Predictors of Weaning Success in Mechanically Ventilated Adult Patients at the Kenyatta National Hospital

Abstract

Background: Mechanical ventilation is commonly used in the Intensive Care Unit (ICU) to sustain lung function. Upon reversal of the primary problem, it becomes a priority for critically ill patients to breathe spontaneously. Predicting successful spontaneous breathing trials is challenging to clinicians. Failed breathing trials are associated with increased mortality and ICU-acquired complications, such as ventilator associated pneumonia with subsequent prolonged ICU stay. Various indices such as airway occlusion pressure (P0.1), work of breathing/minute (WOB), central venous oxygen saturation (ScV02) and the rapid shallow breathing index (RSBI) have been used in isolation to predict spontaneous breathing trial success with variable accuracies. This study aims to analyze their sensitivity, specificity and accuracy in predicting successful spontaneous breathing trials (SBT) when combined. Study Objective: To demonstrate the sensitivity, specificity and accuracy of the various indices in predicting SBT success when combined. Study Design: observational cross-sectional study Study Site: Kenyatta National Hospital Critical Care Unit Study Population: 88 critically ill adult patients on mechanical ventilation at The Kenyatta National Hospital, Critical Care Unit Methodology and Measurements: Patients qualifying for spontaneous breathing trial (RSBI <105) were consented and recruited into the study. Baseline central venous oxygen saturation (ScvO2), P0.1 Index, work of breathing per minute (WOB/min) were recorded. Upon trial of spontaneous breathing (Pressure Support Ventilation) with optimal positive end expiratory pressure (PEEP), serial measurements of ScVO2, P0.1 Index and WOB/min were obtained and recorded at 30 minutes and 2 hours into the spontaneous breathing trial (SBT). Data Analysis: All numerical variables were non-parametric and were summarized into medians and standard deviation. Categorical variables were summarized into frequencies and percentages. Associations between outcome of interest and numeric variables were established using the Wilcoxon rank sum test and Chi-square test for outcome of interest and categorical variables, respectively. Results: The proportion of patients with weaning success was 75.0%. After checking for collinearity, ScVO2 69.5% (t=0), Delta ScVO2 -3.8% (t=30), WOB 17.5 J/L (t=30) and P0.1 - 4.25 (t=30) were established independent predictors of SBT success. They all had high sensitivity (>90%) and low specificity (<40%) in predicting SBT success. Following multivariate logistic regression, the sensitivity was 93.9% with a specificity of 59.1%.

P0.1(t=30) of >-4.25 and Delta ScVO2 (t=30) of >-3.8 were associated with increased odds of SBT success. Conclusion: Delta ScVO2, and P0.1 are independent predictors of weaning success. All have high sensitivity but remarkably low specificity with varied accuracy. When combined, all three indices improved the sensitivity, specificity and accuracy in predicting spontaneous breathing trial outcomes.