Table 3.

Estimates of validity of non-occupational continuous physical activity measurements derived from the EPIC questionnaire when compared to accelerometer measurements of total physical activity

Accelerometer versus 10-month EPIC	MET-hours/week
	Correlation (ρ)	95% CI
Total non-occupational activitya
Overallb (n = 182)	0.21	0.07, 0.35	**
Gender
Males (n = 100)	0.24	0.05, 0.42	*
Females (n = 82)	0.16	-0.06, 0.36
Body mass index
< 27.2 (n = 89)	0.33	0.14, 0.51	**
≥ 27.2 (n = 92)	0.12	-0.09, 0.32
Age
< 58 years (n = 95)	0.25	0.05, 0.43	*
≥ 58 years (n = 87)	0.18	-0.03, 0.37
Employment status
Full-time work (n = 113)	0.17	-0.02, 0.34
Other (n = 68)	0.30	0.07, 0.50	*
Vigorous activity, self-ratedc	0.18	0.04, 0.32	*
Vigorous activity, MET-assignedc	0.23	0.09, 0.37	**
Light-moderate activityd	0.19	0.05, 0.33	**

* P < 0.05; ** P < 0.01; *** P < 0.001

EPIC, European Prospective Investigation into Cancer and Nutrition; ρ, Spearman rank correlation coefficient; CI, confidence interval

^a EPIC total non-occupational activity = recreational + household physical activity; compared to total accelerometer activity (sum of light, moderate and vigorous activity)

^b There were no statistically significant differences in correlations for non-occupational activity between subgroups (P_diff >0.10)

^c Time spent in vigorous non-occupational activity was measured (1) in a separate question about time in activities causing sweating or faster heartbeat, and (2), using the sum of time spent in activities with MET values ≥ 6 (i.e. cycling, sports and stair climbing); these estimates were compared to vigorous-intensity accelerometer activity.

^d Time spent in light-moderate non-occupational activity was estimated using the sum of time spent in activities with MET values < 6 (i.e. housework, walking, gardening, home repair); these estimates were compared to light+moderate-intensity accelerometer activity.