Chest · Nov 18, 2026
Atherosclerotic cardiovascular disease risk prediction in individuals with and without impaired spirometry
By Krishnan S, Oelsner E, Smith B et al.
BACKGROUND: Cardiovascular disease risk scores likely underestimate risk in individuals with impaired spirometry (airflow obstruction and preserved ratio impaired spirometry (PRISm)). RESEARCH QUESTION: How do risk scores perform between individuals with impaired and normal spirometry, and can they be improved by adding spirometry and dyspnea? STUDY DESIGN AND METHODS: Ten-year incidence of major cardiovascular events (MACE) ie., myocardial infarction, stroke or death from coronary heart disease, from 7 US population-based cohorts was used for this analysis. The pooled cohorts' equations' (PCE) ability to predict MACE was compared between normal and impaired spirometry. Cox-models adding spirometry, dyspnea to PCE variables were evaluated for improvement in MACE prediction. RESULTS: Among 26523 participants, those with impaired (N=7547) compared with normal (N=18932) spirometry had a higher incidence of MACE (14% vs 7%), hazard ratio 95%CI 1.25[1.17,1.34] adjusted for PCE variables. Discrimination using area under the receiver operating characteristic curve (AUC)95%CI of the PCE for MACE was 0.792[0.780,0.803] vs 0.740[0.724,0.755] in normal vs impaired spirometry. Discrimination of MACE improved slightly by adding to the PCE: dyspnea plus 1)impaired spirometry 0.792[0.783,0.800], 2)PRISm, mild, moderate+ airflow obstruction 0.792[0.783,0.801], 3)FEV₁%predicted, FEV₁/FVC splines 0.793[0.784,0.802], compared with PCE variables 0.790[0.781,0.798]. The differences in AUC[95%CI] were: 1)0.002[0.001,0.003]p=0.002, 2)0.002[0.001,0.004]p=0.002, 3)0.003[0.002,0.005]p=0.0005 respectively. Among participants with impaired spirometry, only the model that added FEV₁%predicted, FEV₁/FVC continuous and dyspnea to the PCE showed improvement in AUC[95%CI], 0.750[0.735,0.765] vs 0.746[0.730,0.761] for PCE variables, difference of 0.004[0.001,0.008]p=0.008. There were no differences among models in calibration, sensitivity, specificity, overall performance, and after bootstrapping 1000 samples. INTERPRETATION: Performance of the PCE was better among individuals with normal vs impaired spirometry, despite the latter having an excess burden of MACE. Adding spirometry and dyspnea to PCE variables showed a small statistically significant improvement in MACE prediction, which may not be meaningful clinically, so other risk factors should be considered.
doi.org/10.1016/j.chest.2026.03.049PubMed 43051450Biomarkers