This study is a multicenter, prospective, observational trial to develop an ultrasound-based predictive model for extubation failure in mechanically ventilated ICU patients. Extubation failure, defined as re-intubation or death within 48 hours post-extubation, remains a significant challenge, occurring in 15-35% of cases. Prolonged mechanical ventilation is associated with increased morbidity, infection risk, and hospital costs. Traditional predictors, such as maximal inspiratory pressure (Pimax), rapid shallow breathing index (RSBI or Tobin index), and other clinical parameters, have shown limited accuracy in identifying patients who are ready for extubation.
Recent advances in bedside ultrasound have allowed for the assessment of key physiological functions, such as pulmonary aeration, diaphragmatic function, and cardiac performance, offering a more comprehensive approach to predicting extubation outcomes. In this study, ultrasound will be used to evaluate the modified Lung Ultrasound Score (LUS) for lung aeration, diaphragm excursion, and thickening fraction (TFdi) for diaphragmatic function, and left ventricular diastolic function (E/e') to assess cardiac performance.
Study Procedures The study will include approximately 15 intensive care units (ICUs) across Argentina, Chile, Peru, Ecuador, Italy, Spain, the United States, and Canada. Ethical approval has been obtained from each participating institution. Adult patients (≥18 years) on invasive mechanical ventilation (MV) for at least 48 hours and ready to initiate a spontaneous breathing trial (SBT) will be eligible. Ultrasound assessments will be performed between 20 and 30 minutes after initiating the trial.
Key ultrasound measures include:
Pulmonary Ultrasound (LUS score): Evaluation of anterior and lateral lung regions to assess loss of lung aeration during the SBT.
Cardiac Ultrasound: Assessment of systolic function using mitral annular plane systolic excursion (MAPSE) and tricuspid annular plane systolic excursion (TAPSE) to evaluate right ventricular (RV) and left ventricular (LV) systolic function. Diastolic function will be measured through the E/e' ratio.
Diaphragm Ultrasound: Evaluation of diaphragmatic function through excursion and thickening fraction.
Data from these ultrasound assessments will be collected prospectively and entered into a centralized database. Each patient may be included in the study for multiple extubations if applicable.
Quality Assurance and Data Management
The study employs rigorous data collection and validation procedures, including:
A quality assurance plan that ensures data accuracy and completeness, including site monitoring and auditing.
Data validation checks to ensure consistency with predefined rules. Source data verification through comparison with external data sources, such as medical records and case report forms.
A data dictionary detailing each variable\'s definition, coding, and range, to ensure uniform data entry across all sites.
Standard operating procedures (SOPs) guide the processes of patient recruitment, data collection, and data management.
A sample size assessment determined that the trial requires a sufficient number of patients to ensure the detection of statistically significant differences in extubation outcomes. Additionally, a plan for managing missing data has been developed to address cases where information is unavailable or inconsistent.
Statistical Analysis The primary goal of this study is to create a predictive model for extubation failure, based on lung, diaphragm, and cardiac ultrasound parameters. The primary outcome is extubation failure, defined as the need for re-intubation or death within 48 hours. Secondary outcomes include the predictive performance of the model in patients not requiring preventive non-invasive ventilation (NIV) and in neurocritical patients.
The statistical analysis plan will incorporate both univariate and multivariate analyses to determine the predictive value of each ultrasound parameter. Cox proportional hazard models will be employed to assess time-to-event data, while logistic regression models will be used for binary outcomes such as extubation failure.
By combining these diverse ultrasound measurements, the study aims to provide a robust, non-invasive tool for predicting extubation success or failure, thereby enabling more personalized, evidence-based management of critically ill patients in the ICU.
