Development and Validation of a Predictive Model for Hypoxemia in the Postanesthesia Care Unit Following Thoracoscopic Lung Surgery: A Multicenter Cohort Study.

OBJECTIVE

This study was designed to identify risk factors for hypoxemia (SpO₂ ≤90% lasting >1 minute) in the postanesthesia care unit following thoracoscopic lung surgery and to develop a predictive model to facilitate early clinical intervention.

DESIGN

Retrospective multicenter cohort study.

SETTING

Two tertiary grade A hospitals in China.

PARTICIPANTS

A total of 2,563 adult patients (≥18 years; ASA I-III) undergoing elective thoracoscopic lung resection between January 2020 and January 2022 were included. Patients with preoperative hypoxemia; prior thoracic surgery; anesthetic contraindications: (1) uncontrolled coagulopathy (international normalized ratio >1.5 or platelets <80×10⁹/L), (2) documented malignant hyperthermia susceptibility, (3) severe allergy to protocol agents (propofol, sufentanil, cisatracurium, or sevoflurane), and (4) critical airway compromise necessitating awake intubation; or incomplete records were excluded.

INTERVENTIONS: Routine general anesthesia protocol including preoxygenation, intravenous induction (propofol, sufentanil, cisatracurium), double-lumen endotracheal intubation, lung-protective mechanical ventilation during one-lung ventilation, maintenance with remifentanil infusion and sevoflurane, and neuromuscular blockade reversal before extubation. Postextubation supplemental oxygen was administered as needed based on SpO₂ monitoring.

MEASUREMENTS AND MAIN

RESULTS

Patients from Center 1 (n = 1,668) were systematically allocated to training (n = 1,251) and internal validation (n = 417) cohorts; patients from Center 2 (n = 895) formed the external validation cohort. The incidence of hypoxemia was 28.38% in the training cohort. Independent predictors included age, body mass index, smoking history, forced expiratory volume ratio, exercise tolerance, postoperative pain severity, and delayed awakening. The nomogram demonstrated robust discrimination, with areas under the receiver operating characteristic curve of 0.841 (training), 0.864 (internal validation), and 0.812 (external validation). Decision curve analysis and clinical impact curves analyses confirmed superior clinical net benefit over pulse oximetry alone when the risk of hypoxemia exceeded 42%, supporting clinical utility.

CONCLUSIONS

Seven risk factors were independently associated with hypoxemia in the postanesthesia care unit following thoracoscopic lung surgery. The validated nomogram facilitates early identification of high-risk patients, enabling timely perioperative interventions to optimize recovery outcomes.