More on patients' preferences in treating atrial fibrillation

To the editor,

We are writing to respond to the commentary written by Professor Parducci that accompanied our recently published article.^{1, 2} We would like to clarify a few points.

Probabilities in the decision tree

Although we were unable to assign risk of stroke according to severity of atrial fibrillation, we did estimate each patient's probability of stroke according to their risk factor profile. This was based on probabilities derived from cohort studies and randomized trials (see table 1 and references 1, 11-13 from our original article). Thus, we feel that Professor Parducci's criticism that we “lumped together” patients with different risk factor profiles is not correct.

Utilities

His suggestion that input regarding utilities should be sought from “someone who knows the patient and who has observed how similar patients have handled the consequences” is interesting. We agree that input from patient and doctor followed by joint deliberation more closely conforms to a model of shared decision making.² However, there is some evidence that treatment preferences may differ substantially between patients and their health professionals.³ Decision analysis has the benefit of explicitly quantifying each patient's own preferences. We suggest that decision analysis be regarded as a useful starting point for discussion between patients and health professionals about the benefits and potential side effects of treatment, rather than a prescriptive decision that the patient should follow.

We agree that utility assessment using a rating scale may be easier to achieve by practicing physicians in a 10-minute consultation than the time trade-off of standard gamble methods. However, rating scale scores require appropriate correction to be valid,⁴ thereby removing some of the advantage in terms of saving time. It may be that the best way to obtain valid utilities is to allow patients more time to complete the utility assessment procedure, possibly using a computer on their own or with the help of another health professional, before consulting their physician for discussion about treatment.