Reproducibility and associated regression dilution bias of accelerometer-derived physical activity and sleep in UK Biobank.

Overall physical activity had good reproducibility with an ICC of 0.75, while other phenotypes showed moderate reproducibility.

• ICC (95% CI) for overall activity was 0.75 (0.74–0.76)
• ICCs for other phenotypes ranged from 0.58 (0.56–0.59) for sleep efficiency to 0.69 (0.68–0.70) for sedentary behaviour
• Nine physical activity and sleep phenotypes were extracted to capture different movement behaviours
• Reproducibility was assessed using intraclass correlation coefficients (ICCs) across up to four repeat accelerometer measurements after 3–4 years

Correcting for regression dilution bias substantially attenuated the association between daily step count and coronary heart disease risk.

• Before correction, there was a 13% lower risk of CHD per usual 4000 steps
• After correction for regression dilution, the association strengthened to a 20% lower risk of CHD per usual 4000 steps
• The analysis used Cox regression with 87,038 participants and 3,879 CHD events
• The correction represented a meaningful change in the estimated effect size, highlighting the importance of accounting for regression dilution in physical activity research

The study analysed data from 3138 UK Biobank participants with up to four repeat accelerometer measurements over 3–4 years.

• 51% of the 3138 participants were women
• Mean (SD) age was 63.1 (9.4) years
• Participants were from the main accelerometry sub-study of UK Biobank
• This represents a larger sample size and longer follow-up than previous reproducibility studies of 7-day accelerometer measurements

Previous studies on the reproducibility of 7-day accelerometer measurements have been limited by small sample sizes and short follow-up periods.

• The current study addressed these limitations by using a large cohort (n=3138) with up to four repeat measurements
• Follow-up period extended to 3–4 years between measurements
• The authors describe prior work as having been 'limited by small sample sizes and short follow-up periods'

The reproducibility of accelerometer-derived physical activity and sleep measures is comparable to that of established clinical measures such as blood pressure.

• The authors state that 'accelerometer measurements are moderately reproducible and comparable to measures such as blood pressure'
• This comparison contextualizes the observed ICCs (0.58–0.75) within the broader epidemiological literature on measurement reproducibility
• The implication is that regression dilution correction, routinely applied to blood pressure in disease association studies, should similarly be applied to accelerometer data