Development and validation of a predictive model for pneumonia risk in ICU patients with stroke: A retrospective cohort study using the MIMIC-IV 3.0 database

Abstract

Background: Stroke is a leading cause of death and disability worldwide, with poststroke complications such as pneumonia significantly increasing mortality and healthcare burden. Existing models for predicting pneumonia risk in stroke patients have limitations, particularly in ICU settings.

Methods: This retrospective study utilized the Medical Information Mart for Intensive Care IV (MIMIC-IV) database to construct a nomogram predicting pneumonia risk in stroke patients within the ICU. The nomogram integrated multiple risk factors identified through univariate and multivariate logistic regression analyses.

Results: The study included 6542 stroke patients, with 11.5% developing pneumonia. Key predictive factors included breath rate, white blood cell, calcium, mean corpuscular hemoglobin concentration (MCHC), mechanical ventilation, antibiotics, pulmonary circulation disorders, metastatic cancer, and weight loss. The nomogram demonstrated good discrimination ability and calibration, with an AUC of 0.821 in the training set and 0.809 in the test set.

Conclusions: The nomogram provides a valuable tool for clinicians to assess pneumonia risk in stroke patients in the ICU, potentially improving patient outcomes and reducing the burden of pneumonia. Further research is needed for external validation and to incorporate additional variables.