Unravelling the Information Contained in the Single Items of the COPD Assessment Test for Different Outcomes and Smoking Status in Patients with COPD: Results from COSYCONET.

Abstract

Background: The COPD Assessment Test (CAT) comprises eight questions. We evaluated the information that each of the questions and the total score contributed to outcomes and characteristics of chronic obstructive lung disease (COPD), including their dependence on smoking status.

Methods: Patients with COPD of the COSYCONET cohort with Global Initiative for Chronic Obstructive Lung Disease (GOLD) grades 1-4 and the former grade 0 were included. The evaluated outcomes included mortality, exacerbation risk, the comorbidities asthma, cardiac disease (coronary artery disease/heart failure), osteoporosis, and emphysema, for which a reduction in carbon monoxide transfer coefficient (KCO) <55% predicted was considered as marker. Analyses were performed by Cox proportional hazard or logistic multiple regression analyses separately for smokers and nonsmokers.

Results: In total, 2509 patients had complete data, among them 1884 nonsmokers (ex or never; 38.4% female; mean age±SD 66.1±8.5 years) and 625 current smokers (45.1% female, 61.6±7.9 years). The pattern of responses to the single questions of the CAT differed between outcome variables, as well as between smokers and nonsmokers, but in most cases the total score was superior to the single items. The CAT total score was associated with mortality (p<0.05) only in nonsmokers, while for exacerbation frequency/severity, it was of about equal importance in smokers and nonsmokers. Regarding KCO, the total score was indicative (p<0.05) only in nonsmokers. Particularly in smokers, single items could show opposite signs of their coefficients which therefore largely cancelled in the total score.

Conclusion: Our results show in detail for which outcomes single items are informative in nonsmokers and current smokers with COPD, overall being more informative in nonsmokers. Only regarding exacerbation risk, the predictive value was similar in both groups. These results might be helpful to extract as much as possible information from a COPD questionnaire that is often part of routine assessment.

Trial registration: NCT01245933.