Abstract
Introduction
Guidelines recommend the use of risk stratifying tools to aid decision making in patients with Community Acquired Pneumonia (CAP). We aimed to determine if newer pneumonia severity scoring models predict patient centered clinical outcome better than older models in elderly patients?
Methods
We performed a retrospective cohort review of patients aged >65 who presented to the emergency room with CAP at our center between 1st January 2019 and 30th June 2022 using International Classification of Diseases, 10th Revision (ICD-10) codes. Primary outcomes of interest were whether patients died in-hospital or within 30-days of hospitalization, were readmitted within 30 days, required mechanical ventilation or vasopressor infusion, or were discharged to rehabilitation of a nursing facility. To compare the efficacy of the scoring systems, each score was used independently as the sole predictor in a series of logistic regression models. Model accuracy was compared within each outcome using the area under the curve (AUC) as the key outcome.
Results
The sample consisted of 257 patients, with male representing 49.81% of cohort and average age of 75 years (sd = 7.85 years). CHUBA had the highest AUC (0.679) in predicting 30-day mortality with the second highest score being the CURB-65 (0.651); AUC ranged from 0.618 with the CORB-75 to 0.679 with the CHUBA. This pattern was consistent when predicting in-hospital mortality with CHUBA being the most predictive (AUC = 0.758) by a small margin. Discharge to rehabilitation or nursing home was most accurately predicted by CHUBA (0.773). However, CHUBA performed poorly on predicting the need for vasopressors or mechanical ventilation.
Interpretation
In elderly patient population, the CHUBA prediction model may be superior to CURB-65 and PSI in determining the need for hospitalization, but not the need for ICU disposition. Larger prospective studies are needed to substantiate these claims.