Multicenter study comparing case definitions used to identify patients with chronic obstructive pulmonary disease

Abstract

Rationale: Clinical trials in chronic obstructive pulmonary disease (COPD) usually require evidence of airflow obstruction and clinical risk factors. International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes or patient-reported physician diagnoses are often used for epidemiologic studies and performance improvement programs.

Objectives: To evaluate agreement between these case definitions for COPD and to assess the comparability of study populations identified as having COPD not using the clinical trial reference standard.

Methods: We recruited patients from the COPD Outcomes-based Network for Clinical Effectiveness and Research Translation multicenter clinical registry in a cross-sectional study. Demographics, clinical, and post-bronchodilator spirometry data were collected at an in-person study visit. The kappa statistic (κ) was used to evaluate agreement. A multivariable logistic regression model was used to identify patient characteristics associated with meeting the trial reference standard.

Measurements and main results: A total of 998 (82.8%) of 1,206 study participants met at least one case definition for COPD (of the 998: 91% using ICD-9 codes, 73% using patient-reported physician diagnosis, 56% using trial reference standard); agreement between case definitions was poor (κ = 0.20-0.26). Lack of airflow obstruction was the principal (89%) reason patients identified as having COPD did not meet the trial reference standard. Patients who were black (vs. white), obese (vs. normal weight), or had depression (vs. not) were less likely to meet the trial reference standard (odds ratio [95% CI], 0.37 [0.26-0.53], 0.51 [0.34-0.75], 0.53 [0.40-0.71], respectively).

Conclusions: Findings highlight concerns about the applicability of findings in clinical trials to patients meeting other case definitions for COPD.

Keywords: COPD; ICD-9-CM; case definitions; comparative effectiveness; spirometry.

Figure 1.

Percentage of participants identified by each International Classification of Diseases, Ninth Revision (ICD-9) code (n = 909). Most patients (84%) had multiple encounters, and ICD-9 codes may have varied across these encounters. A total of 54% of patients were identified by more than one ICD-9 code and 17% by three or more ICD-9 codes.

Figure 2.

Groups identified as having chronic obstructive pulmonary disease by the different methods (total n = 998). In this study population, 56% met the clinical trial reference standard (A + D + E + G), whereas 44% did not (B + C + F). Only 42% met all three case definitions (G). Overall there was poor agreement between the case definitions for chronic obstructive pulmonary disease: (1) clinical trial reference standard (A + D + E + G) versus patient-reported physician diagnosis (B + D + F + G), κ (95% confidence interval) = 0.26 (0.20–0.31); (2) clinical trial reference standard versus International Classification of Diseases, Ninth Revision (ICD-9) codes (C + E + F + G), κ = 0.20 (0.15–0.24); and (3) patient-reported physician diagnosis versus ICD-9 codes, κ = 0.25 (0.15–0.31).