Skip to main content

Table 3 Differences between calibration methods in regression models of Actigraph counts and VO2 predicted by walking speed

From: Validity of treadmill- and track-based individual calibration methods for estimating free-living walking speed and VO2 using the Actigraph accelerometer

 

Outcome: Actigraph counts · min−1

Outcome: VO2 (ml · kg−1 · min−1)

Parameter

Coef

IPV

(95 % CI)

Coef

IPV

(95 % CI)

Intercept

1442***

418***

(1319, 1565)

8.96***

2.13***

(8.37, 9.55)

Speed (centred at 3.5

km.h−1)

1134***

247***

(1061, 1207)

1.56***

0.61**

(1.28, 1.84)

(Speed)2

–

–

–

0.30***

0.22***

(0.21, 0.40)

C. method

262***

302***

(158, 366)

−2.25***

1.10***

(−2.62, −1.89)

C. method × Speed

198***

145***

(147, 249)

−0.33

0.37*

(−0.68, 0.02)

C. method × (Speed)2

–

–

–

0.08

0.12*

(−0.03, 0.19)

  1. These regression models report on differences between calibration methods in Actigraph counts and VO2 at 3.5 km.h−1walking speed (see regression coefficients for C. method), and differences between calibration methods in associations of walking speed with Actigraph counts and VO2 (see regression coefficients for C. method by Speed and C. method by (Speed2)). Coef = regression coefficient; 95 % CI = 95 % confidence interval; BPV = between-participant variability in regression coefficients (expressed as standard deviations); – = not applicable; C. method = calibration method (treadmill is reference category); *p < .05; **p < .01; ***p < .001