International classification of diseases codes showed modest sensitivity for detecting community-acquired pneumonia

Objective: To estimate the sensitivity of International Classification of Diseases (ICD-9-CM) coding for detecting hospitalized community-acquired pneumonia and to assess possible determinants for misclassification. Study Design and Setting: Based on microbiological analysis data, 293 patients with a principal diagnosis of community-acquired pneumonia at seven hospitals in the Netherlands were assigned to three categories (pneumococcal pneumonia, pneumonia with other organism, or pneumonia with no organism specified). For these patients, the assigned principal and secondary ICD-9-CM codes in the hospital discharge record were retrieved and the corresponding sensitivity was calculated. Furthermore, pneumonia-related patient characteristics were compared between correctly and incorrectly coded subjects. Results: The overall sensitivity was 72.4% for the principal code and 79.5% for combined principal and secondary codes. For pneumococcal pneumonia (ICD-9-CM code 481) and pneumonia with specified organism (ICD-9-CM code 482-483), the sensitivities were 35% and 18.3%, respectively. Patient characteristics were not significantly different between correctly and incorrectly coded subjects except for duration of hospital stay, which correlated negatively with coding sensitivity (P = 0.01). Conclusion: ICD-9-CM codes showed modest sensitivity for detecting community-acquired pneumonia in hospital administrative databases, leaving at least one quarter of pneumonia cases undetected. Sensitivity decreased with longer duration of hospital stay. (C) 2007 Elsevier Inc. All rights reserved.