Background: Allergic rhinitis (AR) is the most common allergic disease. Allergen immunotherapy (AI) is the only available treatment that modifies the allergic process instead of just suppressing symptoms, but it is associated with rare anaphylaxis. Monitoring new therapies for anaphylaxis signals requires knowledge of population-specific background incidence, but few estimates are available.
Objectives: To estimate the incidence rate (IR) of anaphylaxis in general and allergic populations.
Methods: This retrospective cohort study used administrative claims data from the HealthCore Integrated Research DatabaseSM for members age ≥ 5 years. Cohorts included (1) patients with neither AR nor AI (non-allergic cohort), (2) AR patients unexposed to AI (AR cohort), and (3) AI patients with or without AR. We identified anaphylaxis based on a revision of the Harduar-Morano algorithms. The IR of anaphylaxis was defined as the number of events in the study population divided by person-years (PY) at-risk. Patients could contribute > 1 event.
Results: There were 6,717,911 non-allergic, 3,237,972 AR, and 352,219 AI patients in the study. The IR of anaphylaxis per 10,000 PY was 2.5 (95% CI: 2.4–2.6) for the non-allergic cohort and ranged from 13.8 (95% CI: 13.6–14.1, AR cohort) to 36.0 (95% CI: 34.8–37.2, AI cohort) in the allergic groups. There were 14,081 (0.1%) patients with ≥ 1 anaphylaxis event; 1,122 (8.0%) had multiple events. Emergency department or inpatient encounters accounted for 69.7% of events in the non-allergic cohort, but only 39.1% and 32.5% in the AR and AI cohorts, respectively. Of the 3,331 events in the AI cohort, 39.0% occurred on the same day as AI administration. Incidence of anaphylaxis was consistently higher for patients with a history of anaphylaxis or asthma in the baseline period.
Conclusions: The claims-based results suggest that anaphylaxis rates are higher among allergic patients, especially those receiving AI. However, the high proportion of events presenting in outpatient settings suggests that the reported rates may be overestimated. Validation of the Harduar-Morano algorithms based on medical records adjudication, now underway, will better inform our understanding of these findings.
