleep and exercise, but physical activity is more closely aligned than sleep-related information. This may indicate that FitBit is more valid for measuring certain behaviours. If these findings are replicated in T2D, then large-scale collection of certain objective HRQoL measures is possible, but data limitations will need to be acknowledged.