Discriminative performance and clinical utility of chronic obstructive pulmonary disease exacerbation categories for predicting future exacerbations
*Prospective cohort prediction study; Level 2 (OCEBM).
Citation
Bhatt SP, Adibi A, Bodduluri S, et al. American Journal of Respiratory and Critical Care Medicine. 2026;212(8):1721-1729. doi:10.1093/ajrccm/aamag118Background
Preventing chronic obstructive pulmonary disease flares depends on identifying patients at high future risk. Current guidance often uses a 1-year history of at least 2 moderate or 1 severe flare, but this cutoff has limited supporting evidence.Patients
Adults with chronic obstructive pulmonary disease in 2 prospective cohorts: COPDGene (n=3035) and NOVELTY (n=3080). COPDGene excluded major lung disease other than chronic obstructive pulmonary disease or asthma and required airflow obstruction; NOVELTY used physician diagnosis.Intervention
Risk classification using 18 prior flare-history categories, varying by number, severity, and 1-year versus 2-year recall.Control
Current standard: at least 2 moderate or at least 1 severe flare in the prior year.Outcome
Primary outcome: at least 2 moderate or at least 1 severe flare in the next 12 months. Prediction was assessed with area under the curve (0.5 to 1.0; higher is better) and net benefit.Follow-up Period
Three to 4 years, depending on cohort and analysis.Results
| Cohort | Best predictor | Area under curve | Improvement versus current standard |
|---|---|---|---|
| COPDGene | Any moderate or severe flare in past 2 years | 0.69 (95% CI 0.67 to 0.71) | Δ 0.03 (95% CI 0.01 to 0.05) |
| NOVELTY | Any moderate or severe flare in past 2 years | 0.87 (95% CI 0.85 to 0.88) | Δ 0.12 (95% CI 0.10 to 0.15) |
Decision analysis also favored 2-year recall across clinically plausible treatment thresholds. At a 20% risk threshold, the 2-year “any flare” rule yielded about 21 more correct treatment decisions per 1000 in COPDGene and 18 more per 1000 in NOVELTY than the current standard. Analyses used complete cases.