LATE-BREAKING ABSTRACT: An algorithm for the identification of clinical COPD phenotypes in daily practice

Abstract

Background: Several cluster analyses in cohorts of patients with chronic obstructive pulmonary disease (COPD) have identified possible clinical phenotypes, but their results have not been translated for use in daily practice. Objectives: To develop simple clinical rules for allocating COPD patients to specific phenotypes. Methods: Data from 2409 COPD patients from a French/Belgian COPD cohort were analysed using cluster analysis, resulting in the identification of phenotypes validated using all-cause mortality at 3 years. Classification and regression trees (CARTs) were used for developing a set of clinical rules for allocating patients to these phenotypes. The clinical value of this set of clinical rules was further examined in an independent sample of 3651 COPD patients from the COPD Cohorts Collaborative International Assessment (3CIA) initiative. Results: Five phenotypes of COPD patients with different respiratory characteristics, age, comorbidities, and 3-year mortality were identified. A set of clinical rules that included cardiovascular comorbidities and/or diabetes, forced expiratory volume in 1 sec, dyspnoea grade, and age and body mass index allowed accurate classification of approximately 80% of patients. External validation of these clinical rules was obtained in the 3CIA cohorts, in which they identified subgroups of patients with different clinical characteristics and prognosis. Conclusion: A simple set of rules using easily available clinical data may allow the identification of clinically relevant COPD phenotypes.