Hormone therapy and ovarian cancer

Abstract

www.thelancet.com Vol 386 September 12, 2015 1037 be increased with long-term use, and that the UK guidelines are set to be reviewed. Unfortunately the study missed an opportunity to inform these regulatory bodies. After estimating the relative risks in this case-control comparison, a glaring omission is that the attributable risk was not reported. If the attributable risk is greater than zero, its value shows the number of cases of ovarian cancer among hormone therapy users that can be attributed to hormone therapy use itself, or alternatively the number of cases of ovarian cancer among hormone therapy users that could be eliminated if hormone therapy use is eliminated. This missed opportunity means that the public health eff ect of advice to women to consider stopping hormone therapy because of the risk of ovarian cancer will remain undefi ned. Additionally, the attributable risk is the only measure that can be used to sustain the present advice that women should use the lowest dose of hormone therapy for the shortest possible time. This advice still has no basis and the present central analysis missed an opportunity to inform on this advice. Why did the study not report on this simple extension to the relative risk? For the attributable risk to be of value, an assumption of causality has to exist between exposure and disease. Although the study strongly states a support for causality, an interpretation of the results cannot stand up to causality scrutiny. The study is unlikely to be repeated or surpassed by a randomised controlled study. Therefore it missed many opportunities, including the clarifi cation of the diff erence between minimally increased epidemiological risk and causation.