Integrating Accelerometers to Measure Physical Activity in a National Health Surveillance System: Investigating...

This study examined the feasibility of integrating device-based physical activity (PA) measurement into the Dutch national public health monitoring system by assessing (1) selection bias across study phases and (2) agreement between self-reported and device-based adherence to the guidelines of ≥150 minutes per week moderate to vigorous PA.

A random sample of 125,531 adults from 2 municipal health services was invited to participate in the 2024 Dutch Public Health Monitor, which included a questionnaire to assess PA. Respondents indicated willingness to participate in a 9-day activPAL protocol. From those willing to participate, 1100 individuals were selected through stratified random sampling. Logistic regression analyses examined factors associated with willingness, participation, and agreement between questionnaire and activPAL.

Of those invited, 40,018 (31.9%) completed the Public Health Monitor, of whom 6679 (16.7%) respondents were willing to participate in the activPAL study. Willingness was higher among women, people with higher income/education, and people meeting PA guidelines. It was lower among older adults and people with a migrant background. According to the activPAL, 48.0% met the guideline versus 64.4% according to self-report. Agreement was 60.9% (κ = .23), with most mismatches reflecting overreporting of PA by self-report. Lower odds of agreement between the methods were observed for women and people with a migrant background.

Integrating activPALs into national surveillance was feasible but prone to selection bias. Agreement between activPAL and self-report was modest, with self-report overestimating the proportion meeting the guideline. Improved recruitment strategies, such as oversampling underrepresented groups, may strengthen population-level device-based PA surveillance by countering selection bias.