Join us on March 2nd for an ActiGraph webinar:
Wearable Data Gone Awry: Cautionary Tales from the Clinical Research Trenches.Register Now
Comparison Of Four "Time In Intensity" Physical Activity Indices As Predictors Of Cardiometabolic Health
- Presented on June 17, 2013
Introduction Recent physical activity (PA) guidelines incorporate recognition of the greater health benefits from higher intensity PA, but still dichotomize intensity into moderate or vigorous categories. New ubiquitous wearable accelerometry enables the estimation of intensity along the continuum from moderate intensity upwards.
Purpose To compare two versions of a MET-min points system to two interpretations of recent guidelines, in relation to several cardiometabolic health indices.
Methods NHANES 2003-2004 accelerometry data were screened according to previous NHANES protocols, and converted into four PA indices: a) minutes in moderate intensity or above (3+ METs; minsMPA); b) weighted minutes, where vigorous (6+ METs) minutes were scored double (minsMVPA); c) a MET-min points system (3+ METs) based on gross MET-min (MET-minsGross); and d) a similar points system based on net MET-min (MET-minsNet). Participants with 1+ days, 13+ hours and who engaged in at least one 10+ min bout of MVPA were included. These data processing criteria produced a final sample of N = 440. All four indices required PA to occur in at least 10 min bouts in order to be counted. These indices were regressed on six measures of cardiometabolic health: BMI; HDL-C; triglycerides; fasting glucose; mean arterial pressure; and waist circumference. Logistic regression was also used to compare the associations between the four PA indices and metabolic syndrome.
Results All PA indices were highly interrelated (r = .97 -.99). Correlations with health measures were systematically higher for the MET-mins points systems, especially MET-minsNet, in comparison to minsMPA. This pattern was consistent for RMSE, AIC, BIC, and associated F-values. However, explained variances were low, ranging from r2 = .003 for HDL-C to r2 = .015 for fasting glucose. Similar patterns were evident from logistic regression, with several indices showing improvement for MET-minsNet compared to minsMPA and minsMVPA. However, diagnostic accuracy was unimpressive, with area under the curve being only 0.60.
Conclusions It appears that minsMVPA measures incorporating the new PA guidelines (giving double-credit for time in VPA) are more strongly related to cardiometabolic health than measures based on the previous 150 mins MPA guidelines. Two MET-mins outcomes were even more strongly related to cardiometabolic health. Although all four measures were weakly related to cardiometabolic health, the MET-mins index could easily be converted to an activity points system and may hold promise for converting PA guidelines into a simple metric for popular use, incorporating ubiquitous technology such as mobile phones.