We presented all relevant results so that readers could decide for themselves whether their interpretation of our findings fits with ours. Brind thinks that the analysis in which both ever and never users lost to follow-up before the age of 38 were excluded (adjusted relative risk 0.95, 95% confidence interval 0.88 to 1.02) shows serious bias and invalidates the main dataset results in which only never users younger than 38 were excluded (0.88, 0.83 to 0.94). Since the total population in each analysis is different, their results should not be compared directly (in the same way that the results from the main and the general practitioner observation dataset should not be compared directly, as indicated in the footnote to table 2).
Furthermore, the different standard populations inevitably result in some variation in the point estimates from each analysis. The point estimate of 0.95 from the “fully excluded” analysis is not materially different from that of 0.88 from the “partially excluded” analysis, although the latter is based on more data. In addition, the upper 95% confidence interval of the fully excluded point estimate was just above unity, suggesting no increased risk of overall cancer.
The subgroup analyses of duration and time since last use of oral contraception included a large number of comparisons, some of which may have reached significance by chance. Although the relative risk of breast cancer was raised among ever users who had stopped 15-20 years previously, it was decreased in those who had stopped more than 20 years previously (0.54, 0.35 to 0.82) and the trend over time was not significant. We cannot explain the increased risk of central nervous system/pituitary cancer among ever users, although the number of women affected was small (49 of the 3877 cancers in the main dataset). We highlighted and discussed the increased risk of any cancer among women using oral contraception for more than eight years in the paper, press releases, and media interviews. Our interpretation remains that oral contraception was not associated with an overall increased risk of cancer—indeed it may even produce a net public health gain.