Intended for healthcare professionals

Endgames Statistical Question

Bias in observational study designs: prospective cohort studies

BMJ 2014; 349 doi: (Published 19 December 2014) Cite this as: BMJ 2014;349:g7731
  1. Philip Sedgwick, reader in medical statistics and medical education1
  1. 1Institute for Medical and Biomedical Education, St George’s, University of London, London, UK
  1. p.sedgwick{at}

The Women’s Health Initiative Observational Study, a prospective cohort study, was designed to investigate causes of morbidity and mortality in postmenopausal women. In total, 93 676 women aged 50-79 years were recruited at 40 clinical centres throughout the United States between 1993 and 1998. Women were not recruited if they had conditions that were predictive of survival less than three years or had complicating conditions such as alcoholism, drug dependency, or dementia.

Researchers used the data collected for this study to investigate the association between smoking and the risk of invasive breast cancer. For this analysis, 13 686 women from the original cohort were excluded, including 12  075 with a history of cancer (except non-melanoma skin cancer) at baseline and 1168 whose smoking status was missing. In addition, 443 women were lost to follow-up. The size of the remaining cohort for analysis was 79 990.1

The primary outcome was pathologically diagnosed invasive breast cancer. Smoking behaviour had been assessed at baseline using self reported measures of lifetime passive and active smoking exposure. Information on other risk factors as potential confounders had also been collected at baseline, including age, ethnicity, body mass index, physical activity, and alcohol intake. The women were followed prospectively until 14 August 2009, or until they were diagnosed with invasive breast cancer, whichever came first. The average length of follow-up was 10.3 years. In total, 3520 incident cases of invasive breast cancer were identified. Compared with women who had never smoked, the risk of breast cancer was significantly higher in former smokers (adjusted hazard ratio 1.09, 95% CI 1.02 to 1.17) and in current smokers (1.16, 1.00 to 1.34). Among women who had never smoked, those with the most extensive exposure to passive smoking had a significantly increased risk of breast cancer compared with those who had never been exposed …

View Full Text

Log in

Log in through your institution


* For online subscription