Higher intraoperative arterial-to-end-tidal carbon dioxide gradients are associated with postoperative respiratory morbidity after lung cancer resection: a retrospective INSPIRE cohort study

BackgroundThe arterial-to-end-tidal carbon dioxide gradient may reflect impaired gas exchange during thoracic surgery. We examined its association with postoperative respiratory morbidity after lung cancer resection.MethodsWe conducted a retrospective cohort study using INSPIRE version 1.4.2. Each intraoperative arterial carbon dioxide (PaCO2) measurement was paired with the nearest positive end-tidal carbon dioxide (EtCO2) measurement within 10 min, and the primary exposure was each patient's median gradient. Respiratory morbidity was defined from new respiratory diagnoses and qualifying postoperative ventilation records during the index admission. We used multivariable regression, repeated-measurement analyses, sensitivity analyses, and exploratory internally validated prediction models.ResultsAmong 2,665 eligible patients, 85 (3.2%) had respiratory morbidity. A valid gradient was available in 2,527 patients; the primary complete-case analysis included 2,436 patients and 80 events. Each 5 mmHg higher gradient was associated with greater morbidity (adjusted odds ratio 1.53, 95% confidence interval 1.17–2.00; adjusted risk ratio 1.48, 1.14–1.91). Standardised risk increased from 2.28% at the first gradient quartile to 3.46% at the third, a difference of 1.18 percentage points (0.49–1.83). In a post hoc comparison, adding the gradient to a model already containing PaCO2 and clinical covariates did not materially change internally cross-validated discrimination (AUROC 0.7324 vs. 0.7325).ConclusionA higher intraoperative PaCO2–EtCO2 gradient was associated with postoperative respiratory morbidity after lung cancer resection, but its incremental prognostic value beyond PaCO2 was limited in exploratory analyses. Prospective external validation is required before clinical use.