Predicting probability of perirectal colonization with carbapenem-resistant Enterobacteriaceae (CRE) and other carbapenem-resistant organisms (CROs) at hospital unit admission

Machine learning-based antibiotic resistance prediction models: An updated systematic review and meta-analysis

BACKGROUND

The widespread use of antibiotics has led to a gradual adaptation of bacteria to these drugs, diminishing the effectiveness of treatments.

OBJECTIVE

To comprehensively assess the research progress of antibiotic resistance prediction models based on machine learning (ML) algorithms, providing the latest quantitative analysis and methodological evaluation.

METHODS

Relevant literature was systematically retrieved from databases, including PubMed, Embase and the Cochrane Library, from inception up to December 2023. Studies meeting predefined criteria were selected for inclusion. The prediction model risk of bias assessment tool was employed for methodological quality assessment, and a random-effects model was utilised for meta-analysis.

RESULTS

The systematic review included a total of 22 studies with a combined sample size of 43,628; 10 studies were ultimately included in the meta-analysis. Commonly used ML algorithms included random forest, decision trees and neural networks. Frequently utilised predictive variables encompassed demographics, drug use history and underlying diseases. The overall sensitivity was 0.57 (95% CI: 0.42-0.70; p< 0.001; I2= 99.7%), the specificity was 0.95 (95% CI: 0.79-0.99; p< 0.001; I2 = 99.9%), the positive likelihood ratio was 10.7 (95% CI: 2.9-39.5), the negative likelihood ratio was 0.46 (95% CI: 0.34-0.61), the diagnostic odds ratio was 23 (95% CI: 7-81) and the area under the receiver operating characteristic curve was 0.78 (95% CI: 0.74-0.81; p< 0.001), indicating a good discriminative ability of ML models for antibiotic resistance. However, methodological assessment and funnel plots suggested a high risk of bias and publication bias in the included studies.

CONCLUSION

This meta-analysis provides a current and comprehensive evaluation of ML models for predicting antibiotic resistance, emphasising their potential application in clinical practice. Nevertheless, stringent research design and reporting are warranted to enhance the quality and credibility of future studies. Future research should focus on methodological innovation and incorporate more high-quality studies to further advance this field.

Tackling the Antimicrobial Resistance "Pandemic" with Machine Learning Tools: A Summary of Available Evidence

Antimicrobial resistance is recognised as one of the top threats healthcare is bound to face in the future. There have been various attempts to preserve the efficacy of existing antimicrobials, develop new and efficient antimicrobials, manage infections with multi-drug resistant strains, and improve patient outcomes, resulting in a growing mass of routinely available data, including electronic health records and microbiological information that can be employed to develop individualised antimicrobial stewardship. Machine learning methods have been developed to predict antimicrobial resistance from whole-genome sequencing data, forecast medication susceptibility, recognise epidemic patterns for surveillance purposes, or propose new antibacterial treatments and accelerate scientific discovery. Unfortunately, there is an evident gap between the number of machine learning applications in science and the effective implementation of these systems. This narrative review highlights some of the outstanding opportunities that machine learning offers when applied in research related to antimicrobial resistance. In the future, machine learning tools may prove to be superbugs' kryptonite. This review aims to provide an overview of available publications to aid researchers that are looking to expand their work with new approaches and to acquaint them with the current application of machine learning techniques in this field.

A multi-center validation of the electronic health record admission source and discharge location fields against the clinical notes for identifying inpatients with long-term care facility exposure

Identifying long-term care facility (LTCF)-exposed inpatients is important for infection control research and practice, but ascertaining LTCF exposure is challenging. Across a large validation study, electronic health record data fields identified 76% of LTCF-exposed patients compared to manual chart review.

OBJECTIVE

Residence or recent stay in a long-term care facility (LTCF) is an important risk factor for antibiotic-resistant bacterial colonization. However, absent dedicated intake questionnaires or resource-intensive chart review, ascertaining LTCF exposure in inpatients is challenging. We aimed to validate the electronic health record (EHR) admission and discharge location fields against the clinical notes for identifying LTCF-exposed inpatients.

METHODS

We conducted a retrospective study of 1020 randomly sampled adult admissions between 2016 and 2021 across 12 University of Maryland Medical System hospitals. Using study-developed guidelines, we categorized the following data for LTCF exposure: each admission’s history & physical (H&P) note, each admission’s EHR-extracted “Admission Source,” and (3) the EHR-extracted admission and discharge locations for previous admissions (≤90 days). We estimated sensitivities, with 95% CIs, of H&P notes and of EHR admission/discharge location fields for detecting “current” and “any recent” (≤90 days, including current) LTCF exposure.

RESULTS

For detecting current LTCF exposure, the sensitivity of the index admission’s EHR-extracted “Admission Source” was 46% (95% CI: 35%–58%) and of the H&P note was 92% (83%–97%). For detecting any recent LTCF exposure, the sensitivity of “Admission Source” across the index and previous admissions was 32% (24%–41%), “Discharge Location” across previous admission(s) was 57% (47%–66%), and of the H&P note was 68% (59%–76%). The combined sensitivity of admission source and discharge location for detecting any recent LTCF exposure was 76% (67%–83%).

CONCLUSIONS

The EHR-obtained admission source and discharge location fields identified 76% of LTCF-exposed patients compared to chart review but disproportionately missed currently exposed patients.

A review of the mechanisms that confer antibiotic resistance in pathotypes of E. coli

The dissemination of antibiotic resistance in Escherichia coli poses a significant threat to public health worldwide. This review provides a comprehensive update on the diverse mechanisms employed by E. coli in developing resistance to antibiotics. We primarily focus on pathotypes of E. coli (e.g., uropathogenic E. coli) and investigate the genetic determinants and molecular pathways that confer resistance, shedding light on both well-characterized and recently discovered mechanisms. The most prevalent mechanism continues to be the acquisition of resistance genes through horizontal gene transfer, facilitated by mobile genetic elements such as plasmids and transposons. We discuss the role of extended-spectrum β-lactamases (ESBLs) and carbapenemases in conferring resistance to β-lactam antibiotics, which remain vital in clinical practice. The review covers the key resistant mechanisms, including: 1) Efflux pumps and porin mutations that mediate resistance to a broad spectrum of antibiotics, including fluoroquinolones and aminoglycosides; 2) adaptive strategies employed by E. coli, including biofilm formation, persister cell formation, and the activation of stress response systems, to withstand antibiotic pressure; and 3) the role of regulatory systems in coordinating resistance mechanisms, providing insights into potential targets for therapeutic interventions. Understanding the intricate network of antibiotic resistance mechanisms in E. coli is crucial for the development of effective strategies to combat this growing public health crisis. By clarifying these mechanisms, we aim to pave the way for the design of innovative therapeutic approaches and the implementation of prudent antibiotic stewardship practices to preserve the efficacy of current antibiotics and ensure a sustainable future for healthcare.

Machine learning to predict antimicrobial resistance: future applications in clinical practice?

INTRODUCTION

Machine learning (ML) is increasingly being used to predict antimicrobial resistance (AMR). This review aims to provide physicians with an overview of the literature on ML as a means of AMR prediction.

METHODS

References for this review were identified through searches of MEDLINE/PubMed, EMBASE, Google Scholar, ACM Digital Library, and IEEE Xplore Digital Library up to December 2023.

RESULTS

Thirty-six studies were included in this review. Thirty-two studies (32/36, 89 %) were based on hospital data and four (4/36, 11 %) on outpatient data. The vast majority of them were conducted in high-resource settings (33/36, 92 %). Twenty-four (24/36, 67 %) studies developed systems to predict drug resistance in infected patients, eight (8/36, 22 %) tested the performances of ML-assisted antibiotic prescription, two (2/36, 6 %) assessed ML performances in predicting colonization with carbapenem-resistant bacteria and, finally, two assessed national and international AMR trends. The most common inputs were demographic characteristics (25/36, 70 %), previous antibiotic susceptibility testing (19/36, 53 %) and prior antibiotic exposure (15/36, 42 %). Thirty-three (92 %) studies targeted prediction of Gram-negative bacteria (GNB) resistance as an output (92 %). The studies included showed moderate to high performances, with AUROC ranging from 0.56 to 0.93.

CONCLUSION

ML can potentially provide valuable assistance in AMR prediction. Although the literature on this topic is growing, future studies are needed to design, implement, and evaluate the use and impact of ML decision support systems.

Comparison of Daily versus Admission and Discharge Surveillance Cultures for Multidrug-Resistant Organism Detection in an Intensive Care Unit

BACKGROUND

Admission and discharge screening of patients for asymptomatic gut colonization with multidrug-resistant organisms (MDROs) is a traditional approach to active surveillance, but its sensitivity for detecting colonization is uncertain.

METHODS

Daily rectal or fecal swab samples and clinical data were collected over 12 months from patients in one 25-bed intensive care unit (ICU) in Chicago, IL USA and tested for the following multidrug-resistant organisms (MDROs): vancomycin-resistant enterococci (VRE); third-generation cephalosporin-resistant Enterobacterales, including extended-spectrum β-lactamase-producing Enterobacterales (ESBL); and carbapenem-resistant Enterobacterales (CRE). MDRO detection by (1) admission/discharge surveillance cultures or (2) clinical cultures were compared to daily surveillance cultures. Samples underwent 16S rRNA gene sequencing to measure the relative abundance of operational taxonomic units (OTUs) corresponding to each MDRO.

RESULTS

Compared with daily surveillance cultures, admission/discharge cultures detected 91% of prevalent MDRO colonization and 63% of incident MDRO colonization among medical ICU patients. Only a minority (7%) of MDRO carriers were identified by clinical cultures. Higher relative abundance of MDRO-associated OTUs and specific antibiotic exposures were independently associated with higher probability of MDRO detection by culture.

CONCLUSION

Admission and discharge surveillance cultures underestimated MDRO acquisitions in an ICU. These limitations should be considered when designing sampling strategies for epidemiologic studies that use culture-based surveillance.

Fecal carriage and genetic characteristics of carbapenem-resistant enterobacterales among adults from four provinces of China

Carbapenem-resistant Enterobacterales (CRE) is a global concern. This study investigated the prevalence of fecal colonization carriage and clonal dissemination of CRE among population in four provinces of China. A total of 685 stool samples were collected from four provinces in China. Among these samples, 141 and 544 were obtained from healthy and hospitalized individuals, respectively. The overall fecal carriage rate was 9.6% (65/685) with 4.26% (95% CI: 0.9-7.6) in healthy individuals and 10.84% (95% CI: 8.2-13.5) in hospitalized patients. The highest prevalence was in Henan province (18.35%, 95% CI: 9%-18.7%). Sixty-six CRE isolates were identified in Escherichia coli (56.06%, 37/66), Klebsiella (15.15%, 10/66), Citrobacter (13.63%, 9/66), Enterobacter (12.12%, 8/66), and Atlantibacter (1.51%, 1/66). All CRE strains carried carbapenemase genes and multiple antibiotics resistance genes, blaNDM-5 (77.27%, 51/66) was the most common carbapenemase gene, followed by blaNDM-1 (19.69%, 13/66). Antibiotic resistance genes, including blaIMP-4, and the colistin colistin resistance (mcr-1) gene were also identified. All CRE isolates belonged to different sequence types (STs). ST206 (36.84%, 14/38) in E. coli and ST2270 (60%, 6/10) in Klebsiella were significantly dominant clones. The results indicated the prevalence of CRE fecal carriage among adults of China, mostly blaNDM-producing E coli, which pose significant challenges for clinical management. Screening for CRE colonization is necessary to control infection.

Model-based evaluation of admission screening strategies for the detection and control of carbapenemase-producing Enterobacterales in the English hospital setting

BACKGROUND

Globally, detections of carbapenemase-producing Enterobacterales (CPE) colonisations and infections are increasing. The spread of these highly resistant bacteria poses a serious threat to public health. However, understanding of CPE transmission and evidence on effectiveness of control measures is severely lacking. This paper provides evidence to inform effective admission screening protocols, which could be important in controlling nosocomial CPE transmission.

METHODS

CPE transmission within an English hospital setting was simulated with a data-driven individual-based mathematical model. This model was used to evaluate the ability of the 2016 England CPE screening recommendations, and of potential alternative protocols, to identify patients with CPE-colonisation on admission (including those colonised during previous stays or from elsewhere). The model included nosocomial transmission from colonised and infected patients, as well as environmental contamination. Model parameters were estimated using primary data where possible, including estimation of transmission using detailed epidemiological data within a Bayesian framework. Separate models were parameterised to represent hospitals in English areas with low and high CPE risk (based on prevalence).

RESULTS

The proportion of truly colonised admissions which met the 2016 screening criteria was 43% in low-prevalence and 54% in high-prevalence areas respectively. Selection of CPE carriers for screening was improved in low-prevalence areas by adding readmission as a screening criterion, which doubled how many colonised admissions were selected. A minority of CPE carriers were confirmed as CPE positive during their hospital stay (10 and 14% in low- and high-prevalence areas); switching to a faster screening test pathway with a single-swab test (rather than three swab regimen) increased the overall positive predictive value with negligible reduction in negative predictive value.

CONCLUSIONS

Using a novel within-hospital CPE transmission model, this study assesses CPE admission screening protocols, across the range of CPE prevalence observed in England. It identifies protocol changes-adding readmissions to screening criteria and a single-swab test pathway-which could detect similar numbers of CPE carriers (or twice as many in low CPE prevalence areas), but faster, and hence with lower demand on pre-emptive infection-control resources. Study findings can inform interventions to control this emerging threat, although further work is required to understand within-hospital transmission sources.

Detection of Patients at Risk of Multidrug-Resistant Enterobacteriaceae Infection Using Graph Neural Networks: A Retrospective Study

Background: While Enterobacteriaceae bacteria are commonly found in the healthy human gut, their colonization of other body parts can potentially evolve into serious infections and health threats. We investigate a graph-based machine learning model to predict risks of inpatient colonization by multidrug-resistant (MDR) Enterobacteriaceae. Methods: Colonization prediction was defined as a binary task, where the goal is to predict whether a patient is colonized by MDR Enterobacteriaceae in an undesirable body part during their hospital stay. To capture topological features, interactions among patients and healthcare workers were modeled using a graph structure, where patients are described by nodes and their interactions are described by edges. Then, a graph neural network (GNN) model was trained to learn colonization patterns from the patient network enriched with clinical and spatiotemporal features. Results: The GNN model achieves performance between 0.91 and 0.96 area under the receiver operating characteristic curve (AUROC) when trained in inductive and transductive settings, respectively, up to 8% above a logistic regression baseline (0.88). Comparing network topologies, the configuration considering ward-related edges (0.91 inductive, 0.96 transductive) outperforms the configurations considering caregiver-related edges (0.88, 0.89) and both types of edges (0.90, 0.94). For the top 3 most prevalent MDR Enterobacteriaceae, the AUROC varies from 0.94 for Citrobacter freundii up to 0.98 for Enterobacter cloacae using the best-performing GNN model. Conclusion: Topological features via graph modeling improve the performance of machine learning models for Enterobacteriaceae colonization prediction. GNNs could be used to support infection prevention and control programs to detect patients at risk of colonization by MDR Enterobacteriaceae and other bacteria families.

Clinical Risk Factors and Microbiological and Intestinal Characteristics of Carbapenemase-Producing Enterobacteriaceae Colonization and Subsequent Infection

Gastrointestinal colonization with carbapenem-resistant Enterobacteriaceae (CRE) is always a prerequisite for the development of translocated infections. Here, we sought to screen for fecal carriage of CRE and identify the risk factors for CRE colonization as well as subsequent translocated pneumonia in critically ill patients admitted to the intensive care unit (ICU) of a university hospital in China. We further focused on the intestinal flora composition and fecal metabolic profiles in CRE rectal colonization and translocated infection patients. Animal models of gastrointestinal colonization with a carbapenemase-producing Klebsiella pneumoniae (carbapenem-resistant K. pneumoniae [CRKP]) clinical isolate expressing green fluorescent protein (GFP) were established, and systemic infection was subsequently traced using an in vivo imaging system (IVIS). The intestinal barrier, inflammatory factors, and infiltrating immune cells were further investigated. In this study, we screened 54 patients hospitalized in the ICU with CRE rectal colonization, and 50% of the colonized patients developed CRE-associated pneumonia, in line with the significantly high mortality rate. Upon multivariate analysis, risk factors associated with subsequent pneumonia caused by CRE in patients with fecal colonization included enteral feeding and carbapenem exposure. Furthermore, CRKP colonization and translocated infection influenced the diversity and community composition of the intestinal microbiome. Downregulated propionate and butyrate probably play important and multiangle roles in regulating immune cell infiltration, inflammatory factor expression, and mucus and intestinal epithelial barrier integrity. Although the risk factors and intestinal biomarkers for subsequent infections among CRE-colonized patients were explored, further work is needed to elucidate the complicated mechanisms. IMPORTANCE Carbapenem-resistant Enterobacteriaceae have emerged as a major threat to modern medicine, and the spread of carbapenem-resistant Enterobacteriaceae is a clinical and public health problem. Gastrointestinal colonization by potential pathogens is always a prerequisite for the development of translocated infections, and there is a growing need to assess clinical risk factors and microbiological and intestinal characteristics to prevent the development of clinical infection by carbapenem-resistant Enterobacteriaceae.

Predictive score for patients with carbapenemase-producing enterobacterales colonization upon admission in a tertiary care hospital in an endemic area

OBJECTIVES

Carbapenemase-producing Enterobacterales (CPE) comprise important nosocomial pathogens worldwide. Colonized patients are the source of further dissemination in healthcare settings. Considering that timely detection of CPE carriers is pivotal but universal screening is unfeasible, we aimed to develop and validate a prediction score to detect patients harbouring CPE on hospital admission.

METHODS

The study was conducted in a tertiary care hospital located in a CPE endemic area. Rectal swabs were obtained from 2303 patients, screened shortly after hospital admission. The Enterobacterales isolated in cultures were examined for the presence of blaVIM, KPC, NDM, OXA-48 by PCR. Demographic data and patient history of the previous 6 months were recorded. Risk factors for CPE carriage were identified using a multivariable logistic regression model and a points-system risk score was developed. The discriminative ability of the risk score was assessed using the AUC and its predictive performance was validated in a second dataset of 1391 patients in a different time period.

RESULTS

Seven predictors were identified: previous CPE colonization or infection, prior hospitalization, stay in a long-term health care facility, history of ≥2 interventions, renal replacement therapy, diabetes with end-organ damage and Karnofsky score. The developed risk score in the derivation dataset ranged between 0 and 79 points, with an AUC of 0.84 in the derivation and 0.85 in the validation dataset.

CONCLUSIONS

This prediction tool may assist in identifying patients who are at risk of harbouring CPE on hospital admission in an endemic area and guide clinicians to implement prompt and appropriate infection control measures.

Machine learning in predicting antimicrobial resistance: a systematic review and meta-analysis

INTRODUCTION

Antimicrobial resistance (AMR) is a global health threat; rapid and timely identification of AMR improves patient prognosis and reduces inappropriate antibiotic use.

METHODS

Relevant literature in PubMed, Web of Science, Embase and Institute of Electrical and Electronics Engineers prior to 28 September 2021 was searched. Any study that deployed machine learning (ML) or a risk score as a tool to predict AMR was included in the final review; there were 25 studies that employed the ML algorithm to predict AMR.

RESULTS

Extended spectrum β-lactamases, methicillin-resistant Staphylococcus aureus (MRSA) and carbapenem resistance were the most common outcomes in studies with a specific resistance pattern. The most common algorithms in ML prediction were logistic regression (n = 14 studies), decision tree (n = 14) and random forest (n = 7). The area under the curve (AUC) range for ML prediction was 0.48-0.93. The pooled AUC for ML prediction was 0.82 (0.78-0.85). Compared with risk score, higher specificity [87% (82-91) vs. 37% (25-51)] was indicated for ML prediction, but not sensitivity [67% (62-72) vs. 73% (67-79)].

CONCLUSIONS

Machine learning might be a potential technology for AMR prediction; however, retrospective methodology for model development, nonstandard data processing and scarcity of validation in a randomised controlled trial or real-world study limit the application of these models in clinical practice.

A Seven-Year Microbiological and Molecular Study of Bacteremias Due to Carbapenemase-Producing Klebsiella Pneumoniae: An Interrupted Time-Series Analysis of Changes in the Carbapenemase Gene's Distribution after Introduction of Ceftazidime/Avibactam

Background: Ceftazidime/avibactam (CZA) is a new option for the treatment of KPC-producing Klebsiella pneumoniae. The aim of this study was to determine resistance patterns and carbapenemase genes among K. pneumoniae (CP-Kp) bacteremic isolates before and after CZA introduction. Methods: K. pneumoniae from blood cultures of patients being treated in a Greek university hospital during 2015−21 were included. PCR for blaKPC, blaVIM, blaNDM and blaOXA-48 genes was performed. Results: Among 912 K. pneumoniae bacteremias: 725 (79.5%) were due to carbapenemase-producing isolates; 488 (67.3%) carried blaKPC; 108 (14.9%) blaVIM; 100 (13.8%) blaNDM; and 29 (4%) carried a combination of blaKPC, blaVIM or blaNDM. The incidence of CP-Kp bacteremias was 59 per 100,000 patient-days. The incidence of CP-Kp changed from a downward pre-CZA trend to an upward trend in the CZA period (p = 0.007). BSIs due to KPC-producing isolates showed a continuous downward trend in the pre-CZA and CZA periods (p = 0.067), while BSIs due to isolates carrying blaVIM or blaNDM changed from a downward trend in the pre-CZA to an upward trend in the CZA period (p < 0.001). Conclusions: An abrupt change in the epidemiology of CP-Kp was observed in 2018, due to the re-emergence of VIM-producing isolates after the suppression of KPC-producing ones via the use of CZA.

Epidemiological characteristics of carbapenem-resistant Enterobacteriaceae and carbapenem-resistant Acinetobacter baumannii in a tertiary referral hospital in Korea

Objectives

This study aimed to identify the epidemiological characteristics of patients with carbapenem-resistant Enterobacteriaceae and Acinetobacter baumannii (CRE/CRAB) isolates in a tertiary referral hospital in Korea.

Methods

We collected and analyzed data from 528 adults admitted to a tertiary referral hospital from August 1, 2018 to February 29, 2020. The CRE/CRAB isolates were confirmed as being present at the time of patients’ admission or acquired during hospitalization based on their medical records. The t-test, chi-square test, or Fisher exact test and stepwise multiple logistic regression were performed.

Results

While the proportion of community-acquired CRE/CRAB was low (6%), 20% of CRE/CRAB isolates were identified in patients at the time of hospitalization. The risk of CRAB isolation was positively associated with mechanical ventilator use (odds ratio [OR], 3.52; 95% confidence interval [CI], 1.96−6.33) and total parenteral nutrition use (OR, 3.64; 95% CI, 1.87−7.08).

Conclusion

Over 20% of CRE/CRAB isolates in a tertiary referral hospital in Korea were found at the time of patients’ admission. Furthermore, patients with mechanical ventilation and/or total parenteral nutrition tended to acquire CRAB more frequently. Thus, active surveillance for CRE/CRAB at the time of hospitalization is strongly required, particularly for patients who are expected to require mechanical ventilation or total parenteral nutrition.

Epidemiological Characteristics of Carbapenemase Producing Carbapenem-Resistant Enterobacteriaceae Colonization

PURPOSE

This study identified the epidemiological characteristics, including the size and major strains, of carbapenemase-producing carbapenem-resistant Enterobacteriaceae (CP-CRE) and CP-CRE-related factors by comparing the characteristics of patients in the CP-CRE and non-CP-CRE groups and the CP-CRE and non-CRE groups.

METHODS

This secondary data analysis study included 24 patients in the CP-CRE group, 113 patients in the non-CP-CRE group, and 113 in the non-CRE group. The size and type of CP-CRE were analyzed in terms of frequency and percentage, and CP-CRE risk factors were identified using multiple logistic regression analysis.

RESULTS

The rate of CP-CRE positivity among patients with CRE was 17.5%, and the most common causative organism in the CP-CRE group was Klebsiella pneumoniae (81.8%). There were no significant differences between patients in the CP-CRE and non-CP-CRE groups. When compared with the non-CRE group, the isolation of multidrug-resistant organisms except for CRE, particularly vancomycin resistant Enterococcus, was confirmed as a major risk factor.

CONCLUSION

To prevent CP-CRE acquisition, patients with multidrug-resistant organisms require treatment with more thorough adherence to CRE prevention and management guidelines.

Artificial intelligence and machine learning assisted drug delivery for effective treatment of infectious diseases

In the era of antimicrobial resistance, the prevalence of multidrug-resistant microorganisms that resist conventional antibiotic treatment has steadily increased. Thus, it is now unquestionable that infectious diseases are significant global burdens that urgently require innovative treatment strategies. Emerging studies have demonstrated that artificial intelligence (AI) can transform drug delivery to promote effective treatment of infectious diseases. In this review, we propose to evaluate the significance, essential principles, and popular tools of AI in drug delivery for infectious disease treatment. Specifically, we will focus on the achievements and key findings of current research, as well as the applications of AI on drug delivery throughout the whole antimicrobial treatment process, with an emphasis on drug development, treatment regimen optimization, drug delivery system and administration route design, and drug delivery outcome prediction. To that end, the challenges of AI in drug delivery for infectious disease treatments and their current solutions and future perspective will be presented and discussed.

StenoSCORE: Predicting Stenotrophomonas maltophilia Bloodstream Infections in the Hematologic Malignancy Population

Stenotrophomonas maltophilia bloodstream infections (BSI) are associated with considerable mortality in the hematologic malignancy population. Trimethoprim-sulfamethoxazole (TMP-SMX) is the treatment of choice; however, it is not routinely included in empirical treatment regimens, both because of its adverse event profile and the relative rarity of S. maltophilia infections. We developed a risk score to predict hematologic malignancy patients at increased risk for S. maltophilia BSI to guide early (TMP-SMX) therapy. Patients ≥12 years of age admitted to five hospitals between July 2016 and December 2019 were included. Two separate risk scores were developed, (i) a "knowledge-driven" risk score based upon previously identified risk factors in the literature in addition to variables identified by regression analysis using the current cohort, and (ii) a risk score based upon automatic variable selection. For both scores, discrimination (receiver operator characteristic [ROC] curves and C statistics) and calibration (Hosmer-Lemeshow goodness-of-fit test and graphical calibration plots) were assessed. Internal validation was assessed using leave-one-out cross-validation. In total, 337 unique patients were included; 21 (6.2%) had S. maltophilia BSI. The knowledge-driven risk score (acute leukemia, absolute neutrophil count category, mucositis, central line, and ≥3 days of carbapenem therapy) had superior performance (C statistic = 0.75; 0.71 after cross-validation) compared to that of the risk score utilizing automatic variable selection (C statistic = 0.63; 0.38 after cross-validation). A user-friendly risk score incorporating five variables easily accessible to clinicians performed moderately well to predict hematologic malignancy patients at increased risk for S. maltophilia BSI. External validation using a larger cohort is necessary to create a refined risk score before broad clinical application.

[External Validation of Carbapenem-Resistant Enterobacteriaceae Acquisition Risk Prediction Model in a Medium Sized Hospital]

PURPOSE

This study was aimed to evaluate the external validity of a carbapenem-resistant Enterobacteriaceae (CRE) acquisition risk prediction model (the CREP-model) in a medium-sized hospital.

METHODS

This retrospective cohort study included 613 patients (CRE group: 69, no-CRE group: 544) admitted to the intensive care units of a 453-beds secondary referral general hospital from March 1, 2017 to September 30, 2019 in South Korea. The performance of the CREP-model was analyzed with calibration, discrimination, and clinical usefulness.

RESULTS

The results showed that those higher in age had lower presence of multidrug resistant organisms (MDROs), cephalosporin use ≥ 15 days, Acute Physiology and Chronic Health Evaluation II (APACHE II) score ≥ 21 points, and lower CRE acquisition rates than those of CREP-model development subjects. The calibration-in-the-large was 0.12 (95% CI: - 0.16~0.39), while the calibration slope was 0.87 (95% CI: 0.63~1.12), and the concordance statistic was .71 (95% CI: .63~.78). At the predicted risk of .10, the sensitivity, specificity, and correct classification rates were 43.5%, 84.2%, and 79.6%, respectively. The net true positive according to the CREP-model were 3 per 100 subjects. After adjusting the predictors' cutting points, the concordance statistic increased to .84 (95% CI: .79~.89), and the sensitivity and net true positive was improved to 75.4%. and 6 per 100 subjects, respectively.

CONCLUSION

The CREP-model's discrimination and clinical usefulness are low in a medium sized general hospital but are improved after adjusting for the predictors. Therefore, we suggest that institutions should only use the CREP-model after assessing the distribution of the predictors and adjusting their cutting points.

Risk Factors Associated with Carbapenemase Producing Carbapenem-Resistant Enterobacteriaceae (CP-CRE) Positive Cultures in a Cohort of U.S. Veterans

INTRODUCTION

Carbapenem-resistant Enterobacterales (CRE) cause approximately 13,100 infections with 8% mortality in the United States annually. The subset of carbapenemase-producing CRE (CP-CRE) infections have much higher mortality rates (40% -50%). There has been little research on characteristics unique to CP-CRE. The goal of this study was to assess differences between those with nonCP-CRE and CP-CRE cultures in U.S. Veterans.

METHODS

A retrospective cohort of Veterans with CRE cultures from 2013-2018 and their demographic, medical, and facility level covariates were collected. Clustered multiple logistic regression models were used to assess independent factors associated with CP-CRE.

RESULTS

3,096 unique patients with cultures positive for either nonCP-CRE or CP-CRE were included. Being African American (Odds Ratio (OR)=1.44 (95% Confidence Interval (CI) 1.15,1.80), diagnosis in 2017 (OR=3.11 (95% CI 2.13,4.54)) or 2018 (OR=3.93 (95%CI 2.64,5.84)), congestive heart failure (OR=1.35 (95%CI 1.11,1.64)), and gastroesophageal reflux disease (OR=1.39 (95%CI 1.03,1.87)) were associated with CP-CRE cultures. 752 (24.3%) patients had no known antibiotic exposure in the year before culture; these individuals had a comparatively increased frequency of prolonged PPI use (17.3% vs 5.6%).

DISCUSSION

Among a cohort of patients with CRE, African Americans, individuals with congestive heart failure, and patients with gastroesophageal reflux disease had greater odds of having a CP-CRE culture. Roughly one in four patients with CP-CRE had no known antibiotic exposure in the year before their positive culture.

A Selective Culture Medium for Screening Carbapenem Resistance in Pseudomonas spp

Purpose: The SuperCP medium containing meropenem (2 mg/L) was evaluated for screening of carbapenem-resistant Pseudomonas species. Materials and Methods: It was evaluated using 29 meropenem-susceptible and 56 meropenem-nonsusceptible Pseudomonas-like clinical isolates, the latter exhibiting a variety of carbapenem-resistance mechanisms. Results: Its sensitivity and specificity of detection were found to be 91% and 100%, respectively. By testing spiked stools, an excellent performance of the medium was also observed for detection of carbapenem-resistant Pseudomonas aeruginosa, with a lowest detection limit ranging from 10⁰ to 10² CFU/mL. Conclusion: This screening medium provides the opportunity to select carbapenem-resistant Pseudomonas and Pseudomonas-related isolates regardless of their resistance mechanism.

Development of a Nomogram for Carbapenem-Resistant Enterobacteriaceae Acquisition Risk Prediction Among Patients in the Intensive Care Unit of a Secondary Referral Hospital

PURPOSE

This study aimed to identify the risk factors of carbapenem-resistant Enterobacteriaceae (CRE) acquisition to build a nomogram for CRE acquisition risk prediction and evaluate its performance.

METHODS

This unmatched case-control study included 352 adult patients (55 patients and 297 controls) admitted to the intensive care unit (ICU) of a 453-bed secondary referral hospital between January 1, 2018, and September 31, 2019, in Busan, South Korea. The nomogram was built with the identified risk factors using multiple logistic regression analysis. Its performance was analyzed using calibration-in-the-large, the slope of the calibration plot, concordance statistic (c-statistic), and the sensitivity and specificity of the training set, subsets, and a new test set.

RESULTS

The risk factors of CRE acquisition among ICU patients at a secondary referral hospital were Acute Physiology and Chronic Health Evaluation II score at the time of admission, use of a central venous catheter and a nasogastric tube, as well as use of cephalosporin antibiotics. At 20.0% of the predicted CRE acquisition risk in the training set, the calibration-in-the-large was 0, slope of the calibration plot was 1, c-statistic was .93, sensitivity was 85.5%, and specificity was 84.8%. The performance was relatively good in the subsets and new test set.

CONCLUSION

The nomogram can be used to monitor the CRE acquisition risk for ICU patients who have a similar case mix to patients in the study hospitals. Future studies need to involve more rigorous methodology and larger samples.

Clinical and Demographic Characteristics of Patients With a New Diagnosis of Carriage or Clinical Infection With Carbapenemase-Producing Enterobacterales: A Retrospective Study

Background: To examine the clinical, demographic, and microbiologic characteristics of new rectal carbapenemase-producing carbapenem-resistant Enterobacterales (CP-CRE) carriers vs. those with a clinical infection, hospitalized at Padeh-Poriya Medical Center between 2014 and 2017 and to examine the susceptibility profiles of isolates from clinical infections.

Methods: In this retrospective, chart analysis, demographic and clinical data were collected from medical charts of 175 adult patients with either new- onset carbapenemase-producing Enterobacterales (CPE) carriage or clinical CPE infection. Collected data included age, ethnic group, place of residence, hospitalizations in the past 90 days, and 30-day mortality. Microbiological analyses considered bacterial genus, molecular resistance mechanism and antibiotic susceptibility.

Results: A significantly higher percentage (42.4%) of CPE carriers were long-term care facility residents, and had been recently hospitalized (56.3%), as compared to patients with clinical CPE infection (29.2 and 45.9%, respectively). Additionally, we noted a high (58.3%) acquision of CPE in our hospital. The most common bacterial isolate was K. pneumoniae and the most common resistance mechanism was Klebsiella pneumoniae (K. pneumoniae) carbapenemases (KPC). High susceptibility rates to amikacin and chloramphenicol were also noted.

Conclusions: This study reaffirmed the importance of CPE screening and infection control measures. The observed antibiotic susceptibility profile suggests amikacin and chloramphenicol as potential treatments for CPE infection.

Prolonged Carriage of Carbapenemase-Producing Enterobacteriaceae: Clinical Risk Factors and the Influence of Carbapenemase and Organism Types

Prolonged carriage of carbapenemase-producing Enterobacteriaceae (CPE) constitutes a substantial epidemiologic threat. This study aimed to evaluate whether the types of carbapenemase and organism can affect the duration of carriage and to evaluate the clinical factors associated with prolonged carriage. We retrospectively reviewed data for patients admitted between May 2013 and August 2018 who were identified as CPE carriers. A total of 702 patients were identified; the major types of carbapenemase and organism were Oxacillinase (OXA)-48-like (n = 480, 68.4%) and Klebsiella pneumoniae (K. pneumoniae) (n = 584, 83.2%). The analyses of time to spontaneous decolonization using the Kaplan–Meier method showed that OXA-48-like and K. pneumoniae were significantly associated with prolonged carriage (log rank, p = 0.001 and p < 0.001). In multivariable logistic analysis to assess the risk factors for CPE prolonged carriage in the 188 patients with available follow-up culture data for 3 months, K. pneumoniae (adjusted odds ratio [aOR] 6.58; 95% confidence interval [CI], 1.05–41.27; p = 0.044), CPE positive clinical specimen (aOR 11.14; 95% CI, 4.73–26.25; p < 0.001), and concurrent Clostridioides difficile infection (CDI) (aOR 3.98, 95% CI 1.29–12.26; p = 0.016) were predictive of prolonged carriage. Our results suggest that CP-K. pneumoniae may have higher probability of prolonged carriage, while the effect of OXA-48-like CPE is inconclusive. Furthermore, patients with CP-K. pneumoniae who had positive clinical specimen or concurrent CDI can cause a vicious circle in prolonged carriage.

Independent, external validation of clinical prediction rules for the identification of extended-spectrum β-lactamase-producing Enterobacterales, University Hospital Basel, Switzerland, January 2010 to December 2016

Background

Algorithms for predicting infection with extended-spectrum β-lactamase-producing Enterobacterales (ESBL-PE) on hospital admission or in patients with bacteraemia have been proposed, aiming to optimise empiric treatment decisions.

Aim

We sought to confirm external validity and transferability of two published prediction models as well as their integral components.

Methods

We performed a retrospective case–control study at University Hospital Basel, Switzerland. Consecutive patients with ESBL-producing Escherichia coli or Klebsiella pneumoniae isolated from blood samples between 1 January 2010 and 31 December 2016 were included. For each case, three non-ESBL-producing controls matching for date of detection and bacterial species were identified. The main outcome measure was the ability to accurately predict infection with ESBL-PE by measures of discrimination and calibration.

Results

Overall, 376 patients (94 patients, 282 controls) were analysed. Performance measures for prediction of ESBL-PE infection of both prediction models indicate adequate measures of calibration, but poor discrimination (area under receiver-operating curve: 0.627 and 0.651). History of ESBL-PE colonisation or infection was the single most predictive independent risk factor for ESBL-PE infection with high specificity (97%), low sensitivity (34%) and balanced positive and negative predictive values (80% and 82%).

Conclusions

Applying published prediction models to institutions these were not derived from, may result in substantial misclassification of patients considered as being at risk, potentially leading to wrong allocation of antibiotic treatment, negatively affecting patient outcomes and overall resistance rates in the long term. Future prediction models need to address differences in local epidemiology by allowing for customisation according to different settings.

Evaluation of the Association Between Gastric Acid Suppression and Risk of Intestinal Colonization With Multidrug-Resistant Microorganisms: A Systematic Review and Meta-analysis

IMPORTANCE

Acid suppressants inhibit gastric acid secretion and disrupt the intestinal microbiome. Whether acid suppression increases the risk of colonization with multidrug-resistant microorganisms (MDROs) is unclear.

OBJECTIVES

To systematically examine the association of use of acid suppressants with the risk of colonization with MDROs and to perform a meta-analysis of current evidence.

DATA SOURCES

PubMed, Embase, the Web of Science Core Collection, and the Cochrane Central Register of Controlled Trials were searched from database inception through July 8, 2019.

STUDY SELECTION

Study selection was performed independently by 2 authors (R.P.J.W. and C.M.J.E.V.-G.) on the basis of predefined selection criteria; conflicts were resolved by consensus or by an adjudicator (K.v.D.). Human observational studies (case control, cohort, and cross-sectional) and clinical trial designs were selected if they quantified the risk of MDRO colonization in users of acid suppressants in comparison with nonusers.

DATA EXTRACTION AND SYNTHESIS

The Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) and Meta-analysis of Observational Studies in Epidemiology (MOOSE) recommendations were followed. Data were extracted independently by the same 2 authors, and adjudication was conducted when necessary. Risk of bias was assessed according to a modified Newcastle-Ottawa Scale. Pooled odds ratios (ORs) were estimated using random-effects models; heterogeneity was evaluated using the I2 method.

MAIN OUTCOMES AND MEASURES

The primary outcome measure was intestinal colonization with MDROs of the Enterobacterales order (producing extended-spectrum β-lactamases, carbapenemases, or plasmid-mediated AmpC β-lactamases), vancomycin-resistant enterococci, methicillin-resistant or vancomycin-resistant Staphylococcus aureus, or multidrug-resistant Pseudomonas or Acinetobacter species.

RESULTS

A total of 26 observational studies including 29 382 patients (11 439 [38.9%] acid suppressant users) met the selection criteria. Primary meta-analysis of 12 studies including 22 305 patients that provided adjusted ORs showed that acid suppression increased the odds of intestinal carriage of MDROs of the Enterobacterales order and of vancomycin-resistant enterococci by roughly 75% (OR = 1.74; 95% CI, 1.40-2.16; I2 = 68%). The odds were concordant with the secondary pooled analysis of all 26 studies (OR = 1.70; 95% CI, 1.44-1.99; I2 = 54%). Heterogeneity was partially explained by variations in study setting and the type of acid suppression.

CONCLUSIONS AND RELEVANCE

Acid suppression is associated with increased odds of MDRO colonization. Notwithstanding the limitations of observational studies, the association is plausible and is strengthened by controlling for confounders. In view of the global increase in antimicrobial resistance, stewardship to reduce unnecessary use of acid suppressants may help to prevent MDRO colonization.

Machine learning in infection management using routine electronic health records: tools, techniques, and reporting of future technologies

BACKGROUND

Machine learning (ML) is increasingly being used in many areas of health care. Its use in infection management is catching up as identified in a recent review in this journal. We present here a complementary review to this work.

OBJECTIVES

To support clinicians and researchers in navigating through the methodological aspects of ML approaches in the field of infection management.

SOURCES

A Medline search was performed with the keywords artificial intelligence, machine learning, infection∗, and infectious disease∗ for the years 2014-2019. Studies using routinely available electronic hospital record data from an inpatient setting with a focus on bacterial and fungal infections were included.

CONTENT

Fifty-two studies were included and divided into six groups based on their focus. These studies covered detection/prediction of sepsis (n = 19), hospital-acquired infections (n = 11), surgical site infections and other postoperative infections (n = 11), microbiological test results (n = 4), infections in general (n = 2), musculoskeletal infections (n = 2), and other topics (urinary tract infections, deep fungal infections, antimicrobial prescriptions; n = 1 each). In total, 35 different ML techniques were used. Logistic regression was applied in 18 studies followed by random forest, support vector machines, and artificial neural networks in 18, 12, and seven studies, respectively. Overall, the studies were very heterogeneous in their approach and their reporting. Detailed information on data handling and software code was often missing. Validation on new datasets and/or in other institutions was rarely done. Clinical studies on the impact of ML in infection management were lacking.

IMPLICATIONS

Promising approaches for ML use in infectious diseases were identified. But building trust in these new technologies will require improved reporting. Explainability and interpretability of the models used were rarely addressed and should be further explored. Independent model validation and clinical studies evaluating the added value of ML approaches are needed.

The Likelihood of Developing a Carbapenem-Resistant Enterobacteriaceae Infection during a Hospital Stay

Of 1,455 unique patients in U.S. intensive care units (ICUs), 4% were rectally colonized with CRE on admission. A total of 297 patients were initially negative for carbapenem-resistant Enterobacteriaceae (CRE) and remained in the ICU long enough to contribute additional swabs; 22% of these patients had a subsequent CRE-positive swab, with a median time to CRE colonization of 13 days (interquartile range, 7 to 21 days). Patients colonized with carbapenemase-producing CRE were more likely than those colonized with non-carbapenemase-producing CRE to develop CRE infections during their hospitalizations (36% versus 3%; P < 0.05).