Evaluation of ActiGraph’s Low-Frequency Filter in Laboratory and Free-Living Environments

Document Type


Publication Date



The ActiGraph (AG) is the most commonly used research-grade physical activity monitor. Although several investigators have examined the effects of the “low-frequency extension” (LFE) on step counts in the free-living environment, a direct comparison with a valid criterion method is lacking. We sought to determine the accuracy of the AG’s LFE to measure step counts during laboratory and free-living activities in two versions of the device (GT1M and GT3X). Methods: Twenty-four participants wore the StepWatch 3 and two versions of the AG, with the LFE turned on (GT1M-LFE and GT3X-LFE) and the LFE turned off (GT1M-N and GT3X-N), on different days. Direct observation of steps and the ankle-worn StepWatch 3 served as the criterion methods for the treadmill and free-living condition, respectively. All statistical analyses were performed on the percent difference between the devices as [(measured - actual)/actual]. Results: During treadmill walking, the GT1M-N and GT3X-N underestimated steps by approximately 60% at 40 m•min-1 (P< 0.001) and by 31% at 54 m•min-1 (P < 0.001). With the LFE turned on, this underestimation was reduced to 7% and 4% for the GT1M and GT3X, respectively. Under free-living conditions, both the GT1M-LFE and GT3X-LFE overestimated steps by approximately 32% (P < 0.001) whereas the GT1M-N and GT3X-N steps were underestimated by 30% and 25%, respectively (P< 0.001). Conclusions: Turning the LFE on lessens the underestimation of steps recorded at walking speeds <54 m•min-1 for both the GT3X and GT1M. However, the increased sensitivity provided by the AG’s LFE results in overestimation of steps taken throughout the day, when compared with the criterion device. Meanwhile, failure to turn the LFE on results in an underestimation of steps taken throughout the day. [ABSTRACT FROM AUTHOR]