As clinicians like Latha Palaniappan assess their patients' susceptibility to cardiovascular disease, they enter the patients' race into an automated risk calculator. It's how things have been done for more than a decade, with race included as one of the data points used to predict if someone will have an adverse heart event in the next decade.
Many social determinants of health can influence a patient's risk, but Palaniappan and fellow researchers have noticed, from working with data from patients around the nation, that race is not among the most accurate or equitable.
They decided to design statistical models tha evaluate a patient's heart health more holistically. Those models were recently incorporated into a new risk calculator implemented by the American Heart Association and available worldwide.
"We're excited to use a broader set of health markers to better predict heart attacks, strokes and heart failure," said Palaniappan, MD, professor of cardiovascular medicine.
To develop the models, researchers incorporated standard measurements such as blood pressure and cholesterol levels. But they added kidney and metabolic measurements; body mass index; and social determinants of health such as ZIP code, which can be indicative of education, income, employment, and access to transportation.
"One of the major innovations of the new models is to assess risk based on place, not race," Palaniappan said.
Their new calculations also factor in measurements of kidney and metabolic diseases. Conditions like obesity, chronic kidney disease and diabetes are associated with cardiovascular disease risk and disproportionately impact disenfranchised groups, Palaniappan and her colleagues wrote in a package of papers published Nov. 10 in Circulation.
The decision to eliminate race from the prediction reflects a larger shift in the medical community to avoid the misconception that a patient's race is biologically responsible for their heart health.
One of the major innovations of the new models is to assess risk based on place, not race.Latha Palaniappan
Such thinking has the potential to encourage treatment decisions based on a patient's race, rather than the social determinants that actually influence their condition, Palaniappan and colleagues wrote. Palaniappan hopes the change will help clinicians personalize cardiovascular disease prevention efforts more equitably.
'Triumph in Big Data'
In their studies, Palaniappan and her research colleagues from multiple institutions across the country detailed how they developed their statistical models. It included training them on historic health data set records from more than 3 million adult patients nationally.
"It's really a triumph in Big Data," she said, adding that the previous risk estimator was trained on a significantly smaller data set of participants from cardiovascular cohort studies. "We have already shown that these risk calculations work very well in electronic health record data sets that are more representative of the populations that we serve."
The models learned to weigh a patient's health metrics and predict their likelihood of heart failure, stroke and heart attack within the next 10 to 30 years. The researchers tested the models' performance using records from another 3 million patients and found them to be better at predicting cardiovascular risk than the traditional AHA risk calculator.
"Our measurements show our new models are significantly better at risk prediction." Palaniappan said.
Our measurements show our new models are significantly better at risk prediction.Latha Palaniappan
Not only did the new models perform better but they could also predict risk for more patients -- between the ages of 30 and 79, rather than the previous range of 40 to 75. And they could predict heart failure, a chronic problem with the heart's pumping ability, a prediction not offered by the heart association's current risk calculator.
"It's important that we've expanded the definition of cardiovascular risk to include heart failure," Palaniappan said. "Though ischemic heart disease rates have been going down in this country thanks to preventive therapies, heart failure rates are going up."
The AHA released the new risk calculator for clinician use in January. Next, Palaniappan said, Stanford Health Care clinics can weave information gathered from using it into patients' electronic health records.
"It will be seamless for clinicians to incorporate these new comprehensive risk calculations into their everyday practice, to better predict and prevent cardiovascular disease." she said.
Image: Yurchanka Siarhei