Mendelian diseases, those caused by a single deleterious genetic variant, often present as constellations of clinical features.
In a previous study, Lisa Bastarache, MS, Joshua Denny, MD, MS, and colleagues demonstrated a computational method, phenotype risk scoring, or PheRS, that uses hospital billing codes to identify likely Mendelian disease cases in an electronic health record (EHR) population. In that study, comparison of risk scores with genotypes revealed previously unknown deleterious variants and suggested a potential expanded role for genetic testing in the clinic.
Back with a new study in the Journal of the American Medical Informatics Association, the investigators confirm cases of 16 different Mendelian diseases within a de-identified EHR group of 2.5 million patients and proceed to compare four variations on PheRS.
Not uncommonly, PheRS can recognize Mendelian disease cases before clinical teams do, the study shows. All methods served well to distinguish cases from controls; performance tended to improve when billing codes were augmented with clinical lab results.
Bastarache and Denny were joined by Jacob Hughey, PhD, MS, Jeffrey Goldstein, MD, PhD, Julie Bastarache, MD, Satya Das, MD, Neil Charles Zaki, MD, Chenjie Zeng, MPhil, MPH, Leigh Anne Tang and Dan Roden, MD. The study was supported by the National Institutes of Health (LM010685, GM115305).
