Results from a blind and a non-blind randomised trial run in parallel: experience from the Estonian Postmenopausal Hormone Therapy (EPHT) Trial

Effect of blinding was calculated as the hazard ratio of selected chronic diseases, total mortality and all outcomes. For analysing the effect of blinding on external validity, the hazard ratios from women recruited to the placebo arm and to the non-treatment arm were compared with those not recruited; for analysing the effect of blinding on internal validity, the hazard ratios from the blind trial were compared with those from the non-blind trial.The women recruited to the placebo arm had less cerebrovascular disease events (HR 0.43; 95% CI: 0.26-0.71) and all outcomes combined (HR 0.76; 95% CI: 0.63-0.91) than those who were not recruited. Among women recruited or not recruited to the non-treatment arm, no differences were observed for any of the outcomes studied.Among women recruited to the trial, the risk for coronary heart disease events (HR 0.77; 95% CI: 0.64-0.93), cerebrovascular disease events (HR 0.66; 95%CI: 0.47-0.92), and all outcomes combined (HR 0.82; 95% CI: 0.72-0.94) was smaller among participants in the blind trial than in the non-blind trial. There was no difference between the blind and the non-blind trial for total cancer (HR 0.95; 95% CI: 0.64-1.42), bone fractures (0.93; 95% CI: 0.74-1.16), and total mortality (HR 1.03; 95% CI: 0.53-1.98).The results from blind and non-blind trials may differ, even if the target population is the same. Blinding may influence both internal and external validity. The effect of blinding may vary for different outcome events.[ISRCTN35338757]Evidence from randomised controlled trials (RCTs) is an essential source of information used by public health policy makers as well as by clinicians and patients for making treatment decisions. RCTs are conducted against the backdrop of difficult methodological choices related to the balance between internal validity (reliability of the results) and external validity (generalisability) [1].Internal validity is maximized by reducing bias by using randomisation and blinding [2-4