a1 Cancer Research UK Epidemiology Unit, University of Oxford, Radcliffe Infirmary, Oxford, OX2 6HE, UK
Objective: To assess the accuracy of self-reported waist and hip circumferences and the waist-to-hip ratio (WHR) by comparison with measured waist and hip circumferences and WHR in a sample of middle-aged men and women.
Design: Analysis of measured and self-reported waist and hip data from participants in the Oxford cohort of the European Prospective Investigation into Cancer and Nutrition (EPIC–Oxford).
Participants: Four thousand four hundred and ninety-two British men and women aged 35–76 years.
Results: Spearman's rank correlation coefficients between measured and self-reported waist, hip and WHR were 0.80, 0.74 and 0.44, respectively, for men and 0.83, 0.86 and 0.62 for women. Waist was underestimated, on average, by 3.1 (standard deviation (SD) 5.6) cm in men and 1.9 (SD 5.4) cm in women. The extent of underestimation was greater in participants with larger waists, older participants and women with greater body mass index (BMI). Hip was underestimated by a mean of 1.8 (SD 4.9) cm in men and 1.2 (SD 4.5) cm in women; the extent of underestimation was greater in participants with larger hip circumference and older participants. On average, WHR was underestimated by less than 2% by men and women; the extent of underestimation was greater among those with larger WHR, older people and those with greater BMI. Using self-reported values, the proportion of classification to the correct tertile was over 65% for waist and hip measurements. For WHR this proportion was 50% among men and 60% among women.
Conclusions: Self-reported waist and hip measurements in EPIC–Oxford are sufficiently accurate for identifying relationships in epidemiological studies.
(Received August 26 2003)
(Accepted January 07 2004)