Associations between the lived environment and wearable accelerometer-derived physical activity volume and intensity during long-term continuous monitoring in the SafeHeart cohort

Abstract Background SafeHeart is a multicentre, prospective observational clinical study in the Netherlands (NL) and Denmark (DK) which collected wrist-worn accelerometry for 6 months from 300 participants with implanted cardioverter-defibrillators (ICD). Seasonal and weather-related physical activity (PA) patterns have previously been described in healthy populations but so far not in a population similar to SafeHeart. NL and DK are relatively similar both socio-economically and in terms of climate with some differences in seasonal changes due primarily to different latitudes. Purpose Describe seasonal, weather- and ICD event-related patterns of physical activity volume and intensity distribution in a cohort with ICDs across two countries. Methods Daily physical activity (PA) volume (Average Acceleration, AvAcc) and intensity distribution (Intensity Gradient, IG) were derived from raw accelerometer data. Higher values of IG indicate a larger proportion of PA accumulated at higher intensities. Associations of AvAcc and IG with season, weather, and proximity to ICD therapy events were estimated using linear mixed model. Results 34,856 total days of observation were included from 266 participants (N=52 female) with a median age of 63 (IQR: 57, 71), including 38 days where one or more ICD therapy events occurred between 28 different participants. AvAcc (Fig. 1a) was correlated to season, with higher volumes of PA during the spring in NL and during the summer in DK, with a wider range across the year for DK compared to the NL. IG was slightly higher in summer for both countries, but to a lesser degree (Fig 1b). Warmer days (relative to seasonal expectation) were positively correlated to AvAcc (Fig 2a-d) for the NL in winter and spring, and for DK in summer but only slightly positively correlated to IG (Fig 2e-h) for NL in winter and spring, and for DK in winter. Both AvAcc and IG tended to be lower in the month after an event (each day categorised as week 1 or weeks 2-4 following an event), but only IG in weeks 2-4 after an event reached significance (-0.030 (CI: -0.050, -0.009). Conclusions Participants tended to be more active in spring in NL and summer in DK – broadly consistent with seasonal patterns seen in healthy populations and likely reflecting slightly larger seasonal weather differences in DK compared to NL. Seasonal differences in PA volume were not primarily driven by an increased proportion of PA accumulated at higher intensities of PA, indicating that changes in incidental PA and lower-intensity activities may equally contribute. Temperature-related increases in PA volume partially correspond to an increase in IG in NL in winter and spring, but not in DK in summer, indicating that behavioural changes driving a temperature-related increase in PA volume vary by season and country. PA volume and intensity tended to drop in the month following an ICD therapy event, but more data following an event is needed to accurately estimate the effect.Figure 1 Figure 2