ACC 2018: Physician judgment more accurate than traditional risk score in predicting coronary artery disease

March 12, 2018 – Integration of physician judgment and risk estimation may optimize stable chest pain evaluation.

Suspected coronary artery disease (CAD) is one of the most common, potentially life-threatening diagnostic problems encountered by clinicians. Informal physician judgment is often used to estimate risk and guide treatment in these patients. However, the relationship between physician estimates and guideline-recommended scores to predict CAD, such as Diamond-Forrester (D-F), was not known.

A study led by DCRI researchers evaluated 4,533 patients with stable chest pain who underwent computed tomography (CT) coronary or invasive angiography in the PROspective Multicenter Imaging Study for Evaluation of chest pain (PROMISE) trial. This randomized, pragmatic comparative effectiveness trial included stable, symptomatic outpatients without known coronary artery disease referred for noninvasive testing at 193 sites in North America. The median follow-up time was 25 months.

Researchers categorized a priori the pre-test likelihood of obstructive CAD for each patient; D-F estimates were also calculated. Overall, the agreement rate between the physician and D-F estimates was poor.

The findings were presented today at the American College of Cardiology 67th Annual Scientific Session in Orlando, Florida.

Compared to the D-F score, physician judgment more accurately identified obstructive CAD and worse patient outcomes. The highest and lowest rates of obstructive CAD – and the highest rates of death, myocardial infarction and unstable angina – were in patients with physician and D-F agreement.

“We believe this is the first study to show that physician judgment in CAD is more accurate than the use of risk scores, and that this was associated with fewer events,” said lead author Christopher Fordyce, MD, a former cardiology fellow at the DCRI who is now clinical assistant professor in the Division of Cardiology at the University of British Columbia. “From the physician standpoint, this validates the need to take into account all available information for each patient, rather than rely on a risk score alone.

“This analysis confirms that clinical expertise still has a very important role in evaluating patients,” noted senior author Pamela S. Douglas, MD, MACC, FACC, Geller Professor of Research in Cardiovascular Diseases at the DCRI. “As we move increasingly towards algorithmic care, we must leave room for physician expertise to individualize approaches. We also need new risk scores, such as the PROMISE Minimal-Risk Tool, which uses readily available pretest variables to discriminates minimal-risk patients, for whom deferred testing may be considered.”

Other Duke authors of the presentation included C. Larry Hill; Adrian Coles, PhD; Kerry L. Lee, PhD; Daniel B. Mark, MD; and Manesh R. Patel, MD.