Accuracy of administrative data for identifying patients with pneumonia

Area of Science:

• Health Services Research
• Medical Informatics
• Epidemiology

Background:

• Accurate identification of pneumonia cases in administrative databases is crucial for research and quality assessment.
• Existing claims-based algorithms, including International Classification of Diseases, Version 9, (ICD-9) and diagnosis-related group (DRG) codes, vary in their precision.
• The impact of these different definitions on patient characteristics and outcomes is not fully understood.

Purpose of the Study:

• To evaluate the accuracy of five distinct claims-based pneumonia definitions.
• To compare the performance of ICD-9 and DRG-based algorithms against a clinical reference standard.
• To assess the impact of these algorithms on patient-related variables and healthcare utilization.

Main Methods:

• Compared three ICD-9-based and two DRG-based case identification algorithms against a clinical pneumonia reference standard.
• Analyzed data from 10,748 patients, with 272 confirmed pneumonia cases.
• Evaluated sensitivity, positive predictive values, patient variables, hospital admission rates, mortality, length of stay, and costs.

Main Results:

• Sensitivity of claims-based algorithms ranged from 47.8% to 66.2%; positive predictive values ranged from 72.6% to 80.8%.
• ICD-9 algorithms did not significantly alter patient variables compared to the reference standard.
• DRG algorithms demonstrated significantly lower hospital admission rates, 30-day mortality, shorter length of stay, and reduced costs compared to ICD-9 algorithms.

Conclusions:

• Claims-based algorithms for pneumonia identification in administrative data are imprecise.
• While ICD-9 algorithms showed moderate accuracy, DRG-based algorithms are significantly less precise for identifying pneumonia patients.
• The choice of claims-based algorithm impacts the estimation of patient outcomes and healthcare resource utilization.