Predictive Model for Multidrug Resistance in Patients Admitted to the Emergency Department With Sepsis

Overview

Introduction: Timely and accurate antibiotic administration in emergency department (ED) patients with sepsis or septic shock is vital, given mortality rates of 20% and over 40%, respectively. In high antimicrobial resistance (AMR) settings, selecting effective empirical antibiotics is challenging, requiring a balance between efficacy and minimizing multidrug-resistant organism (MDRO) emergence. A predictive model estimating AMR probability could optimize antibiotic use, improve outcomes, and reduce resistance. Although risk factors are known, no single validated model exists for predicting multidrug resistance in sepsis. Accurate prediction must integrate patient history, pathogen profiles, infection source, and antibiotic characteristics. Objectives: To estimate AMR prevalence in adult ED patients with sepsis or septic shock and develop a validated predictive model estimating AMR probability and likely pathogens. The model will follow a three-phase approach: (1) predict culture positivity, (2) estimate pathogen likelihood, and (3) predict AMR. Additionally, we aim to describe individual-level statistics for both predictable and unpredictable cases based on model performance. Methods: A cross-sectional study will be conducted at Hospital Italiano's adult ED over 70 months (Jan 1, 2017-Mar 20, 2020 and May 1, 2022-Aug 10, 2025), excluding the COVID-19 period. Primary outcomes include culture positivity, bacterial species, and MDRO prevalence. Frequency analyses will use positive cultures, species, and resistance classifications (MDRO, MDR, XDR, PDR), including mechanisms (e.g., MRSA, ESBL, KPC, MBL, OXA). Denominators will include all sepsis patients and, separately, culture-positive cases. Confidence intervals (95%) will be calculated using normal approximation. Multivariate logistic regression with backward stepwise selection will identify predictors and interactions. A hierarchical model will be developed based on culture results, pathogen identification, and resistance profiles.

Detailed description

Introduction:

Timely and accurate antibiotic administration in emergency department (ED) patients with sepsis or septic shock is vital, given mortality rates of 20% and over 40%, respectively. In high antimicrobial resistance (AMR) settings, selecting effective empirical antibiotics is challenging, requiring a balance between efficacy and minimizing multidrug-resistant organism (MDRO) emergence. A predictive model estimating AMR probability could optimize antibiotic use, improve outcomes, and reduce resistance. Although risk factors are known, no single validated model exists for predicting multidrug resistance in sepsis. Accurate prediction must integrate patient history, pathogen profiles, infection source, and antibiotic characteristics.

Objectives

In adult patients who present to an emergency department in a tertiary care center with sepsis or septic shock:

1-Prevalence and Associated Factors

1a- Estimate the prevalence of AMR/resistance patterns with clinical significance.

1b- Describe the predictive factors associated with AMR in this population.

1c- Generally, and in clinically relevant subgroups: by probable focus, clinically relevant pathogens, severity.

2- Generation and Validation of Predictive Models 2a- Generate and validate clinically useful predictive models to predict the probability of AMR.

2b- Generate and validate clinically useful predictive models to predict the probability of common/relevant pathogens.

2c- Evaluate the performance of stepwise predictive models in three stages: 1. Prediction of positive culture, 2. Intermediate prediction of pathogen, and 3. Prediction of AMR.

2d- Describe and evaluate point statistics on deterministic and unpredictable individuals based on the best predictive models.

Methods:

A cross-sectional study will be conducted at Hospital Italiano's adult ED over 70 months (Jan 1, 2017-Mar 20, 2020 and May 1, 2022-Aug 10, 2025), excluding the COVID-19 period. Primary outcomes include culture positivity, bacterial species, and MDRO prevalence. Frequency analyses will use positive cultures, species, and resistance classifications (MDRO, MDR, XDR, PDR), including mechanisms (e.g., MRSA, ESBL, KPC, MBL, OXA). Explanatory variables - Potential predictors of resistance include: Patient characteristics, Invasive devices, Immunosuppression, Comorbidities, Therapeutic adequacy, Medical history, Antibiotic use, Clinical status and Diagnostic studiesDenominators will include all sepsis patients and, separately, culture-positive cases. Confidence intervals (95%) will be calculated using normal approximation. Multivariate logistic regression with backward stepwise selection will identify predictors and interactions.

A hierarchical model will be developed based on culture results, pathogen identification, and resistance profiles. The sample will be randomly divided into a generation sample (2/3 of the sample) and a validation sample (1/3 of the sample). For the generation and validation of predictive models, the positive culture, each selected relevant bacteria, MDRO, MDR, XDR, PDR will be used as outcome variables.

List of Abbreviations (Abbreviation - Meaning) ABA - Acinetobacter baumannii AMR - Antimicrobial Resistance ESBL - Extended Spectrum Beta-Lactamase-producing Enterobacterales ESKAPE - Acronym summarizing the main clinically relevant resistant germs currently, each letter represents the initial of the scientific name of the bacterium: Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, Enterobacter spp.

CPE - carbapenemase-producing Enterobacteriaceae GNB - Gram-Negative Bacilli GPC - Gram-Positive Cocci KPC - Carbapenem-resistant Klebsiella pneumoniae MBL - Metallo-beta-lactamase MDR - Multidrug-resistant MOR - Multidrug-resistant Organisms MRSA - Methicillin-resistant Staphylococcus aureus OXA - Oxacillinase-type Carbapenemase PAE MR - Multidrug-resistant Pseudomonas aeruginosa PDR - Pan-resistant SOFA - Sepsis-related Organ Failure Assessment SSC - Surviving Sepsis Campaign VRE - Vancomycin-resistant Enterococci XDR - Extremely resistant

Primary outcome measures

• Proportion of episodes with multidrug-resistant organisms (MDRO) [Time frame: Baseline (within the first 48 hours of admission)]

Secondary outcome measures (7)

• Culture positivity [Time frame: Baseline]
• Bacterial species (Number of Participants with each microorganism species) [Time frame: Baseline]
• Enzymatic resistance mechanisms (Number of Participants with each enzymatic mechanism detected) [Time frame: Baseline]
• Carbapenemase genotypes (Number of participants with each genotypic resistance mechanism -KPC, NDM, VIM, OXA-48, IMP- detected by multiplex PCR assay) [Time frame: Baseline]
• Proportion of episodes with MDR organisms [Time frame: Baseline]
• Proportion of episodes with XDR organisms [Time frame: Baseline]
• Proportion of episodes with PDR organisms [Time frame: Baseline]

Eligibility criteria

Inclusion criteria

• Adults aged 18 years or older.
• Attended at the Adult Emergency Department of Hospital Italiano de Buenos Aires during the periods:

• January 1, 2017 - March 20, 2020, or
• May 1, 2022 - August 10, 2025.
• Sepsis or septic shock at presentation and at least 48 hours of observation or hospital admission.
• Bacterial cultures obtained during the initial evaluation.

Exclusion criteria

• No indication for antibiotic therapy within the first 48 hours of hospital admission.
• No bacterial cultures performed within the first 48 hours of hospital admission.
• SARS-CoV-2 infection diagnosed within the first 72 hours of hospital admission.

Criteria are shown verbatim from the registry (in English). Final eligibility is always assessed by the study center.

Healthy volunteers: No

Study design

Observational model

Other

Study locations

Center list to be confirmed — check the primary protocol.

Publications

• Jung K, Kashyap S, Avati A, Harman S, Shaw H, Li R, Smith M, Shum K, Javitz J, Vetteth Y, Seto T, Bagley SC, Shah NH. A framework for making predictive models useful in practice. J Am Med Inform Assoc. 2021 Jun 12;28(6):1149-1158. doi: 10.1093/jamia/ocaa318. PMID 33355350
• Rubin LG, Levin MJ, Ljungman P, Davies EG, Avery R, Tomblyn M, Bousvaros A, Dhanireddy S, Sung L, Keyserling H, Kang I; Infectious Diseases Society of America. 2013 IDSA clinical practice guideline for vaccination of the immunocompromised host. Clin Infect Dis. 2014 Feb;58(3):309-18. doi: 10.1093/cid/cit816. PMID 24421306
• Im Y, Kang D, Ko RE, Lee YJ, Lim SY, Park S, Na SJ, Chung CR, Park MH, Oh DK, Lim CM, Suh GY; Korean Sepsis Alliance (KSA) investigators. Time-to-antibiotics and clinical outcomes in patients with sepsis and septic shock: a prospective nationwide multicenter cohort study. Crit Care. 2022 Jan 13;26(1):19. doi: 10.1186/s13054-021-03883-0. PMID 35027073
• Taylor SP, Kowalkowski MA, Skewes S, Chou SH. Real-World Implications of Updated Surviving Sepsis Campaign Antibiotic Timing Recommendations. Crit Care Med. 2024 Jul 1;52(7):1002-1006. doi: 10.1097/CCM.0000000000006240. Epub 2024 Feb 22. PMID 38385751
• Rose N, Matthaus-Kramer C, Schwarzkopf D, Scherag A, Born S, Reinhart K, Fleischmann-Struzek C. Association between sepsis incidence and regional socioeconomic deprivation and health care capacity in Germany - an ecological study. BMC Public Health. 2021 Sep 7;21(1):1636. doi: 10.1186/s12889-021-11629-4. PMID 34493250
• Cornistein W, Santonato D, Novau PA, Fabbro LG, Jorge MF, Malvicini MA, Vilches V, Iudica FM. Synergy between infection control and antimicrobial stewardship programs to control carbapenem-resistant Enterobacterales. Antimicrob Steward Healthc Epidemiol. 2023 Sep 26;3(1):e162. doi: 10.1017/ash.2023.439. eCollection 2023. PMID 37771737
• Kherabi Y, Thy M, Bouzid D, Antcliffe DB, Rawson TM, Peiffer-Smadja N. Machine learning to predict antimicrobial resistance: future applications in clinical practice? Infect Dis Now. 2024 Apr;54(3):104864. doi: 10.1016/j.idnow.2024.104864. Epub 2024 Feb 12. PMID 38355048
• Peiffer-Smadja N, Delliere S, Rodriguez C, Birgand G, Lescure FX, Fourati S, Ruppe E. Machine learning in the clinical microbiology laboratory: has the time come for routine practice? Clin Microbiol Infect. 2020 Oct;26(10):1300-1309. doi: 10.1016/j.cmi.2020.02.006. Epub 2020 Feb 12. PMID 32061795

Identifiers

NCT: NCT07167173 · 7441

Primary sources (government registries)

• ClinicalTrials.gov (US NIH)

View this study on ClinicalTrials.gov ↗