Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Comparative Study
, 16 (6), 183-8

Identifying Patients With Physician-Diagnosed Asthma in Health Administrative Databases

Affiliations
Comparative Study

Identifying Patients With Physician-Diagnosed Asthma in Health Administrative Databases

Andrea S Gershon et al. Can Respir J.

Abstract

Background: Asthma imposes a heavy and expensive burden on individuals and populations. A population-based surveillance and research program based on health administrative data could measure and study the burden of asthma; however, the validity of a health administrative data diagnosis of asthma must first be confirmed.

Objective: To evaluate the accuracy of population-based provincial health administrative data in identifying adult patients with asthma for ongoing surveillance and research.

Methods: Patients from randomly selected primary care practices were assigned to four categories according to their previous diagnoses: asthma, chronic obstructive pulmonary disease, related respiratory conditions and nonasthma conditions. In each practice, 10 charts from each category were randomly selected, abstracted, then reviewed by a blinded expert panel who identified them as asthma or nonasthma. These reference standard diagnoses were then linked to the patients' provincial records and compared with health administrative algorithms designed to identify asthma. Analyses were performed using the concepts of diagnostic test evaluation.

Results: A total of 518 charts, including 160 from individuals with asthma, were reviewed. The algorithm of two or more ambulatory care visits and/or one or more hospitalization(s) for asthma in two years had a sensitivity of 83.8% (95% CI 77.1% to 89.1%) and a specificity of 76.5% (95% CI 71.8% to 80.8%).

Conclusion: Definitions of adult asthma using health administrative data are sensitive and specific for identifying adults with asthma. Using these definitions, cohorts of adults with asthma for ongoing population-based surveillance and research can be developed.

Comment in

Similar articles

See all similar articles

Cited by 82 PubMed Central articles

See all "Cited by" articles

Publication types

Feedback