Accuracy of Direct Observation to Assess Physical Activity in Older Adults

Abstract

PURPOSE: To determine the accuracy of direct observation (DO) to estimate MET level and intensity category during lab-based and free-living activity in older adults. METHODS: Adults over 60 yrs engaged in 1 hour of unstructured activity in the laboratory and 3 hours of unstructured free-living activity within the community. During data collection, participants wore a portable metabolic system to measure energy expenditure and were directly observed by a trained observer. DO recorded MET-level point estimates and intensity category (sedentary, light, moderate, vigorous). Chi-square analyses and intraclass correlation coefficients were used to determine agreement between the two measurements. RESULTS: In-lab analysis included 7 (3F,4M) participants (72.3 yr, 26.0 kg·m-2) for a total of 25,215 observation points and free-living analysis included 4 (2F, 2M) participants (70.8 yr, 24.2 kg.m-2) for a total of 41,190 observation points. There was a significant difference between measured METs and DO estimated METs and intensity categories during in-lab (p<0.001) and free-living (p<0.001) observation, however, there was a moderate relationship between the two MET assessments (ICC=0.54-0.64, p<0.001). When activities were examined individually, highest DO correct MET point estimate classification was seen during standing and sitting still (27.9% and 22.5%, respectively) in lab, and sitting and lying (31.2% and 40.2%, respectively) in free-living. The lowest DO correct classification for MET point estimates was seen in walking up stairs, both in lab (1.9%) and in free-living (1%) settings. The highest percent classification intensity estimates were seen in low intensity activities (sitting-still 56.6%, laundry 61.2%) and the lowest intensity classification was seen during stair climbing (4.6%). CONCLUSION: Use of direct observation to estimate MET level and intensity categories in older adults showed consistent low accuracy in a lab and free-living setting. Discrepancies between measured MET values and compendium MET values (DO) contributed to the large inaccuracies seen between the two measurements. Future research using DO in older adults should utilize older adult specific intensity values.