BACKGROUND

Ventilator-associated pneumonia (VAP) is a common nosocomial infection in the intensive care unit (ICU), with high in-hospital mortality. Current scoring systems are limited in predicting nosocomial death of VAP. This study aimed to develop and validate a more accurate and effective prediction model for in-hospital mortality in ICU patients with VAP.

METHODS

This was a retrospective cohort study. The demographic and clinical data of 8,182 adult patients with VAP were extracted from the Medical Information Mart for Intensive Care (MIMIC-III) database. All patients were randomly classified as a training set (n=4,629) and a test set (n=1,984) with a ratio of 7:3. The outcome was in-hospital mortality and the follow-up was terminated at discharge. Univariate and multivariate logistic regression analyses were used to identify the independent predictors and develop the prediction model in the training set, and internal validation was carried out in the test set. The receiver operating characteristic (ROC) curve and calibration curve were plotted to evaluate the performance of the model.

RESULTS

Ethnicity, lung cancer history, septicemia history, hospital length of stay (LOS), fraction of inspired oxygen (FIO2) level, oxygen saturation (SPO2) level, Simplified Acute Physiology Score (SAPS II) score, Sequential Organ Failure Assessment (SOFA) score, and duration of invasive ventilation were all independently associated with the mortality of VAP. The algorithm of the prediction model was as follows: lnP/(1-P) = -0.700 + 0.493 Other Ethnicity + 0.789 Lung Cancer (Yes) + 0.693 Septicemia (Yes) - 0.074 Hospital LOS - 0.008 FIO2 - 0.032 SPO2 + 0.104 SOFA Score + 0.047 SAPS II + 0.004 Invasive Ventilation. The AUC was 0.837 in the training set and 0.817 in the test set, which indicated that the model performed well. The calibration curve also confirmed good calibration.

CONCLUSIONS

A model with good performance was developed to predict the individual death risk of VAP patients in the ICU, which might have the potential to provide ancillary data to support decision-making by physicians. External validation requires further evaluation of the model performance.