Identifying the need for infection-related consultations in intensive care patients using machine learning models.

Infection-related consultations on intensive care units (ICU) have a positive impact on quality of care and clinical outcome. However, timing of these consultations is essential and to date they are typically event-triggered and reactive. Here, we investigate a proactive approach to identify patients in need for infection-related consultations by machine learning models using routine electronic health records. Data was retrieved from a mixed ICU at a large academic tertiary care hospital including 9684 admissions. Infection-related consultations were predicted using logistic regression, random forest, gradient boosting machines, and long short-term memory neural networks (LSTM). Overall, 7.8% of admitted patients received an infection-related consultation. Time-sensitive modelling approaches performed better than static approaches. Using LSTM resulted in the prediction of infection-related consultations in the next clinical shift (up to eight hours in advance) with an area under the receiver operating curve (AUROC) of 0.921 and an area under the precision recall curve (AUPRC) of 0.541. The successful prediction of infection-related consultations for ICU patients was done without the use of classical triggers, such as (interim) microbiology reports. Predicting this key event can potentially streamline ICU and consultant workflows and improve care as well as outcome for critically ill patients with (suspected) infections.

Impact of reduced antibiotic treatment duration on antimicrobial resistance in critically ill patients in the randomized controlled SAPS-trial.

Background: In the previously reported SAPS trial (https://clinicaltrials.gov/ct2/show/NCT01139489), procalcitonin-guidance safely reduced the duration of antibiotic treatment in critically ill patients. We assessed the impact of shorter antibiotic treatment on antimicrobial resistance development in SAPS patients. Materials and methods: Cultures were assessed for the presence of multi-drug resistant (MDR) or highly resistant organisms (HRMO) and compared between PCT-guided and control patients. Baseline isolates from 30 days before to 5 days after randomization were compared with those from 5 to 30 days post-randomization. The primary endpoint was the incidence of new MDR/HRMO positive patients. Results: In total, 8,113 cultures with 96,515 antibiotic test results were evaluated for 439 and 482 patients randomized to the PCT and control groups, respectively. Disease severity at admission was similar for both groups. Median (IQR) durations of the first course of antibiotics were 6 days (4-10) and 7 days (5-11), respectively (p = 0.0001). Antibiotic-free days were 7 days (IQR 0-14) and 6 days (0-13; p = 0.05). Of all isolates assessed, 13% were MDR/HRMO positive and at baseline 186 (20%) patients were MDR/HMRO-positive. The incidence of new MDR/HRMO was 39 (8.9%) and 45 (9.3%) in PCT and control patients, respectively (p = 0.82). The time courses for MDR/HRMO development were also similar for both groups (p = 0.33). Conclusions: In the 921 randomized patients studied, the small but statistically significant reduction in antibiotic treatment in the PCT-group did not translate into a detectable change in antimicrobial resistance. Studies with larger differences in antibiotic treatment duration, larger study populations or populations with higher MDR/HRMO incidences might detect such differences.

Better antimicrobial resistance data analysis and reporting in less time.

Objectives: Insights about local antimicrobial resistance (AMR) levels and epidemiology are essential to guide decision-making processes in antimicrobial use. However, dedicated tools for reliable and reproducible AMR data analysis and reporting are often lacking. We aimed to compare traditional data analysis and reporting versus a new approach for reliable and reproducible AMR data analysis in a clinical setting. Methods: Ten professionals who routinely work with AMR data were provided with blood culture test results including antimicrobial susceptibility results. Participants were asked to perform a detailed AMR data analysis in a two-round process: first using their software of choice and next using our newly developed software tool. Accuracy of the results and time spent were compared between both rounds. Finally, participants rated the usability using the System Usability Scale (SUS). Results: The mean time spent on creating the AMR report reduced from 93.7 to 22.4 min (P < 0.001). Average task completion per round changed from 56% to 96% (P < 0.05). The proportion of correct answers in the available results increased from 37.9% in the first to 97.9% in the second round (P < 0.001). Usability of the new tools was rated with a median of 83.8 (out of 100) on the SUS. Conclusions: This study demonstrated the significant improvement in efficiency and accuracy in standard AMR data analysis and reporting workflows through open-source software. Integrating these tools in clinical settings can democratize the access to fast and reliable insights about local microbial epidemiology and associated AMR levels. Thereby, our approach can support evidence-based decision-making processes in the use of antimicrobials.

Trends in Occurrence and Phenotypic Resistance of Coagulase-Negative Staphylococci (CoNS) Found in Human Blood in the Northern Netherlands between 2013 and 2019.

BACKGROUND: For years, coagulase-negative staphylococci (CoNS) were not considered a cause of bloodstream infections (BSIs) and were often regarded as contamination. However, the association of CoNS with nosocomial infections is increasingly recognized. The identification of more than 40 different CoNS species has been driven by the introduction of matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry. Yet, treatment guidelines consider CoNS as a whole group, despite increasing antibiotic resistance (ABR) in CoNS. This retrospective study provides an in-depth data analysis of CoNS isolates found in human blood culture isolates between 2013 and 2019 in the entire region of the Northern Netherlands. METHODS: In total, 10,796 patients were included that were hospitalized in one of the 15 hospitals in the region, leading to 14,992 CoNS isolates for (ABR) data analysis. CoNS accounted for 27.6% of all available 71,632 blood culture isolates. EUCAST Expert rules were applied to correct for errors in antibiotic test results. RESULTS: A total of 27 different CoNS species were found. Major differences were observed in occurrence and ABR profiles. The top five species covered 97.1% of all included isolates: S. epidermidis, S. hominis, S. capitis, S. haemolyticus, and S. warneri. Regarding ABR, methicillin resistance was most frequently detected in S. haemolyticus (72%), S. cohnii (65%), and S. epidermidis (62%). S. epidermidis and S. haemolyticus showed 50-80% resistance to teicoplanin and macrolides while resistance to these agents remained lower than 10% in most other CoNS species. CONCLUSION: These differences are often neglected in national guideline development, prompting a focus on 'ABR-safe' agents such as glycopeptides. In conclusion, this multi-year, full-region approach to extensively assess the trends in both the occurrence and phenotypic resistance of CoNS species could be used for evaluating treatment policies and understanding more about these important but still too often neglected pathogens.

Mapping twenty years of antimicrobial resistance research trends.

OBJECTIVE: Antimicrobial resistance (AMR) is a global threat to health and healthcare. In response to the growing AMR burden, research funding also increased. However, a comprehensive overview of the research output, including conceptual, temporal, and geographical trends, is missing. Therefore, this study uses topic modelling, a machine learning approach, to reveal the scientific evolution of AMR research and its trends, and provides an interactive user interface for further analyses. METHODS: Structural topic modelling (STM) was applied on a text corpus resulting from a PubMed query comprising AMR articles (1999-2018). A topic network was established and topic trends were analysed by frequency, proportion, and importance over time and space. RESULTS: In total, 88 topics were identified in 158,616 articles from 166 countries. AMR publications increased by 450% between 1999 and 2018, emphasizing the vibrancy of the field. Prominent topics in 2018 were Strategies for emerging resistances and diseases, Nanoparticles, and Stewardship. Emerging topics included Water and environment, and Sequencing. Geographical trends showed prominence of Multidrug-resistant tuberculosis (MDR-TB) in the WHO African Region, corresponding with the MDR-TB burden. China and India were growing contributors in recent years, following the United States of America as overall lead contributor. CONCLUSION: This study provides a comprehensive overview of the AMR research output thereby revealing the AMR research response to the increased AMR burden. Both the results and the publicly available interactive database serve as a base to inform and optimise future research.

The Visual Dictionary of Antimicrobial Stewardship, Infection Control, and Institutional Surveillance Data.

Objectives: Data and data visualization are integral parts of (clinical) decision-making in general and stewardship (antimicrobial stewardship, infection control, and institutional surveillance) in particular. However, systematic research on the use of data visualization in stewardship is lacking. This study aimed at filling this gap by creating a visual dictionary of stewardship through an assessment of data visualization (i.e., graphical representation of quantitative information) in stewardship research. Methods: A random sample of 150 data visualizations from published research articles on stewardship were assessed (excluding geographical maps and flowcharts). The visualization vocabulary (content) and design space (design elements) were combined to create a visual dictionary. Additionally, visualization errors, chart junk, and quality were assessed to identify problems in current visualizations and to provide improvement recommendations. Results: Despite a heterogeneous use of data visualization, distinct combinations of graphical elements to reflect stewardship data were identified. In general, bar (n = 54; 36.0%) and line charts (n = 42; 28.1%) were preferred visualization types. Visualization problems comprised color scheme mismatches, double y-axis, hidden data points through overlaps, and chart junk. Recommendations were derived that can help to clarify visual communication, improve color use for grouping/stratifying, improve the display of magnitude, and match visualizations to scientific standards. Conclusion: Results of this study can be used to guide data visualization creators in designing visualizations that fit the data and visual habits of the stewardship target audience. Additionally, the results can provide the basis to further expand the visual dictionary of stewardship toward more effective visualizations that improve data insights, knowledge, and clinical decision-making.

The Impacts of Deep Surgical Site Infections on Readmissions, Length of Stay, and Costs: A Matched Case-Control Study Conducted in an Academic Hospital in the Netherlands.

OBJECTIVE: This study aimed to evaluate the impacts of deep surgical site infections (dSSIs) regarding hospital readmissions, prolonged length of stay (LoS), and estimated costs. PATIENTS AND METHODS: We designed and applied a matched case-control observational study using the electronic health records at the University Medical Center Groningen in the Netherlands. We compared patients with dSSI and non-SSI, matched on the basis of having similar procedures. A prevailing topology of surgeries categorized as clean, clean-contaminated, contaminated, and dirty was applied. RESULTS: Out of a total of 12,285 patients, 393 dSSI were identified as cases, and 2864 patients without SSIs were selected as controls. A total of 343 dSSI patients (87%) and 2307 (81%) controls required hospital readmissions. The median LoS was 7 days (P25-P75: 2.5-14.5) for dSSI patients and 5 days (P25-P75: 1-9) for controls (p-value: <0.001). The estimated mean cost per hospital admission was €9,016 (SE±343) for dSSI patients and €5,409 (SE±120) for controls (p<0.001). Independent variables associated with dSSI were patient's age ≥65 years (OR: 1.334; 95% CI: 1.036-1.720), the use of prophylactic antibiotics (OR: 0.424; 95% CI: 0.344-0.537), and neoplasms (OR: 2.050; 95% CI: 1.473-2.854). CONCLUSION: dSSI is associated with increased costs, prolonged LoS, and increased readmission rates. Elevated risks were seen for elderly patients and those with neoplasms. Additionally, a protective effect of prophylactic antibiotics was found.

The phylogenetic landscape and nosocomial spread of the multidrug-resistant opportunist Stenotrophomonas maltophilia.

Recent studies portend a rising global spread and adaptation of human- or healthcare-associated pathogens. Here, we analyse an international collection of the emerging, multidrug-resistant, opportunistic pathogen Stenotrophomonas maltophilia from 22 countries to infer population structure and clonality at a global level. We show that the S. maltophilia complex is divided into 23 monophyletic lineages, most of which harbour strains of all degrees of human virulence. Lineage Sm6 comprises the highest rate of human-associated strains, linked to key virulence and resistance genes. Transmission analysis identifies potential outbreak events of genetically closely related strains isolated within days or weeks in the same hospitals.

Random glucose sampling as screening tool for diabetes among disadvantaged tuberculosis patients residing in urban slums in India.

Recently, a two-step diagnostic algorithm to diagnose diabetes among TB patients was proposed comprising random glucose and point-of-care HbA1c. This study evaluates the first part of this algorithm among disadvantaged TB patients. http://ow.ly/UI7d30nK1UN.