Development and Validation of a Predictive Model for Poor Pulmonary Rehabilitation Outcomes in Patients After Radical...

This study aims to develop and validate a nomogram model based on surgical parameters and clinical features for predicting the risk of poor pulmonary rehabilitation outcomes in patients with non-small cell lung cancer (NSCLC) after radical surgery.

This retrospective cohort study included 320 patients who underwent radical lung cancer surgery between June 2021 and December 2024. Patients were divided into a development cohort (n = 224) and a validation cohort (n = 96) at a 7:3 ratio. A composite surgical trauma score (STS) was developed to integrate key intraoperative variables. Independent predictors were identified through multivariate logistic regression analysis, including surgical trauma score, operative time, intraoperative blood loss, and other clinical factors. A nomogram model was constructed and validated using the bootstrap resampling (1000 repetitions). Model performance was assessed using the area under the curve (AUC) and calibration metrics.

The incidence of poor pulmonary rehabilitation was 38.4% in the development cohort and 36.5% in the validation cohort. In the development cohort, 11.2% of patients had a forced expiratory volume in 1 second (FEV1) recovery rate <80% alone, 15.2% had a modified Medical Research Council (mMRC) score ≥2 alone, and 12.0% met both criteria. Multivariable analysis predicted five independent variables: high versus low surgical trauma score (odds ratio (OR) = 3.09, 95% confidence interval (CI): 1.71-5.58), preoperative FEV1% predicted <70% (OR = 2.67, 95% CI: 1.53-4.67), operative time ≥180 minutes (OR = 2.52, 95% CI: 1.41-4.50), intraoperative blood loss ≥80 mL (OR = 2.25, 95% CI: 1.27-3.99), and age ≥65 years (OR = 1.92, 95% CI: 1.10-3.35). The nomogram demonstrated good discrimination, with an AUC of 0.84 (95% CI: 0.78-0.90) in the development cohort and 0.79 (95% CI: 0.70-0.88) in the validation cohort. Calibration was satisfactory in both cohorts (p = 0.23 in the development cohort; p = 0.31 in the validation cohort). Decision curve analysis revealed meaningful net benefits across a wide range of threshold probabilities. Predictive performance remained consistent across subgroups defined by age groups, surgical trauma strata, and tumor-node-metastasis (TNM) stages. Furthermore, sensitivity analyses using alternative endpoint definitions and various approaches for handling missing data further confirmed the robustness of the model.

We developed and validated a predictive nomogram for poor pulmonary rehabilitation outcomes after radical lung cancer surgery. This model incorporates readily available clinical and surgical parameters, providing individualized risk assessment that may facilitate personalized rehabilitation strategies and improve postoperative management.