Randomized study concerning the effect of the number of risk diagrams (with treatment +/- without treatment), the period of stroke risk estimation (one year or five years) and the target of prescription (the patient with atrial fibrillation or the physician himself, imagining she/he has atrial fibrillation) on the intention to prescribe or not oral anticoagulation.

Objectives:

To answer the questions:

1. Concerning the decision to prescribe (take) treatment on the basis of a decision aid, are there necessary both diagrams (that with risk without treatment, and that with risk under treatment), or it is enough only the second (risk under treatment, which normally gives all the needed information)?
2. Concerning the decision to prescribe (take) treatment, is it a difference between the effects of the presentation on decision aid of the risk of stroke for the next 1 year, and the presentation of the estimated risk of stroke for the next 5 years?
3. Is the decision different if physicians prescribe the anticoagulant treatment to patients, over if they should take it themselves?

Study: 2x3 factorial randomized controlled trial (RCT) for comparison:

1. Between the decision aid with 2 images (without treatment + with treatment) and the one with 1 image (risk only with treatment)
2. Between the effect of the presentation of the stroke risk chart for 1 year and the stroke risk chart for 5 years.
3. The decision to prescribe to patients over the decision to take the treatment themselves.

The comparison will be made for the spectrum of risks (scores CHA2DS2-VASC) from 1-5.

Sample size: was calculated a sample of 948 participants (474 + 474) for p = 0.05, power = 80% statistical difference between decisions of 5% (from 95% to 90%). The study does not have enough power neither to compare the 5 groups CHA2DS2-VASC (but we will make these comparisons with exploratory purpose), nor to test interactions.

Participants: physicians participating to the National Congress of Internal Medicine, physicians participating to courses, professional manifestations.

Randomization: randomization will be done on graph type (1 or 2 pictures), duration of risk estimate (1 year and 5 years) and the size of CHA2DS2-VASC risk score (1 to 5), and target prescription (patient or the doctor himself), a total of 40 possibilities. Randomization will be done in blocks of 40.

Participants will be asked to decide, depending on the risk chart, if the patient (or himself) will be treated, ignoring the risk of bleeding.

The chart will contain the pictogram according to the CHA2DS2-VASC risk score, without communicating the actual score, and the physician will have to make the decision to treat or not, depending on the perceived risk, and not on treatment guidelines.

No. questionnaire: first digit = number of risk diagrams (1 or 2); second digit = number of years for which the risk of stroke is calculated (1 or 5); third digit = CHA2DS2-VASC score (1-5). Ex: 253: 2 decision aid diagrams (with and without treatment), with an estimated risk of stroke for the next five years, in a patient with CHA2DS2-VASC score =3.

Effect (outcome): decision to treat / not to treat the patient / physician himself, with oral anticoagulants.

Statistical analysis: It will look for differences in bivariate analysis, and multivariate = logistic regression (dependent variable = treatment decision, the independent variables = number of charts, period for risk assessment (one or 5 years), prescription target (patient or the physician himself), CHA2DS2-VASC score, time from graduation, medical/teaching grade, working in hospital / ambulatory, the size of the city the physician works in, specialty, gender, age, if the physician has/had someone close with stroke (data from questionnaires).

Inclusion Criteria:

• physicians who prescribe anticoagulant treatment for atrial fibrillation (cardiology, internal medicine, family medicine, hematology) or who deal with patients with stroke (neurology, pathology) or bleeding (gastroenterology)

Exclusion Criteria:

• physicians who never prescribe anticoagulant treatments, or do not deal with patients with stroke or bleeding because of anticoagulants

At the end, when published

age: median (min, max): 39 (25, 83)

• prescription was only simulated - not real patients, physicians had to imagine they had atrial fibrillation
• results could be different on patients
• participants had to decide prescription/not after seeing the risk diagram for only 5 minutes