Risk score uses EHR data to predict COPD patient outcomes

A new computer-based predictive tool is able to accurately forecast clinical outcomes for patients with chronic obstructive pulmonary disease so that clinicians can optimize their care.

For COPD patients, clinicians have traditionally made their own calculations rating the severity of the disease using scores that—according to researchers at Intermountain Healthcare—aren’t optimal at accurately predicting outcomes.

However, results of a study presented this week at the American Thoracic Society’s conference in Dallas show that electronic health records and computer calculations provide a better, more accurate way to predict clinical outcomes for COPD patients.

“Current risk scores help guide care, but the scores that are typically used are easy to remember and based on simple, often inexact calculations you can do in your head,” says Benjamin Horne, director of cardiovascular and genetic epidemiology at Intermountain Healthcare. “They’re not terribly useful and they don’t do all that well at predicting a prognosis.”

Also See: Anthem’s COPD initiative offers sensors on inhalers to manage illness

The study evaluated a predictive tool—called Summit Score—that analyzes electronic health records, weighing a range of risk factors such as age, body mass index, smoking history, prior COPD hospitalization, history of heart attack, history of heart failure, diagnosis of diabetes, as well as the use of medications.

“The benefit of a calculated, objective score for physicians is that they can look at the score and get a quantified, repeatable value that will indicate where the patient is as far as their health,” adds Horne. “That can help indicate to a clinician whether they should give the standard of care or if they need to be a little bit more aggressive.”

According to Horne, the Summit Score calculates patients’ risk for scenarios such as a sudden exacerbation of COPD symptoms, repeat hospital visits, and mortality on a low-moderate-high risk scale of 0 to 30.

“Instead of using a risk score that doesn’t work all that well, you can use a risk score that’s consistently better at predicting outcomes where the computer does the work for you,” he concludes.