Are general practitioners less likely to recommend treatment when given absolute data on its risks and benefits? A web-based randomized survey.

Doctors need to balance benefits and harms of treatments to advise patients and enable shared decision-making, yet studies indicate general practitioners (GPs) may not be familiar with treatment effect sizes. In atrial fibrillation, national guidelines on anticoagulation recommend risk scoring tools, which do not provide details of the absolute effects. To determine if presenting absolute data on benefits and risks of anticoagulation in atrial fibrillation influences GPs' treatment recommendations. One hundred and fifty-nine GPs and 93 GP trainees in South East England were recruited from conferences and via email invitations. Participants completed a questionnaire with three clinical scenarios of progressively lower stroke risk. They were randomized to receive either guideline recommended risk scores, or descriptions of absolute risk, then asked for their anticoagulation recommendation and to self-rate their recommendation strength. In this study, doctors were less likely to recommend anticoagulation when given absolute data compared to risk scores in all three scenarios (P < .0077, P < .0001, P < .0001). In scenario one and two, even those who recommended treatment rated their recommendation strength lower (P < .0001, P < .012). In scenario two and three, qualified GP's decisions were more likely to concord with guidelines than trainees when given risk scores (P < .0072, P < .023), but no significant differences were seen if informed of absolute risk. Doctors should have ready access to both guidelines and absolute data to tailor their advice to individual patients. Care is needed in how information is presented to avoid discouraging evidence-based practice.