EMRs contain structured [International Classification of Diseases or ICD-9 (billing) codes and Current Procedural Terminology or CPT (procedure) codes; clinical lab test results, semi-structured, and unstructured (clinical notes) data, all of which can be used for electronic phenotyping. The data types currently available in the SD that can be accessed for electronic phenotyping include narratives (such as clinical notes, discharge summaries, history and physicals, problem lists, surgical reports, progress notes, letters), ICD-9 codes, CPT codes, forms (intake, assessment), reports [pathology, electrocardiograms (ECGs), echocardiograms], clinical communications, lab values and vital signs, medication orders, TraceMaster (ECGs), and the tumor registry [41 (link)].
Electronic phenotyping has been previously described for BMI in EAGLE BioVU [42 ]. To extract lipid and type 2 diabetes traits from EAGLE BioVU EMRs, laboratory measurements were queried for glucose, HbA1c, HDL-C, insulin, LDL-C, total cholesterol, and triglycerides. Records were also queried for calculated LDL-C. Prescription medication is available in the SD through MedEx [43 (link)], an algorithm that extracts medications and their signature mentions from free-text entries available in the EMR.
For each adult patient (>18 years), median values were calculated for a) measurements taken when no medications are prescribed (“pre-medication” values) and b) measurements taken at first mention of medication and post mention of medication (“post-medication” values). We used the following medication class and list to identify lipid measurements determined at the time or after patients were prescribed lipid-lowering medications
Electronic phenotyping has been previously described for BMI in EAGLE BioVU [42 ]. To extract lipid and type 2 diabetes traits from EAGLE BioVU EMRs, laboratory measurements were queried for glucose, HbA1c, HDL-C, insulin, LDL-C, total cholesterol, and triglycerides. Records were also queried for calculated LDL-C. Prescription medication is available in the SD through MedEx [43 (link)], an algorithm that extracts medications and their signature mentions from free-text entries available in the EMR.
For each adult patient (>18 years), median values were calculated for a) measurements taken when no medications are prescribed (“pre-medication” values) and b) measurements taken at first mention of medication and post mention of medication (“post-medication” values). We used the following medication class and list to identify lipid measurements determined at the time or after patients were prescribed lipid-lowering medications
Statins (also known as HMG CoA reductase inhibitors, atorvastatin (Lipitor®), fluvastatin (Lescol®), lovastatin (Mevacor®, Altoprev™), pravastatin (Pravachol®), rosuvastatin calcium (Crestor®), simvastatin (Zocor®), lovastatin + niacin (Advicor®), atorvastatin + amlodipine (Caduet®), and simvastatin + ezetimibe (Vytorin™)
Selective cholesterol absorption inhibitors (ezetimibe (Zetia®))
Resins (cholestyramine (Questran®, Questran® Light, Prevalite®, Locholest®, Locholest® Light), colestipol (Colestid®), colesevelam Hcl (WelChol®))
Fibrates (gemfibrozil (Lopid®), fenofibrate (Antara®, Lofibra®, Tricor®, and Triglide™), clofibrate (Atromid-S))
Niacin
