Development Of Wrist And Ankle Cut-points For Youth With The ActiGraph Accelerometer

Purpose: The purpose of this study was to develop and validate physical activity (PA) cut-points for wrist (WR) and ankle (AK) accelerometer placements in youth.

Methods: : For cut-point development 84 girls and 97 boys (mean ± SD; age, 12.0 ± 1.5 yrs) completed 30 min of lying rest and two to seven structured activities (from a list of 25) for 7 min each (ranging from sedentary to vigorous intensity). Cutpoints were validated on 26 boys and 15 girls (age, 12.6 ± 0.8 yrs) who completed approximately 2 hrs of free-living activity. Activity data were collected using an ActiGraph GT3X or GT3X+, positioned on the dominant WR and non-dominant AK and were converted to 5 sec epochs for the vertical axis (VA) and the vector magnitude (VM). Simultaneously EE (METs) was measured using a Cosmed K4b2. Measured METs were calculated as measured activity VO2 divided by measured resting VO2. Receiver Operator Characteristic (ROC) curve analyses were used to develop WR and AK cut-points (VA and VM) for sedentary (SED; <1.5 METs), moderate (MPA; 3.0-5.99 METs), and vigorous (VPA; ≥ 6.0 METs) intensities. Repeated Measures ANOVAs were used to compare measured and predicted (ROC cut-points) time in each intensity category during free-living activity.

Results: WR cut-points for SED, MPA and VPA were: VA ≤ 105, ≥ 262, and ≥ 565 counts/5 sec; VM ≤ 275, ≥ 416, and ≥ 778 counts/5 sec. AK cut-points for SED, MPA and VPA were: VA ≤ 23, ≥ 90, and ≥ 202 counts/5 sec; VM ≤ 127, ≥ 309, and ≥ 515 counts/5 sec. All ROC analyses were similar for area under the curve (≥ 0.818), sensitivity (≥ 0.756), and specifi city (≥ 0.626). WR VA and VM cut-points signifi cantly underestimated measured time in light activity (LPA) by 13.0 – 18.8 min and overestimated VPA by 14.1 -15.9 min (P < 0.05), and were within 7.8 min of measured SED and MPA time (P > 0.05). AK VA and VM cut-points significantly underestimated measured time in LPA and MPA by 11.8 – 18.8 min and overestimated SED and VPA time by 12.0 – 16.5 min (P < 0.05). Individual errors tended to be lower for the AK (95% prediction interval (PI), ± 22.0 – 41.6 min) versus WR (95%PI, ± 30.0 – 49.9 min).

Conclusion: Compared to the Cosmed, the WR or AK cut-points have similar errors on a group basis. Additionally, the WR and AK errors are similar to published errors for ActiGraph cut-points used with a hip location.

Study supported by NIH grant R21HL093407