Use of objective physical activity measures is rising. We investigated the representativeness of survey participants who wore an accelerometer.
4273 adults aged 16+ from a cross-sectional survey of a random, nationally representative general population sample in England in 2008 were categorized as 1) provided sufficient accelerometry data [4−7 valid days (10+ hrs/d), n = 1724], 2) less than that (n = 237), or 3) declined (n = 302). Multinomial logistic regression identified demographic, socioeconomic, health, lifestyle, and biological correlates of participants in these latter 2 groups, compared with those who provided sufficient accelerometry data (4+ valid days).
Those in the random subsample offered the accelerometer were older and more likely to be retired and to report having a longstanding limiting illness than the rest of the adult Health Survey for England participants. Compared with those providing sufficient accelerometery data, those wearing the accelerometer less were younger, less likely to be in paid employment, and more likely to be a current smoker. Those who declined to wear an accelerometer did not differ significantly from those who wore it for sufficient time.
We found response bias in wearing the accelerometers for sufficient time, but refusers did not differ from those providing sufficient data. Differences should be acknowledged by data users.
The authors are with the Dept of Epidemiology and Public Health, University College London, United Kingdom.