Construction and validation of a tracheostomy prediction model in mechanically ventilated stroke patients and the impact of early versus late tracheostomy on clinical outcomes: an IPTW-based analysis.

Stroke, a major global public health challenge, has become the second leading cause of death and the third leading cause of disability worldwide (). Data from the Global Burden of Disease (GBD) study () show that stroke causes approximately 7.3 million deaths annually. The incidence and disability burden remain on the rise in some regions, with a growing trend toward stroke occurring in younger adults under 55 years of age (, ). For critically ill stroke patients requiring prolonged mechanical ventilation support, tracheotomy is a commonly used and an important clinical intervention (). Studies have reported that the tracheotomy rate is approximately 15% among general ICU patients, while it can be as high as 35% in critically ill stroke patients (). Tracheotomy can reduce endotracheal intubation-related complications and improve patients’ ventilation tolerance and comfort. Tracheostomy is generally associated with a lower risk of ventilator-associated pneumonia (VAP) compared with long-term endotracheal intubation (). Despite its significant benefits in improving ventilation function, tracheotomy may increase the risk of lower respiratory tract infections, thereby adversely affecting patients’ prognosis and rehabilitation (). Previous studies have shown that only approximately one-third of stroke patients who undergo tracheotomy ultimately regain the ability to live independently (). In addition, the timing of tracheotomy is a crucial decision point. Emerging evidence suggests that different timings of tracheotomy exert varying effects on patients with severe stroke, but no unified clinical consensus has been reached to date (). Therefore, scientifically and accurately determining the optimal timing of tracheotomy and formulating individualized intervention strategies for mechanically ventilated stroke patients are of great significance for ensuring patient safety and improving long-term prognosis. Predictive models have demonstrated tremendous potential in the field of stroke, particularly in disease prediction, subtype classification, risk factor identification, and risk stratification (). The present study aims to construct and validate a tracheotomy risk prediction model specifically developed for mechanically ventilated stroke patients. Additionally, it intends to analyze the impact of early versus late tracheotomy on major in-hospital clinical outcomes of these patients, thereby providing a scientific basis for individualized clinical decision-making. A total of 508 mechanically ventilated stroke patients admitted to our hospital from January 2022 to January 2025 were enrolled in this retrospective study.