SummaryThe relationship between fetal growth as indicated by weight and length at birth, and cancer risk in 1080 adult Swedish women was examined. Birth factors were retrieved from original midwife records for the years 1914, 1918, 1922 and 1930, and primary cancer cases were identified by matching with national and regional cancer registries through the year 1998. A positive and statistically significant increased risk for cancer was found with increasing birth weight or birth length for all site cancer and non-hormone related cancer, defined as all cancer sites excluding breast, uterus and ovary. Addition of factors suspected to influence cancer risk, maternal proteinuria, birth order, own parity and age at menarche, did not attenuate this relation. Previously only breast cancer has been reported to be related to size at birth in adult women and this is the first study to report that cancer sites other than the major hormone-related sites may be influenced by size at birth, as measured by either weight or length at birth; these findings warrant further investigation. These reports are compiled at both the regional and national level. By means of the unique 10-digit personal number assigned to all residents of Sweden, the population study database could be matched with the cancer registries to identify primary incident cases. Date of death (from any cause) was determined by matching personal identification numbers with the Swedish Cause of Death Registry and by confirmation with parish records. Cancer cases were analysed as combined all site cancer morbidity and also divided into 'hormone-related' and 'nonhormonal' cancers and are referred to as such in the following analysis and discussion; the former comprises cancers of the breast, uterus and ovaries, the major cancer sites with a hormonal aetiology (Miller, 1978). Non-hormonal cancers, as defined here, are thus all other cancer sites. As there is evidence that cancer of the colon may be hormone-related (Potter, 1995), the defined hormone-related sites and colon cancer are also examined in combination. Cancers of the 'digestive system', ICD7 codes 150-158, are analysed separately to examine a sub-group of the 'non-hormone-related' cancers.This study was approved by the Research Ethics Committee of Göteborg University.
Statistical methodsCharacteristics of participants and non-participants were compared using a 2-sample t-test for continuous variables and chisquared test for proportions for factors which may influence cancer risk or birth outcome -maternal age, birth order of the participant, own parity (defined as number of pregnancies) and age of menarche of the participants. Birth weight and birth length were analysed as continuous variables and in population quintiles birth weight and tertiles birth length. Cox proportional hazards modelling was used to examine trend in cancer risk in relation to size at birth, treating weight and length separately, and in modelling cancer risk with consideration of covariates implicated in cancer pathogenesis: maternal p...