Association of Smartphone-Recorded Steps Over Years and Change in Cardiovascular Risk Factors Among Working-Age Adults

Abstract

Background Few data exist on long-term steps and their relation to changes in cardiovascular disease risk factors. We aimed to examine the associations using long-term smartphone-recorded steps. Methods and Results The present analysis made use of data from 2 national databases and a commercial app database. We evaluated the associations between smartphone-recorded daily steps over 2 years and 2-year changes in the cardiovascular disease risk factors. A total of 15 708 participants with mean (SD) age of 44.1 (9.5) and 23.5% women were included. After adjustment for potential confounders, differences in weight were almost linearly associated with 2-year steps in men (estimate [SE] per 1000 steps/d: -0.33 [0.029] kg), and inversely related only above 5000 steps/d in women (-0.18 [0.054] kg). An inverse linear association with systolic blood pressure was observed in men (-0.34 [0.097] mm Hg) but not in women. Greater steps were associated with change in high-density lipoprotein cholesterol and triglycerides (0.61 [0.068] and -3.4 [0.61] mg/dL in men; 0.64 [0.17] and -2.3 [0.67] mg/dL in women), while changes in low-density lipoprotein cholesterol were evident in men only (-0.59 [0.17] mg/dL). A significant negative association with hemoglobin A1c was observed only in women (-0.012 [0.0043] %). Conclusions In a large cohort of Japanese adults, smartphone-recorded steps over years were associated with beneficial changes in cardiovascular disease risk factors, with some differences between men and women in the associational patterns. The findings support the benefit of long-term physical activity for cardiovascular disease health and suggest a useful role of smartphone-recorded steps for monitoring cardiovascular disease risk over the long term.

Keywords: cardiovascular disease; glucose metabolism; mobile health; physical activity; smartphone; step counts; weight loss.

Figure 1

Study design. The exposure consecutively measured averaged step counts over 2 years, and the outcomes were 2‐year changes in CVD risk factors including weight, body mass index, waist, systolic and diastolic blood pressure, high‐density lipoprotein cholesterol, low‐density lipoprotein cholesterol, triglyceride, fasting plasma glucose, and HbA1c. Laboratory data were ascertained in the baseline and follow‐up (24–35 months after baseline) health check‐ups. Models were adjusted for baseline demographics and laboratory data as well as baseline steps measured over 1 year before the participant's baseline health check‐up. CVD indicates cardiovascular disease; and HbA1c, hemoglobin A1c.

Figure 2

Associations of 2‐year step volume and changes in weight and waist. Figures show sex‐specific, nonlinear associations of 2‐year steps as steps/d (x‐axis) and 2‐year changes of weight (kg), BMI (kg/m²), and waist (cm) (each: y‐axis). These associations were assessed by generalized additive models with cubic spline functions of 2‐year steps, adjusted for age, BMI, current smoking, alcohol drinking, use of lipid‐lowering drugs, glucose‐lowering drugs, and antihypertensive drugs at baseline health check‐up, baseline steps, and months between baseline and follow‐up health check‐ups; and additionally for baseline waist in the model of waist changes. Gray shadows show the 95% CIs. BMI indicates body mass index.

Figure 3

Associations of 2‐year step volume and changes in blood pressure. Figures show sex‐specific, nonlinear associations of 2‐year steps as steps/d (x‐axis) and 2‐year changes of sBP and dBP (mm Hg) (each: y‐axis). These associations were assessed by generalized additive models with cubic spline functions of 2‐year steps, adjusted for age, BMI, current smoking, alcohol drinking, use of lipid‐lowering drugs, glucose‐lowering drugs, and antihypertensive drugs at baseline health check‐up, baseline steps, and months between baseline and follow‐up health check‐ups; and additionally for systolic or diastolic BP in the model for systolic and diastolic BP changes, respectively. Gray shadows show the 95% CIs. BMI indicates body mass index; dBP, diastolic blood pressure; and sBP, systolic blood pressure.

Figure 4

Associations of 2‐year step volume and changes in lipid‐related biomarkers. Figures show sex‐specific, nonlinear associations of 2‐year steps as steps/d (x‐axis) and 2‐year changes of HDL‐c (mg/dL), LDL‐c (mg/dL), and triglyceride (mg/dL) (each: y‐axis). These associations were assessed by generalized additive models with cubic spline functions of 2‐year steps, adjusted for age, BMI, current smoking, alcohol drinking, use of lipid‐lowering drugs, glucose‐lowering drugs and antihypertensives at baseline health check‐up, baseline steps, and months between baseline and follow‐up health check‐ups; and additionally for the related lipid biomarker with respect to each outcome. Gray shadows show the 95% CIs. BMI indicates body mass index; HDL‐c, HDL‐cholesterol; LDL‐c, LDL‐cholesterol; and TG, triglyceride.

Figure 5

Associations of 2‐year step volume and changes in glucose‐related biomarkers. Figures show sex‐specific, nonlinear associations of 2‐year steps as steps/d (x‐axis) and 2‐year changes of fasting glucose (FPG, mg/dL) and HbA1c (%) (each: y‐axis). These associations were assessed by generalized additive models with cubic spline functions of 2‐year steps, adjusted for age, BMI, current smoking, alcohol drinking, use of lipid‐lowering drugs, glucose‐lowering drugs, and antihypertensives at baseline health check‐up, baseline steps, and months between baseline and follow‐up health check‐ups; and additionally for the related baseline glucose biomarker with respect to each outcome. Gray shadows show the 95% CIs. BMI indicates body mass index; FPG, fasting plasma glucose; and HbA1c, hemoglobin A1c.