Introduction: The goal of this study was to evaluate the convergent validity of physical activity (PA) assessment methods widely used with older adults with an accelerometer-based pattern recognition monitor, the Sensewear Pro 3 Armband (SP3). The ActiGraph (ACT) accelerometer and the Seven Day Physical Activity Recall (7DAY) physical activity questionnaires (PAQ) are two of the most popular methods of PA assessment; however the understanding of their use with an older adult population is limited. PAQs designed specifically for older adults such as the Community Healthy Activities Model Program for Seniors (CHAMPS), Yale Physical Activity Survey (YPAS), Physical Activity Survey for the Elderly (PASE), and Rapid Assessment of Physical Activity (RAPA) are currently lacking comparison with objective criterion measures. The purpose of the study was to establish validity for PA assessment methods, designed for both the general population and older adults, which are lacking comparison with objective criterion measures in an older adult population.

Methods: The participants (n=27; age=74.4 y 6.5) wore the SP3 and ACT for 7 days of normal activities of daily living and completed the PAQs to recall their activity over the same time frame. ACT physical activity energy expenditure (PAEE) was calculated with the Freedson equation (ACT-F) and Work-Energy-Theorem and Freedson combination (ACT-C). The ACT-F and Matthews cutpoint (ACT-M) were used for estimates of minutes per week of PA. A one way ANOVA was used to find differences between genders. Pearson and Spearman correlations examined significant relationships. Paired samples t-tests and Bland-Altman plots investigated differences between methods.

Results: ACT estimates of step count were highly related at 0.96 (p<0.01) with the SP3 although there were differences between methods. The ACT-C showed moderate relationships (r=0.54, p<0.01) with the SP3 in PAEE estimates. A trend of underestimation was present of 78kcal per day. The ACT-M was moderately related (r=0.64, p<0.01) to the SP3 for minutes of PA. It underestimated 47 minutes per week of PA. No significant differences were found between the ACT-C and ACT-M with the SP3. The 7DAY and YPAS correlated moderately well with the SP3 for PAEE (r=0.69 & 0.85, p<0.01) and for minutes of PA (r=0.68 & 0.84, p<0.01). Spearman rank order was 0.73 (p<0.01), 0.56 (p<0.01) and 0.39 (p<0.05) for the 7DAY, YPAS and PASE with the SP3 for PAEE. Rank order was 0.70 and 0.64 (p<0.01) for the 7DAY and YPAS with the SP3 estimates of PA minutes. Significant differences were not found for 7Day and YPAS with the SP3. Wider limits of agreement existed for the YPAS under Bland-Altman analysis. The CHAMPS showed no significant relationships with the criterion measure. Analysis of the RAPA was not possible due to a small, highly active sample.

Discussion: The ACT had reasonable agreement with the SP3 for estimating PAEE and minutes spent in PA. The results support the use of the 7DAY to estimate PAEE in a healthy older adult population. The PAQs designed specifically for older adults such as the YPAS and PASE may only have use for group rankings, while the RAPA may only be useful for classification as active or sedentary. The CHAMPS requires further research with an objective criterion measure before being used for estimates of PA in an older adult population.