Abstract 270: Successful Development and Implementation of an Algorithm to Identify Resistant Hypertension in the Electronic Medical Record
Resistant hypertension (RH) affects 8.4 to 17.4% of the hypertensive population, and is associated with an elevated risk of end-organ damage. To identify genetic or environmental factors associated with RH, we developed an algorithm to phenotype patients using electronic medical records (EMRs).
We modified a previously validated algorithm for detecting patients with RH from 158,128 adults in BioVU, a DNA databank incorporating clinical data from EMRs. We defined patients as having RH if they had a SBP>140 mmHg or a DBP>90 mmHg for at least one month while receiving at least three anti-hypertensive agents including 25 mg HCTZ or equivalent doses of another thiazide diuretic or amlodipine 5 or 10 mg, or equivalent doses of a similar calcium channel blocker. Patients could be controlled after addition of a fourth drug. Patients were defined as having controlled hypertension if they carried a diagnosis of hypertension, were taking at least one and no more than two anti-hypertensive medications, and had a SBP<135 and a DBP<90 mmHg on repeated measurement. Drug exposures were identified from electronic-prescribing tools and MedEx.
To validate the algorithm, we conducted an independent review of 150 EMRs from patients classified as having resistant or controlled hypertension. The accuracy of the algorithm was 94.7%. Among 55,477 patients with hypertension in BioVU, 14.2% met the criteria for RH. The table shows the characteristics of patients with resistant and controlled hypertension.
This study demonstrates the successful use of a phenotyping algorithm to identify subjects with RH from EMR data and provides important characteristics of patients with RH in a real-world clinical setting.
Author Disclosures: M.M. Shuey: None. J.C. Denny: None. T.L. Edwards: None. N.J. Brown: None.
- © 2014 by American Heart Association, Inc.