Influence of decision aids on oral anticoagulant prescribing among physicians: a randomised trial

Oral anticoagulants (OAC) are underused in treatment of atrial fibrillation (AF), with differences in patient and physician preferences. For risk communication, the graphic showing risks on treatment contains all the information, therefore, the graphic showing risks without treatment may not be necessary. Here, our objective was to assess whether decision aids require information of risks without treatment and specifically whether presentation of 5-year stroke risk in patients with AF increases use of OACs compared with presentation of 1-year risk and whether decisions on treatment are different when physicians decide their own treatment vs. that of the patient.

Randomised controlled trial with 2³ factorial design, performed at 12 university hospitals, one internal medicine course and one national medical conference.

Of 968 physicians who participated, 83·3% prescribed anticoagulation therapy. Treatment decisions were not influenced by the number of graphics or by the time frame of risk estimation, with risk differences of 0·5% (95% confidence interval, -4·0% to 5·4%) and 3·4% (-1·3% to 8·1%). However, physician-to-patient prescription rates were 5·4% (0·2-10·6%) more frequent after seeing the 5-year risk graphic. Physician-to-self intentions to prescribe occurred less frequently, with risk difference of 15·4% (10·8-20%). Physicians considered the baseline risk and the absolute risk reduction only when prescribing to patients but not to themselves.

Risks could be communicated using decision aids with only one graphic. Showing the risk of stroke at 5 years could increase the prescription of OACs to patients with AF. Faced with the same risk of stroke, physicians prescribed less to themselves than to patients.