B3-2: Validation of Administrative and Claims Data for the Identification of Anaphylaxis Cases in the Mini-Sentinel Distributed Database

Abstract

Background/AimsFew published studies have evaluated the validity of health plan administrative and claims data to identify anaphylaxis. Given the severity of the condition, drug-induced anaphylaxis is a major public health concern. We developed and evaluated the positive predictive value (PPV) of algorithms for identifying anaphylaxis in the Mini-Sentinel Distributed Database.MethodsWe conducted a retrospective study among members enrolled in 8 geographically diverse health plans (HealthCore, Inc.; Humana; three member health plans within the Kaiser Permanente Center for Effectiveness and Safety Research; and two member health plans in the HMO Research Network). Diagnosis and procedure codes were used to identify potential cases of anaphylaxis recorded in the Mini-Sentinel Distributed Database between January 1, 2009 and December 31, 2010. A random sample of medical charts (n = 131) was abstracted and adjudicated and positive predictive values (PPVs) were calculated based upon diagnostic criteria by Sampson et al.ResultsOf the 131 potential available cases, adjudicators determined that 77 were anaphylaxis, 45 were not anaphylaxis, and 9 were determined to have inadequate information to confirm whether or not a case was anaphylaxis. Overall, the PPV was 63.1% (95% confidence interval, 53.9%–71.7%). PPVs ranged from 48.1% to 78.9% across the Data Partners. In comparing different criteria, algorithms which included codes for allergy (ICD-9-CM 995.3) or adverse effects of drug, medicinal and biological substance (ICD-9-CM 995.2, E930-E949) in combination with additional treatment or symptom codes, such as injection of epinephrine, had a lower PPV (PPV = 45.8%) than algorithms that included codes specifically indicating anaphylaxis (ICD-9-CM 995.0, 999.4), although confidence intervals were wide and overlapped.ConclusionsWhile the PPV of the developed algorithm was higher than those reported in previously published studies, the PPV of anaphylaxis remains low. Further evaluation of individual codes and combinations of codes is ongoing as part of the present study.