Prediction models for exacerbations in different COPD patient populations: Results of five large databases

Background: Exacerbations are important outcomes in COPD both from a clinical and economic perspective. This study aimed to estimate prediction models for exacerbations to compare predictors between populations that differ in COPD severity. Methods: Members of an existing COPD health economic modelling network who had access to a large database (> 500 patients) with longitudinal data on exacerbations were asked to use two-third of the dataset to estimate several prediction models for total and severe exacerbations. The models differed in terms of predictors (depending on availability) and type of model. One-third of the dataset was used for validation. Results: Five COPD databases participated: two population-based studies (COPDGene and OLIN studies), one study in primary care patients (RECODE) and two clinical trials (ECLIPSE and UPLIFT). FEV1% predicted and previous exacerbations were significant predictors for total exacerbations in all five databases. Disease-specific quality of life and gender were predictors in 4 out of 4 and 3 out of 5 databases, respectively. Age was significant only in the two trials including severe COPD patients. Other significant predictors available in one database were cough, wheeze, pack-years and 6 MWD. Predictors for severe exacerbations were about the same, but in addition presence of cardiovascular disease and emphysema were found to be predictors in severe patients. Conclusion: FEV1% predicted, previous exacerbations and disease-specific quality of life were identified as predictors of exacerbations in COPD patients regardless of COPD severity, while age, cardiovascular disease and emphysema seemed to be predictors in severe patients only.