A 2-year recall of ≥1 moderate or severe exacerbation better predicted future high-risk COPD exacerbations than the 1-year standard (AUC 0.69 vs 0.66 in COPDGene, p<0.001).
Cohort (n=6,115)
Yes
Does a two-year recall of at least 1 moderate or 1 severe exacerbation improve the prediction of future high COPD exacerbation risk compared to the current 1-year standard?
A two-year recall period for any moderate or severe COPD exacerbation provides superior discriminative accuracy and clinical utility for predicting future high exacerbation risk compared to the current one-year standard.
Effect estimate: AUC (95% CI 0.67-0.71)
Absolute Event Rate: 0.69% vs 0.66%
p-value: p=<0.001
Abstract Introduction Treatment initiation or intensification to prevent exacerbation of chronic obstructive pulmonary disease (COPD) is based on the identification of patients with high exacerbation risk. The commonly used high-risk category of at least 2 moderate or 1 severe exacerbation within the prior 12 months has limited supporting evidence. We aimed to test the discriminative accuracy and assess the clinical utility of various COPD exacerbation categories for predicting future exacerbations. Methods In the COPDGene and NOVELTY cohorts, for each 1-year and 2-year recall periods, we estimated 6 distinct categories of exacerbation frequencies: ≥1 moderate (M1), ≥2 moderate (M2), ≥1 severe (S1), ≥1 moderate and ≥1 severe (M1andS1), ≥1 moderate or ≥ 1 severe (M1orS1), and ≥2 moderate or ≥ 1 severe (M2orS1), each ascertained in 3 ways: within 1 year, in each of 2 consecutive years (suffix E), and over a rolling combined 2-year period (suffix R). We used the area under the receiver operating characteristic curve (AUC) and decision curve analysis to evaluate the discriminative accuracy and clinical utility of these 18 categories for predicting the occurrence of M2orS1 (current standard) in the subsequent year. Results In COPDGene (n = 3,035), for the prediction of future M2orS1, baseline M1orS1R had the highest AUC (0.69, 95%CI 0.67-0.71) vs. baseline M2orS1 (0.66, 95%CI 0.64-0.67; Δ = 0.03;p0.001). In NOVELTY (n = 3,080), M1orS1R category had the highest AUC (0.87, 95%CI 0.85-0.88) vs. M2orS1 (0.75, 95%CI 0.72-0.77, Δ = 0.12;p0.001). Decision curve analysis demonstrated that the two-year rolling patterns provided the highest clinical utility across a clinically relevant treatment threshold range of 5% to 30% (Figure). M1orS1R also had the highest AUC for predicting any exacerbation (M1orS1) in both COPDGene (AUC = 0.68, 95%CI 0.66-0.70) and in NOVELTY (AUC = 0.86, 95%CI 0.85-0.88). Conclusions At least 1 moderate or 1 severe exacerbation over the previous 2 years has the highest discrimination and confers the highest clinical utility for predicting high COPD exacerbation risk. Overall, the combination of higher performance of various exacerbation history patterns in terms of their statistical (AUC) and clinical utility (net benefit) indicates that using a two-year recall and a lower threshold for high-risk classification (any moderate/severe events) is superior to the current standard of care. This abstract is funded by: This work was supported by NHLBI R01 HL151421 (SPB and AN), U01 HL089897 and U01 HL089856, by NIH contract 75N92023D00011, and by a Team Grant from the Canadian Institutes of Health Research (PHT 178432). COPDGene is also supported by the COPD Foundation through contributions made to an Industry Advisory Board that has included AstraZeneca, Bayer Pharmaceuticals, Boehringer Ingelheim, Genentech, GlaxoSmithKline, Novartis, Pfizer, and Sunovion. The NOVELTY study was funded by AstraZeneca.
Bhatt et al. (Fri,) conducted a cohort in chronic obstructive pulmonary disease (COPD) (n=6,115). 2-year recall of ≥1 moderate or ≥1 severe exacerbation (M1orS1R) vs. 1-year recall of ≥2 moderate or ≥1 severe exacerbation (M2orS1) was evaluated on occurrence of M2orS1 in the subsequent year (AUC, 95% CI 0.67-0.71, p=<0.001). A 2-year recall of ≥1 moderate or severe exacerbation better predicted future high-risk COPD exacerbations than the 1-year standard (AUC 0.69 vs 0.66 in COPDGene, p<0.001).