RT Journal Article SR Electronic T1 Validity of estimating physical activity intensity using a triaxial accelerometer in healthy adults and older adults JF BMJ Open Sport & Exercise Medicine JO BMJ OPEN SP EX MED FD BMJ Publishing Group Ltd SP e000592 DO 10.1136/bmjsem-2019-000592 VO 5 IS 1 A1 Sho Nagayoshi A1 Yoshitake Oshima A1 Takafumi Ando A1 Tomoko Aoyama A1 Satoshi Nakae A1 Chiyoko Usui A1 Shuzo Kumagai A1 Shigeho Tanaka YR 2019 UL http://bmjopensem.bmj.com/content/5/1/e000592.abstract AB Background A triaxial accelerometer with an algorithm that could discriminate locomotive and non-locomotive activities in adults has been developed. However, in the elderly, this accelerometer has not yet been validated. The aim were to examine the validity of this accelerometer in the healthy elderly, and to compare the results with those derived in a healthy younger sample.Methods Twenty-nine healthy elderly subjects aged 60–80 years (Elderly), and 42 adults aged 20–59 years (Younger) participated. All subjects performed 11 activities, including locomotive and non-locomotive activities with a Douglas bag while wearing the accelerometer (Active style Pro HJA-750C). Physical activity intensities were expressed as metabolic equivalents (METs). The relationship between the METs measured using the Douglas bag and METs predicted using the accelerometer was evaluated.Results A significant correlation between actual and predicted METs was observed in both Elderly (r=0.85, p<0.001) and Younger (r=0.88, p<0.001). Predicted METs significantly underestimated compared with actual METs in both groups (p<0.001). The mean of the errors was −0.6±0.6 METs in Elderly and −0.1±0.5 METs in Younger. The degree of underestimation increased with increasing METs in Elderly (p<0.001). A stepwise multiple regression analysis revealed that predicted METs, age, and weight were related to actual METs in both groups.Conclusion The degree of correlation between predicted and actual METs was comparable in elderly and younger participants, but the prediction errors were greater in elderly participants, particular at higher-intensity activities, which suggests that different predicting equations may be needed for the elderly.