鶹ý

Algorithm Has Promise for Finding Undiagnosed FH

— Cleveland Clinic study used EHRs to screen for familial hypercholesterolemia

MedpageToday

This article is a collaboration between 鶹ý and:

PHILADELPHIA -- Screening for familial hypercholesterolemia (FH) may be possible using electronic health records (EHRs), according to researchers, who developed an EHR-based algorithm to flag potentially undiagnosed cases.

In a retrospective review of more than 2 million EHRs at a single center, the algorithm identified 46 patients with definite FH, 289 individuals with probable FH, and 649 patients with possible FH, reported Kristin Bede, now a medical student at Ohio University's osteopathic medical school.

"There are a lot of familial hypercholesterolemia patients who are missed, and we [find out about them] the hard way -- when they have a cardiac event," Bede told 鶹ý at her group's poster at the American Heart Association meeting. "We decided to be proactive in screening. We wanted to find a way within the EHRs to streamline the process so we can catch these people earlier, rather than after an event that could be fatal."

Bede and colleagues conducted the study when she was at the Cleveland Clinic. They filtered out patients (ages 20 to 49) in the EHR system who had a documented cardiac event, and who had high LDL cholesterol levels (190 mg/dL or greater), but were not currently diagnosed with FH, from 2014 to 2019.

"We were looking for those who already had an event, and also their family history, to see if they were being seen in the cardiology department, and who was being treated with PSCK9 inhibitors, the leading treatment for FH," she said.

The algorithm applied point values to the (DLNC) criteria, based on family history, clinical history, physical examination, cholesterol levels, and DNA analysis. A DLNC score of 8 or more on the algorithm was considered likely for FH; a score below 3 indicated that FH was unlikely.

In the total study population, "there were disparities in appropriate diagnosis, treatment, and specialist referrals," the researchers stated.

Of the patients identified as having definite FH, less than a quarter were diagnosed with FH by ICD-10 codes. Of those, almost three-fourths were on high-intensity statin therapy, and a little less than 20% were on PSCK9 inhibitors.

In the probable FH subset, more than three-fourths were on high-intensity statins yet less than 10% had a formal FH diagnosis by ICD codes.

Also, less than 10% were referred to medical genetics, and less than 20% to preventative cardiology, although well over half were referred to any cardiology specialist.

Bede noted that definite FH has to be confirmed through genetic testing, which families be may reluctant to undergo. "A lot of people will just begin therapy without getting the genetic testing," she said.

"Through EHR data advancements, there is opportunity for more proactive classification and treatment of familial hypercholesterolemia patients," the researchers concluded.

Patrick Kee, MD, PhD, of the University of Texas Health Science Center at Houston, told 鶹ý that he thought the algorithm "had promise," but questioned whether it could be used in large-scale EHR systems with multiple hospitals.

Disclosures

Bede and Kee disclosed no relevant relationships with industry.

Primary Source

American Heart Association

Bede K, et al "Automated Electronic Health Record Screening for Familial Hypercholesterolemia" AHA 2019.