The difference between study recommendations, stated policy, and actual practice in a clinical trial.

BACKGROUND We determined whether physicians involved in a clinical trial adhere to the study recommendations or the stated policy of their treatment centre with respect to the administration of boost radiation after breast conserving surgery. PATIENTS AND METHODS Boost radiation treatment policy was determined by survey at 25 oncology centres involved in a randomised trial of breast or breast plus nodal radiation in Canada. Actual practice was compared with stated policy and study recommendations. RESULTS Among 248 subjects, 201 (81%) were treated according to stated policy [kappa=0.40, 95% confidence intervals (CI) 0.27-0.52; P<0.0001], indicating only a fair to moderate agreement between stated and actual practice, while 232 (94%) were treated according to study recommendations (kappa=0.59, 95% CI 0.40-0.77; P<0.0001), indicating moderate to near substantial agreement between study recommendations and actual practice (P=0.88 for z-test of difference). In a multivariate analysis, subjects who had invasive disease at a resection margin were more likely to get a boost than those with margins clear of invasive tumour by 2 mm [odds ratio (OR) 49, 95% CI 7.6-322; P<0.0001]. CONCLUSIONS Physicians appear compliant with study recommendations for a non-randomised manoeuvre in a clinical trial, possibly at the expense of compliance with stated local policy. Clinical trial protocols should incorporate standard practice.