Development and external validation of a clinical prediction model for MRSA carriage at hospital admission in Southeast Lower Saxony, Germany

Identifying Importation and Asymptomatic Spreaders of Multi-drug Resistant Organisms in Hospital Settings

Healthcare-associated infections (HAIs) due to multi-drug resistant organisms (MDROs) are a significant burden to the healthcare system. Patients are sometimes already infected at the time of admission to the hospital (referred to as "importation"), and additional patients might get infected in the hospital through transmission ("nosocomial infection"). Since many of these importation and nosocomial infection cases may present no symptoms (i.e., "asymptomatic"), rapidly identifying them is difficult since testing is limited and incurs significant delays. Although there has been a lot of work on examining the utility of both mathematical models of transmission and machine learning for identifying patients at risk of MDRO infections in recent years, these methods have limited performance and suffer from different drawbacks: Transmission modeling-based methods do not make full use of rich data contained in electronic health records (EHR), while machine learning-based methods typically lack information about mechanistic processes. In this work, we propose N eur ABM, a new framework which integrates both neural networks and agent-based models (ABM) to combine the advantages of both modeling-based and machine learning-based methods. N eur ABM simultaneously learns a neural network model for patient-level prediction of importation, as well as the ABM model which is used for identifying infections. Our results demonstrate that N eur ABM identifies importation and nosocomial infection cases more accurately than existing methods.

Prevalence and factors associated with methicillin-resistant Staphylococcus aureus colonization on admission to geriatric care units: impact on screening practices

OBJECTIVES

Universal screening for methicillin-resistant Staphylococcus aureus (MRSA) entails additional costs, and there is no consensus for targeted screening for high-risk units. The aims of this study were to determine the prevalence of MRSA in geriatric care units, and to identify the factors associated with MRSA colonization on admission.

METHODS

This retrospective case-control study (1:1) in the geriatric care unit of six Belgian hospitals covered the period from 1st January 2021 to 31st December 2022. Cases were patients with a positive MRSA screening result within 48 h of admission to the geriatric care unit, and controls were patients with a negative screening result.

RESULTS

In total, 556 patients were included in this study (278 in each group). Prevalence per 100 admissions for the total sample was 2.3 [95% confidence interval (CI) 2.2-2.6]. Significant multi-variate factors associated with MRSA carriage on admission were: history of MRSA, nursing home origin, and chronic skin lesions. Applying these three factors would give an area under the receiver operating characteristic (ROC) curve of 0.73 (95% CI 0.71-0.77), and would allow screening to be carried out in only 55.4% of cases (95% CI 51.2-59.6%).

CONCLUSIONS

Using these factors as screening criteria in geriatric care units could significantly reduce the number of patients screened for MRSA, while maintaining satisfactory sensitivity and specificity.

Clinical prediction tools for identifying antimicrobial-resistant organism (ARO) carriage on hospital admissions: a systematic review

BACKGROUND

Increasing prevalence of antimicrobial-resistant organisms (AROs) is a growing economic and healthcare challenge. Increasing utilization of electronic medical record (EMR) systems and improvements in computation and analytical techniques afford an opportunity to reduce the spread of AROs through the development of clinical prediction tools to identify ARO carriers on admission to hospital.

AIM

To identify existing clinical prediction tools for meticillin-resistant Staphylococcus aureus (MRSA) and carbapenemase-producing organisms (CPOs), their predictive performance, and risk factors utilized in these tools.

METHODS

The CHARMS checklist was followed. Medline, EMBASE, Cochrane SR, CRD databases (DARE, NHS EED), CINAHL and Web of Science were searched from database inception to 26^th July 2021. Full-text articles were assessed independently, and quality assessment was conducted using the Prediction Model Risk of Bias Assessment Tool.

FINDINGS

In total, 3809 abstracts were identified and 22 studies were included. Among these studies, risk score models were the most common prediction tool (N=16). Previous admission, recent antibiotic exposure, age and sex were the most common risk factors for ARO carriage. Prediction tools were commonly evaluated on sensitivity and specificity with ranges of 15-100% and 46-98.6%, respectively, for MRSA, and 30-81.3% and 79.8-99.9%, respectively, for CPOs.

CONCLUSION

There is no gold standard ARO prediction tool. However, high-performance clinical prediction tools and identification of key risk factors for the early detection of AROs exist. Risk score models are easier to use and interpret; however, with recent improvements in machine learning techniques, highly robust models can be developed with data stored in an EMR.

Prospective surveillance of colonization and disease by methicillin-resistant Staphylococcus aureus (MRSA) at a European pediatric cancer center

PURPOSE

Children and adolescents undergoing treatment for cancer or allogeneic hematopoietic cell transplantation are at increased risk for methicillin-resistant Staphylococcus aureus (MRSA). We therefore examined the occurrence and outcome of MRSA colonization and infection in patients of a large European pediatric cancer center.

METHODS

In a prospective observational cohort study conducted between 2007 and 2018, nasopharyngeal swabs for culture of MRSA were obtained from all admitted patients. The primary endpoint of the study was the colonization rate over time. Secondary endpoints included genetic relatedness of isolates, time burden of isolation measures, and results of decolonization efforts.

RESULTS

During the study period, MRSA screening identified 34 colonized patients (median age: 10 years; range: 0-21) without trends over time. MRSA colonization was associated with the presence of classical risk factors. There was no molecular evidence of patient-to-patient transmission. A standard MRSA eradication regimen led to a lasting eradication of the organism in 26 of 34 patients. MRSA infection occurred in two patients with no associated fatalities.

CONCLUSION

Prospective monitoring revealed low rates of MRSA colonization and infection at our center. These low rates and the absence of patient-to-patient transmission support the effectiveness of the management bundle of MRSA identification, isolation, and decolonization.

Development and Validation of a Tool for the Prediction of Vancomycin-Resistant Enterococci Colonization Persistence-the PREVENT Score

Vancomycin-resistant enterococci (VRE) are nosocomial pathogens with increasing prevalence worldwide. Extensive hygiene measures have been established to prevent infection transmission in hospitals. Here, we developed a predictive score system (the predictive vancomycin-resistant enterococci [PREVENT] score) to identify the clearance or persistence in patients with a history of VRE carrier status at readmission. Over a cumulative 3-year period, patients with a positive VRE carrier status were included. The study population was recruited in two successive time periods and separated into training data for predictive score development and validation data for evaluation of the predictive power. The risk factors for persistent VRE colonization were analyzed in a univariable analysis before development of a logistic regression model based on the potential risk factors. The score points were determined proportionally to the beta coefficients of the logistic regression model. The data from 448 (79%) patients were used as the training data, and those from 119 (21%) as the validation data. Multivariable analysis revealed the following variables as independent risk factors: age of ≥60 years, hemato-oncological disease, cumulative antibiotic treatment for >4 weeks, and a VRE infection. The resulting logistic regression model exhibited an acceptable area under the curve (AUC) of 0.81 (95% confidence interval [CI], 0.72 to 0.91). The predictive score system had a sensitivity of 82% (95% CI, 65 to 93%) and a specificity of 77% (95% CI, 66 to 85%). The developed predictive score system is a useful tool to assess the VRE carrier status of patients with a history of VRE colonization. On the basis of this risk assessment, more focused and cost-effective infection control measures can be implemented. IMPORTANCE Given the increasing relevance of VRE as nosocomial pathogens worldwide, infection prevention and control measures, including patient isolation and contact precautions, are indispensable to avoid their spread in the hospital setting. In this study, we developed and validated the PREVENT score, a tool for rapid risk assessment of VRE persistence in patients with a history of previous VRE colonization. The score is designed to be easily performed, employing clinical information available in a regular admission setting and immediately providing information to inform the decision of whether to adopt patient isolation and contact precautions during the hospital stay. After validation, the score was shown to accurately identify patients with persistent VRE colonization upon admission, representing a suitable option as (i) a complementary method yielding preliminary results significantly more quickly than culture-based VRE detection techniques and (ii) an alternative strategy for VRE detection in settings in which microbiological VRE screening is not routinely performed due to limited resources.

Modeling transmission of pathogens in healthcare settings

PURPOSE OF REVIEW

Mathematical, statistical, and computational models provide insight into the transmission mechanisms and optimal control of healthcare-associated infections. To contextualize recent findings, we offer a summative review of recent literature focused on modeling transmission of pathogens in healthcare settings.

RECENT FINDINGS

The COVID-19 pandemic has led to a dramatic shift in the modeling landscape as the healthcare community has raced to characterize the transmission dynamics of SARS-CoV-2 and develop effective interventions. Inequities in COVID-19 outcomes have inspired new efforts to quantify how structural bias impacts both health outcomes and model parameterization. Meanwhile, developments in the modeling of methicillin-resistant Staphylococcus aureus, Clostridioides difficile, and other nosocomial infections continue to advance. Machine learning continues to be applied in novel ways, and genomic data is being increasingly incorporated into modeling efforts.

SUMMARY

As the type and amount of data continues to grow, mathematical, statistical, and computational modeling will play an increasing role in healthcare epidemiology. Gaps remain in producing models that are generalizable to a variety of time periods, geographic locations, and populations. However, with effective communication of findings and interdisciplinary collaboration, opportunities for implementing models for clinical decision-making and public health decision-making are bound to increase.