The impact of race-based bias on mortality prediction by pneumonia severity scoring systems
CHEST(2023)
Abstract
SESSION TITLE: Chest Infections Posters 2 SESSION TYPE: Original Investigation Posters PRESENTED ON: 10/10/2023 12:00 pm - 12:45 pm PURPOSE: There are known racial disparities in the morbidity and mortality associated with community-acquired pneumonia (CAP). Compared to white patients, black patients have higher mortality, are less likely to receive pneumococcal/influenza vaccination, and experience higher re-admission rates. Various pneumonia scoring systems have been proposed to assess mortality and direct patient care decisions. Scoring systems such as SOFA and Apache IVa overpredict mortality in black patients compared to white patients. The purpose of this study is to comprehensively evaluate the effect of race on inpatient mortality prediction by common pneumonia severity scoring systems. METHODS: This is a retrospective analysis of two cohorts of patients > 18 years of age who were diagnosed with CAP (SARS-CoV-2 CAP and non-SARS-CoV-2 CAP) in Louisville, KY. Research assistants screened patients for CAP at eight participating hospitals from 1 June 2014 – 31 May 2016 (non-SARS-CoV-2 CAP cohort) and 5 March 2020 – 31 March 2021 (SARS-CoV-2 CAP cohort). Demographics, vitals, and radiological/laboratory data were obtained from admission. Differences between groups with continuous data were assessed by Mann-Whitney U test and groups with categorical variables were analyzed by Fisher’s exact tests. ROC analysis predicting inpatient mortality was performed for the following scoring systems and comparisons were assessed between white and black patients: PSI , CURB-65, CORB, ATS/IDSA 2007, SMART COP, SCAP, and REA ICU. Analysis was performed with R. RESULTS: A total of 6,824 white patients and 1,886 black patients were diagnosed with CAP. Compared to white patients, black patients had an increased incidence of diabetes (45.1% vs 30.8%, p<0.001), heart failure (29.9% vs 25.5%, p<0.001), kidney disease (33.4% vs 26.5%, p<0.001). In-hospital mortality was higher in white patients compared to black patients (8.2% vs 6.7%, p=0.042). There was a wide range of discrimination ability between pneumonia severity scoring systems, with an AUC of 0.689 for REA ICU up to an AUC 0.83 for IDSA/ATS 2007. Comparison between white vs black patients, respectively, using ROC analysis showed AUC values (95% CI) for the following scores: PSI with 0.772 (0.751-0.791) vs 0.762 (0.717-0.805), CURB-65 with 0.717 (0.696-0.738) vs 0.744 (0.700-0.785), CORB with 0.707 (0.685-0.728) vs 0.712 (0.664-0.759), ATS/IDSA 2007 with 0.819 (0.799-0.837) vs 0.890 (0.859-0.917), SMART COP with 0.692 (0.666-0.714) vs 0.730 (0.681-0.776), and REA ICU with 0.689 (0.668-0.710) vs 0.683 (0.630-0.735). CONCLUSIONS: The ATS/IDSA 2007 scoring system predicted mortality better in Black patients than white patients. Otherwise, there were no differences in the discrimination of pneumonia severity scoring systems to predict in-hospital mortality in black and white patients. CLINICAL IMPLICATIONS: This study shows there is minimal impact of race on the discrimination of most pneumonia severity scoring models to predict in-hospital mortality from CAP. Our results suggest that well-validated scoring systems, such as PSI and CURB-65, can continue to be used to compare and assess mortality risk in black and white patient populations. DISCLOSURES: No relevant relationships by James Bradley No relevant relationships by Rodrigo Cavallazzi No disclosure on file for Thomas Chandler No disclosure on file for Stephen Furmanek No relevant relationships by Pooja Gandhi No relevant relationships by Julio Ramirez No relevant relationships by Matthew Wallace
MoreTranslated text
Key words
pneumonia,mortality,prediction,race-based
AI Read Science
Must-Reading Tree
Example
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
Generate MRT to find the research sequence of this paper
Chat Paper
Summary is being generated by the instructions you defined