The following is the summary of “Development and validation of a survival prediction model in elder patients with community-acquired pneumonia: a MIMIC-population-based study” published in the January 2023 issue of Pulmonary medicine by Li, et al.
The goal of this study was to create a model for estimating the likelihood of death during hospitalization for elderly ICU admissions due to community-acquired pneumonia (CAP) admitted to the intensive care unit (ICU). From 2001-2012, information on 619 individuals with CAP aged 65 years was culled from the Medical Information Mart for Intensive Care III (MIMIC III) database for this cohort study. The sample dataset was arbitrarily split into a training set and a testing set to determine the stability of predictor variables (ratio: 6.5:3.5). Multivariate logistic regression was used to assess the predictors, and a model for the future was built based on the results. Positive and negative predictive value, accuracy, area under the curve, and 95% CI were used to compare the prediction model with the commonly used assessments of Sequential Organ Failure Assessment (SOFA), Pneumonia Severity Index (PSI), systolic blood pressure, oxygenation, age, and respiratory rate (SOAR), and CURB-65 scores (CI).
A decision curve analysis (DCA) was performed to evaluate the prediction model’s overall usefulness. Pathogen-based categorization was created for study purposes. Following a median of 8 days, 90 of the 402 elderly CAP patients in the training set died in the hospital during their 30-day stay. In-hospital mortality after 30 days was found to be predictably influenced by age, respiratory rate, international normalized ratio, ventilation, vasopressor use, red cell distribution width/blood urea nitrogen ratio, and Glasgow coma scale. The area under the curve (AUC) for the prediction model, which included the SOFA, SOAR, PSI, and CURB-65 scores, ranged from 0.751 (95% CI: 0.749-0.752) to 0.672 (95% CI: 0.670-0.674) to 0.607 (95% CI: 0.605-0.609) to 0.538 (95% CI: 0.533-0.554) to 0.645 (95% CI: 0.643-0.646).
According to the results of the DCA, patients with CAP who are admitted to the intensive care unit stand to gain more clinical net benefits from using the prediction model. The prediction model also showed improved predictive ability with respect to the pathogen. In addition, their algorithm could help doctors identify the high-risk population among elderly patients with CAP by predicting their 30-day inpatient mortality.
Source: bmcpulmmed.biomedcentral.com/articles/10.1186/s12890-023-02314-w