Minute level smartphone derived exposure to greenness and consumer wearable derived physical activity in a cohort of US women

Abstract

Background: Inconsistent results have been found in the literature on associations of greenness, or vegetation quantity, and physical activity. However, few studies have assessed associations between mobility-based greenness and physical activity from mobile health data from smartphone and wearable devices with fine spatial and temporal resolution.

Methods: We assessed mobility-based greenness exposure and wearable accelerometer data from participants in the US-based prospective Nurses' Health Study 3 cohort Mobile Health (mHealth) Substudy (2018-2020). We recruited 500 female participants with instructions to wear devices over four 7-day sampling periods equally spaced throughout the year. After restriction criteria there were 337 participants (mean age 36 years) with n = 639,364 unique observations. Normalized Difference Vegetation Index (NDVI) data were derived from 30 m x 30 m Landsat-8 imagery and spatially joined to GPS points recorded every 10 min. Fitbit proprietary algorithms provided physical activity summarized as mean number of steps per minute, which we averaged during the 10-min period following a GPS-based greenness exposure assessment. We utilized Generalized Additive Mixed Models to examine associations (every 10 min) between greenness and physical activity adjusting for neighborhood and individual socioeconomic status, Census region, season, neighborhood walkability, daily mean temperature and precipitation. We assessed effect modification through stratification and interaction models and conducted sensitivity analyses.

Results: Mean 10-min step count averaged 7.0 steps (SD 14.9) and greenness (NDVI) averaged 0.3 (SD 0.2). Contrary to our hypotheses, higher greenness exposure was associated non-linearly with lower mean steps per minute after adjusting for confounders. We observed statistically significant effect modification by Census region and season.

Discussion: We utilized objective physical activity data at fine temporal and spatial scales to present novel estimates of the association between mobility-based greenness and step count. We found higher levels of greenness were inversely associated with steps per minute.

Figure 1.

Study participant flow diagram for the Nurses’ Health Study 3 mHealth Substudy and restriction criteria for primary analytic dataset and secondary analytic dataset for cohort population (N) and GPS mobility observations (n).

Figure 2:

Three panel exposure map* a) GPS mobility data over traditional residential buffers of 270 m and 1230 m and a selective daily mobility bias restriction criterion, b) workplace omitted GPS mobility data over traditional residential buffers, and c) active transport (walk to run velocities) GPS mobility metrics of exposure over traditional residential buffers. *This figure does not represent participant data. Data were obtained via the author’s personal collection.

Figure 3.

Associations^a between NDVI^b and average steps per minute across a 10-min period. ^a Controlling for age (years; continuous), socioeconomic status defined as: educationlevel (masters in nursing or higher; binary), and marital status (never [never married]]/ever [married, widowed, divorced]; binary), and area-level measures of neighborhood socioeconomic status (z-score; quartiles), walkability (z-scores; quartiles), mean daily temperature (Celsius; quartiles), daily precipitation (millimeters; binary), season and Census region in the 2018–2020 Nurses’ Health Study mHealth Substudy. ^b NDVI values below 0 represent water, ~0 represent rocks and bare soil including concrete, and values ~0.6–0.8 represent temperate and tropical forests. * Average steps per minute across each 10-min collection period.

Figure 4.

Associations^a between NDVI^b and average steps per minute across a 10-min period stratfiying on season. ^a Controlling for age (years; continuous), socioeconomic status defined as: education level (masters in nursing or higher; binary), and marital status (never [never married]]/ever [married, widowed, divorced]; binary), and area-level measures of neighborhood socioeconomic status (z-score; quartiles), walkability (z-scores; quartiles), mean daily temperature (Celsius; quartiles), daily precipitation (millimeters; binary), and Census region in the 2018–2020 Nurses’ Health Study mHealth Substudy. ^b NDVI values below 0 represent water, ~0 represent rocks and bare soil including concrete, and values ~0.6–0.8 represent temperate and tropical forests *Average steps per minute across each 10-min collection period.

Figure 5.

Associations^a between NDVI^b and average steps per minute across a 10-min period stratifying on region ^a Controlling for age (years; continuous), socioeconomic status defined as: education level (masters in nursing or higher; binary), and marital status (never [never married]]/ever [married, widowed, divorced]; binary), and area-level measures of neighborhood socioeconomic status (z-score; quartiles), walkability (z-scores; quartiles), mean daily temperature (Celsius; quartiles), daily precipitation (millimeters; binary), and season in the 2018–2020 Nurses’ Health Study mHealth Substudy. ^b NDVI values below 0 represent water, ~0 represent rocks and bare soil including concrete, and values ~0.6–0.8 represent temperate and tropical forests *Average steps per minute across each 10-min collection period.

Figure 6.

Associations^a between NDVI^b and average steps per minute across a10-min period, restricting on active transportation (walk to run velocity) GPS mobillity data ^a Controlling for age (years; continuous), socioeconomic status defined as: education level (masters in nursing or higher; binary), and marital status (never [never married]]/ever [married, widowed, divorced]; binary), and area-level measures of neighborhood socioeconomic status (z-score; quartiles), walkability (z-scores; quartiles), mean daily temperature (Celsius; quartiles), daily precipitation (millimeters; binary), season, and Census Region in the 2018–2020 Nurses’ Health Study mHealth Substudy. ^b NDVI values below 0 represent water, ~0 represent rocks and bare soil including concrete, and values ~0.6–0.8 represent temperate and tropical forests *Average steps per minute across each 10-min collection period.