March 3, 2026Open Access

Development and Validation of a Nomogram to Predict Multidrug-Resistant Organism Infection in Severe Pneumonia: A Retrospective Cohort Study

Key Points

Multidrug-resistant organism (MDRO) infections are influenced by factors like types of pneumonia and COPD.
The modeling group's area under curve (AUC) reached 0.910, indicating strong predictive ability.
Logistic regression identified several risk factors including ICU admission and antibiotic use in patients.
The nomogram demonstrates high clinical value, ensuring effective risk evaluation for MDRO infections.

Abstract

Objective: The purpose of this study was to explore the etiological characteristics of multidrug-resistant bacteria (MDRO) infection in patients with severe pneumonia (SP) and to construct a nomogram model based on their risk factors. Methods: A total of 214 severe pneumonia (SP) patients admitted from November 2022 to January 2024 were retrospectively assigned to the modeling group, and 92 patients admitted from February 2024 to March 2025 formed the validation group. Patients were categorized into MDRO and non-MDRO groups based on multidrug-resistant organism infection status. The distribution of pathogens in the MDRO group was analyzed. Logistic regression was performed to analyze the influencing factors. R software was performed to construct nomogram models. Decision curve analysis(DCA) was performed to evaluate the clinical application value of nomogram models. Results: A total of 93 strains of pathogenic bacteria were isolated from 72 MDRO-infected patients. Gram-negative bacteria showed the highest resistance to piperacillin-tazobactam and piperacillin, while Gram-positive bacteria exhibited a high resistance rate to penicillin.Logistic analysis showed types of pneumonia, chronic obstructive pulmonary disease(COPD), invasive examination,intensive care unit(ICU) admission, antimicrobial drug combination, use of carbapenems, and hypoalbuminemia were risk factors (P< 0.05). The modeling group achieved an area under curve(AUC) of 0.910, with the Hosmer–Lemeshow (H-L) test showing χ 2 = 7.423. DCA indicated that the nomogram model had high clinical utility for predicted probabilities between 0.07 and 0.94. In the validation group, the AUC was 0.953, with H-L test χ 2 = 7.032. DCA showed that the nomogram model had high clinical value for predicted probabilities between 0.07 and 0.89. Conclusion: The types of pneumonia, COPD, invasive examination, ICU admission, combination of antimicrobial drugs, carbapenems use, and hypoalbuminemia are influencing factors for MDRO infection in SP patients. The nomogram constructed based on this can better evaluate the risk of MDRO infection. Keywords: severe pneumonia, multidrug resistant organism, risk factors, nomogram

Bookmark

View Full Paper