Does the choice of stepping intensity metric influence dose-response associations with mortality? A UK population cohort study of 70,174 adults

Abstract Background: Research on the health effects of stepping intensity in free–living environments is limited and inconclusive. Inconsistent use of stepping intensity estimation metrics could explain current equivocal results. We aimed to examine and compare a range of different cadence–based metrics in terms of their multivariable–adjusted associations with all–cause (ACM) cardiovascular disease (CVD), cancer and physical–activity (PA)–related cancer mortality. Methods: This prospective cohort study included participants with valid wrist–worn accelerometer data from the UK Biobank. We estimated stepping intensity using ten different cadence–based metrics, including eight peak–cadence metrics (defined as averaged steps / min of the highest but not necessarily consecutive minutes) that most of whom have appeared in prior literature, plus two non–peak–cadence metrics: 1) average daily cadence, defined as steps/accelerometer wearing mins, and 2) average cadence of purposeful steps, defined as averaged steps / min of minutes with ≥ 40 steps. We rescaled each metric into a standardised cadence scale with mean of 0 and standard deviation (SD) of 1, using (absolute–mean)/SD. We compared the dose–response associations of each stepping intensity estimation metrics with mortality outcomes using previously published modelling involving Cox–restricted–cubic–spline model, presented as overlay plots on standardised and absolute cadence scales. Results: Among 70,336 participants (age [SD], 61.6 [7.8] years; female, 40,933 [58%]) followed up for a median of 8.0 years, all cadence–based metrics, besides the average cadence of purposeful steps, exhibited a comparable beneficial dose–response association with ACM/CVD/cancer mortality, with 95% CI largely overlapped (e.g., at –0.2 standardised steps/min, the hazard ratio (HR) of ACM for peak 1– and peak 30–min cadence were: 0.72, 95%CI [0.65, 0.82] and 0.66 [0.60, 0.73], respectively). The average cadence of purposeful steps only did not show dose–response associations with mortality outcomes (e.g., the HR that corresponds to the standardised median for the average cadence of purposeful steps in ACM was 0.98 [95% CI: 0.86, 1.12]. Conclusion: Besides the average cadence of purposeful steps, all stepping intensity estimation metrics demonstrated comparable beneficial dose–response associations with mortality of all–cause, CVD and cancer, suggesting these cadence–based metrics may be used interchangeably for estimating associations of free–living stepping intensity with health outcomes and applied in different research scenarios accordingly.