Ietje Perfors

99 and hospital for weighed randomisation, to ensure optimal comparison of study arms. To minimize the chance of contamination, GPs are only personally informed about the study details after one of their patients is randomised to the intervention. Given the low incidence of cancer in general practice (about 3 new patients meeting our inclusion criteria annually in an average general practice), we accepted the low chance of contamination resulting from the situation were one GP will first have a patient who is randomised to the intervention arm, followed by a patient randomised to the control arm. 4