JMIR Publications

ABSTRACT

Background:

Daily physical activities are related to the development of diabetes. However, traditional research methods are not optimal to document long-term daily routine patterns. mHealth offers an effective approach to collect users’ long-term daily activities and daily routine patterns.

Objective:

To evaluate cross-sectional associations of daily routine variables derived from a smartwatch with diabetes and prediabetes status in the electronic Framingham Heart Study (eFHS).

Methods:

At baseline, diabetes was defined as fasting blood glucose (FBG) ≥126 mg/dL or on self-report of glucose lowering medication; prediabetes as FBG 100-125 mg/dL; the rest were referents. Using smartwatch data, we calculated an average daily step count, and identified the first and last watch time on a given day. We compared average daily step counts and the intra-individual day-to-day variability of the first and last watch time in participants with diabetes and prediabetes to referents, adjusting for age, sex, and race/ethnicity.

Results:

We included 796 participants (mean age 53, 62% women, 41 diabetes, 209 prediabetes, and 546 referents) in up to 3-year of follow-up (median follow-up=219 days) since the baseline. On average, participants with diabetes took 1,611 fewer daily steps (95% CI=863, 2,360; p-value<0.001), and displayed 13 minutes (95% CI=6, 20; p-value<0.001) greater variation in their first watch time, 6 minutes (95% CI=2, 10; p-value=0.005) greater variation in their last watch time, compared to referents. Participants with prediabetes also walked fewer daily steps (p-value=0.04) and displayed larger variation in their first watch time (p-value=0.04) compared to referents.

Conclusions:

Participants with diabetes on average walked significantly fewer daily steps and displayed larger variations in their morning and night watch time routines. These findings may shed light on the roles of physical activity and daily routine in diabetes management.