The impact of missing Global Positioning System (GPS) signal data on relationships between environmental variables and objectively measured physical activity and sedentary behavior

Purpose: Explore environment-related patterns of GPS signal-loss, levels of objectively measured physical activity (PA) during periods of GPS signal loss, and the effect of missing GPS data on relationships between environmental variables, PA and sedentary behavior.

Methods: Eight free-living studies using GPS and accelerometer devices were pooled, representing a range of participants, lifestyles and built environments. Minutes of sedentary time, light intensity PA, and moderate-to-vigorous PA were assessed for each period of missing GPS signal (‘lapse’). A random-effects mixed model was used to account for clustering of lapse epochs nested within days and subjects. Land use, population density, and minutes of sedentary time were modeled as fixed effects on the square-root transformed lapse length.

Results: 563 subjects with missing GPS data were assessed totaling 2,036 days. GPS lapse epochs range = 1 to 1427 min/day (mean = 176±246 min.). Average daily minutes of PA during missing GPS epochs were: Sedentary = 199±184, light PA = 69±81, MVPA = 3.0±9.0. Compared to time spent in residences, mean (square-root transformed) lapse length was lower in commercial facilities (coeff(SE):-1.023 (0.187), p < 0.001) and in parks (-0.645 (0.319), p = 0.045). Sedentary time was positively associated with lapse length (p < 0.001), with each minute increase in sedentary time associated with 0.05 increase in (sqrt) lapse time.

Conclusions: GPS signal loss may yield biased results in studies of participants who spend considerable time at home or have a sedentary lifestyle when relationships between sedentary time, PA and the environment are of interest.