Associations between childhood BMI and adult cancer risk were investigated in a historical cohort study based on the Carnegie ("Boyd Orr") Survey of Diet and Health in Pre-War Britain (1937-9). In 14 centres in England and Scotland, children had their height and weight measured. We included 2,347 individuals aged between 2 and 14 years 9 months at the time of measurement, who were traced through the National Health Service Central Register. Relative cancer risk (registration or death) was estimated in relation to age- and sex-specific BMI SD scores. We studied associations with (i) all cancers, (ii) cancer groups stratified according to their relationship to smoking and (iii) certain site-specific cancers. In the 50 years of follow-up, 188 men and 192 women developed cancer. There was a 9% increase (95% CI -3 to 22%) in risk of cancer in adulthood per SD increase in BMI measured in childhood. There was no evidence of confounding by childhood or adulthood socioeconomic position, other anthropometric variables, childhood energy intake or birth order. There was a 30% increase (95% CI 10-54%) in risk of smoking-related cancers per SD increase in childhood BMI. There was no relationship between BMI and cancers not related to smoking. Associations for all cancers and non-smoking-related cancers tended to be stronger in children who were measured at an older (>8 years) rather than a younger (< or =8 years) age. We conclude that childhood BMI is related to increased risk of cancer in later life, particularly smoking-related cancers.
Copyright 2004 Wiley-Liss, Inc.