Can we identify allergic rhinitis from administrative data: A validation study.

Area of Science:

• Epidemiology
• Health Informatics
• Pharmacovigilance

Background:

• Large-scale database studies are crucial for understanding allergic rhinitis prevalence, progression, and treatment.
• Accurate identification of allergic individuals within these databases is essential for reliable insights.
• Existing algorithms for identifying allergic rhinitis from health records require validation.

Purpose of the Study:

• To assess the validity of 13 algorithms for identifying adults with allergic rhinitis using Danish nationwide prescription and hospital data.
• To compare the performance of different algorithms based on sensitivity, specificity, and predictive values.

Main Methods:

• Utilized two Danish health examination studies (2006-2008 and 2012-2015) with a total of 10,652 adult participants.
• Established primary gold standard using serum specific IgE (≥0.35) and self-reported nasal symptoms.
• Secondary gold standard involved self-reported physician diagnosis of allergic rhinitis.
• Calculated sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for 13 register-based algorithms.

Main Results:

• All algorithms demonstrated low sensitivity (≤0.40) regardless of the gold standard or time period.
• Algorithms requiring both antihistamines and intranasal corticosteroids achieved the highest positive predictive values (PPV).
• For the primary gold standard in 2012-2015, the highest PPV was 0.69, with a corresponding sensitivity of 0.10.

Conclusions:

• Algorithms combining antihistamine and intranasal corticosteroid use show the highest PPVs for identifying allergic rhinitis.
• Despite improved PPVs, these algorithms exhibit moderate performance and low sensitivity when using a strict gold standard (sIgE and nasal symptoms).
• Further refinement of algorithms is needed for accurate identification of allergic rhinitis in large health databases.