Abstract
Aims
The 2007 GOLD classification identifies COPD patients based on lung function. Because lung function correlates poorly with clinically relevant outcomes, in 2011 risk and burden of disease were added to the model. Main objective of our study was to investigate whether the 2011 COPD GOLD model is a better predictor of mortality and morbidity than the 2007 classification. Furthermore, the predictive power of the GOLD 2011 classification with the modified medical research council scale (mMRC) versus Clinical COPD Questionnaire (CCQ) was compared.
Methods
This study was performed in 645 COPD patients from the COMIC study of whom all data for both classification models were known. Cox-regression models with C-statistic were calculated to compare the predictive power of all models.
Results (see figure 1)
Overall, C-statistics were higher in the 2011 models compared to the 2007 model in predicting all cause mortality. Using the CCQ with a cut-off of 1.5 showed higher C-statistics and better discrimination between the four categories (ABCD) than the cut-off of 1.0 and the mMRC. Similar results are observed for hospitalisations and pneumonias in COPD.
Conclusion
In COPD patients, the 2011 COPD GOLD classification is improved in predicting mortality, hospitalizations for COPD exacerbations and pneumoniaswhen compared to the 2007 classification and performs best when the CCQ with cut off of 1.5 is used.
- © 2014 ERS