Thank you very much for giving us the opportunity to comment on the letter of Professor Terao.
Our study is an observational follow-up study, where judgement of causality should not only be based on whether the confidence interval includes the value 1. In our opinion judgement of causality should be based on the risk estimate and confidence interval with careful examination of possible bias and confounding. Only when the exposure is randomized, do we have the solid theoretical basis for probability models from which P values are derived [1].
We agree that the classification of β-adrenoceptor blockers may be too simple from a pharmacological point of view [2]. If there is any difference in risk of suicide between subjects included in the medium lipid solubility group of β-adrenoceptor blockers, and in particular if some of the β-adrenoceptor blockers included in this group are not associated with suicide, this misclassification simply leads to an underestimation of the real risk estimate.
References
- 1.Poole C. Low P-values or narrow confidence intervals: which are more durable? Epidemiology. 2001;12:291–297. doi: 10.1097/00001648-200105000-00005. [DOI] [PubMed] [Google Scholar]
- 2.Svendsen TL. Beta-receptorblokkerende midler. In: Kristensen MB, Friis H, Rassing MR, editors. LægemiddelkatalogetCopenhagen: Dansk Lægemiddel Information. 1998. pp. 179–183. [Google Scholar]