Does awareness of rates of obstetric interventions change practice?

Dr A E Fletcher, Department of Epidemiology and Population Sciences, London School ofHygiene and Tropical Medicine, London WC1E 7HT. atenolol, but in an observational study we cannot discount the possibility of confounding factors. Our model adjusted for the most obvious of these (age and smoking), but there may be other unidentified factors. The strength of experimental studies is that randomisation removes selection bias. In the Medical Research Council's trial the relative risk of dying from cancer for men receiving atenolol was 1 9 compared with placebo (1-2 to 2 8) and 1-4 (0 9 to 2 2) compared with diuretic. In our study the 95% confidence intervals of the relative risks for all cancers and for lung cancer excluded a twofold excess risk from atenolol. We cannot exclude risks lower than this.