Accuracy of Actigraph inclinometer to classify free-living postures and motion in adults with overweight and obesity

Sedentary behaviour (SB) is an independent health risk-factor and interrupting SB seems to be beneficial. SB is both defined by posture and intensity, thus objective measurement of postures remains a priority. The ActivPAL inclinometer (AP) has been used as a reference for postural estimation, but information on the validity of the Actigraph inclinometer (AG_incl) is scarce. We compared postural estimation from waist-worn AG_incl against thigh-worn AP, under free-living. Data from 10 participants (50.4 ± 11.4 years) were used and each 15s-block from 60-valid days was matched for the devices’ comparison. At group level, no differences were found for SB and standing-time between inclinometers (p ≥ 0.05). AG_incl underestimated stepping-time by 26.3 min and overestimated SB-to-upright transitions by 200.7 occasions (p < 0.05). Inter-individual variability was higher for SB and standing-time, and AG_incl was in better agreement with AP for the stepping-time (area under the ROC curve = 0.98, with 100% sensitivity and 94% specificity; CCC = 0.44). These results highlight potential error in estimating individual postures using AG_incl in overweight/obese adults and provide insights on the differences of using specific criteria on data-analysis. Our findings suggest that at the group level, AG_incl provides similar estimates compared to AP for SB and standing-time, but not for SB-to-upright transitions or stepping-time. (ClinicalTrials.govID:NCT02007681)