Physician adjudication of angioedema in patients with heart failure on angiotensin converting enzyme inhibitor therapy

Document Type

Conference Proceeding

Publication Date


Publication Title

Pharmacoepidemiol Drug Saf


Background: Angioedema, a potentially fatal adverse event of angiotensin- converting enzyme inhibitor (ACEI) therapy, occurs more often in patients with heart failure. Cohort studies conducted with healthcare databases can identify possible angioedema events during ACEI therapy using International Classification of Diseases (ICD) diagnosis codes. Few studies have undertaken physician adjudication of events to confirm angioedema and estimate the positive predictive value (PPV) of diagnosis codes. Objectives: Our objective was to calculate the PPV of ICD-9 and 10 diagnosis codes for angioedema according to physicians' confirmation of events after reviewing the text of health records. Methods: We included patients from five health plans in the United States (US) that contribute to the Cardiovascular Research Network (CVRN): Kaiser Permanente (KP) Northern California, KP Southern California, KPNorthwest,KPMid-Atlantic States and the Henry Ford Health System. We assembled a cohort of patients with heart failure based on diagnostic criteria. We identified incident users of ACEIs between July 2006 and December 2017.Wefollowed patients until the first diagnosis of angioedema or censoring: (1) discontinuation of ACEI therapy; (2) initiation of a different renin-angiotensin-aldosterone system (RAAS) blocking therapy; (3) completion of 365 days of ACEI therapy; (4) disenrollment from the health plan; (5) death; or, (6) end of the study on December 31, 2017. We identified angioedema using ICD-9 code 995.1 (Angioneurotic edema not elsewhere classified) or ICD-10 codes in the T78.3 series (Angioneurotic edema). Physicians reviewed coded angioedema events against the text of the electronic health record. To confirm a diagnosis, we required the documentation of signs or symptoms consistent with angioedema (e.g., facial swelling). We calculated the PPV as the number of confirmed events divided by all events with a diagnosis code. When an event could not be confirmed because of incomplete documentation, we classed the event as unconfirmed. Results: We observed 141 possible angioedema events in 45,483 patients (26,039 patient-years). Physicians confirmed 119 events. Twenty-two events were not confirmed, including 13 events with incomplete documentation. The PPV was 0.84 (95% confidence interval, 0.78 to 0.90). Conclusions: The PPV of an ICD-9 or 10 code for angioedema was high. Our cohort's PPV is consistent with a previous US cohort study-conducted 20 years earlier-of patients on ACEI therapy, which adjudicated events with similar clinical criteria.





First Page


Last Page


This document is currently not available here.