Validating measures of free-living physical activity in overweight and obese subjects using an accelerometer.

Abstract

Free-living physical activity can be assessed with an accelerometer to estimate energy expenditure but its validity in overweight and obese subjects remains unknown. Here, we validated published prediction equations derived in a lean population with the TracmorD accelerometer (DirectLife, Philips Consumer Lifestyle) in a population of overweight and obese. We also explored possible improvements of new equations specifically developed in overweight and obese subjects. Subjects were 11 men and 25 women (age: 41±7 years; body mass index: 31.0±2.5 kg m(-2)). Physical activity was monitored under free-living conditions with TracmorD, whereas total energy expenditure was measured simultaneously with doubly-labeled water. Physical activity level (PAL) and activity energy expenditure (AEE) were calculated from total energy expenditure and sleeping metabolic rate. The published prediction equation explained 47% of the variance of the measured PAL (P<0.001). PAL estimates were unbiased (errors (bias±95% confidence interval): -0.02±0.28). Measured and predicted AEE/body weight were highly correlated (r(2)=58%, P<0.001); however, the prediction model showed a significant bias of 8 kJ kg(-1) per day or 17.4% of the average AEE/body weight. The new prediction equation of AEE/body weight developed in the obese group showed no bias. In conclusion, equations derived with the TracmorD allow valid assessment of PAL and AEE/body weight in overweight and obese subjects. There is evidence that estimates of AEE/body weight could be affected by gender. Equations specifically developed in overweight and obese can improve the accuracy of predictions of AEE/body weight.