An integrated pipeline for prediction of Clostridioides difficile infection.

With the expansion of electronic health records(EHR)-linked genomic data comes the development of machine learning-enable models. There is a pressing need to develop robust pipelines to evaluate the performance of integrated models and minimize systemic bias. We developed a prediction model of symptomatic Clostridioides difficile infection(CDI) by integrating common EHR-based and genetic risk factors(rs2227306/IL8). Our pipeline includes (1) leveraging phenotyping algorithm to minimize temporal bias, (2) performing simulation studies to determine the predictive power in samples without genetic information, (3) propensity score matching to control for the confoundings, (4) selecting machine learning algorithms to capture complex feature interactions, (5) performing oversampling to address data imbalance, and (6) optimizing models and ensuring proper bias-variance trade-off. We evaluate the performance of prediction models of CDI when including common clinical risk factors and the benefit of incorporating genetic feature(s) into the models. We emphasize the importance of building a robust integrated pipeline to avoid systemic bias and thoroughly evaluating genetic features when integrated into the prediction models in the general population and subgroups.

Multicenter evaluation of the BIOFIRE Joint Infection Panel for the detection of bacteria, yeast, and AMR genes in synovial fluid samples.

The bioMérieux BIOFIRE Joint Infection (JI) Panel is a multiplex in vitro diagnostic test for the simultaneous and rapid (~1 h) detection of 39 potential pathogens and antimicrobial resistance (AMR) genes directly from synovial fluid (SF) samples. Thirty-one species or groups of microorganisms are included in the kit, as well as several AMR genes. This study, performed to evaluate the BIOFIRE JI Panel for regulatory clearance, provides data from a multicenter evaluation of 1,544 prospectively collected residual SF samples with performance compared to standard-of-care (SOC) culture for organisms or polymerase chain reaction (PCR) and sequencing for AMR genes. The BIOFIRE JI Panel demonstrated a sensitivity of 90.9% or greater for all but six organisms and a positive percent agreement (PPA) of 100% for all AMR genes. The BIOFIRE JI Panel demonstrated a specificity of 98.5% or greater for detection of all organisms and a negative percent agreement (NPA) of 95.7% or greater for all AMR genes. The BIOFIRE JI Panel provides an improvement over SOC culture, with a substantially shorter time to result for both organisms and AMR genes with excellent sensitivity/PPA and specificity/NPA, and is anticipated to provide timely and actionable diagnostic information for joint infections in a variety of clinical scenarios.

Cepheid Xpert Xpress Flu/RSV evaluation performed by minimally trained non-laboratory operators in a CLIA-waived environment.

The COVID-19 pandemic highlighted the significance of readily available and easily performed viral testing for surveillance during future infectious pandemics. The objectives of this study were: to assess the performance of the Xpert Xpress Flu and/or RSV test, a multiplex PCR assay for detecting influenza A and B virus and respiratory syncytial virus nucleic acids in respiratory tract specimens, relative to the Quidel Lyra Influenza A+B assay and the Prodesse ProFlu+ assay, and the system's ease of use by minimally trained operators. Overall, the Xpert Xpress Flu/RSV test demonstrated a high positive and negative percent agreement with the comparator assays, and was easy to use and interpret results, based on the operators' feedback. We concluded that the Xpert Xpress Flu/RSV test is sensitive, specific, and easy to use for the diagnosis of influenza and RSV by minimally trained operators and can be a valuable tool in future infectious clusters or pandemics.

Prediction of Influenza Complications: Development and Validation of a Machine Learning Prediction Model to Improve and Expand the Identification of Vaccine-Hesitant Patients at Risk of Severe Influenza Complications.

Influenza vaccinations are recommended for high-risk individuals, but few population-based strategies exist to identify individual risks. Patient-level data from unvaccinated individuals, stratified into retrospective cases (n = 111,022) and controls (n = 2,207,714), informed a machine learning model designed to create an influenza risk score; the model was called the Geisinger Flu-Complications Flag (GFlu-CxFlag). The flag was created and validated on a cohort of 604,389 unique individuals. Risk scores were generated for influenza cases; the complication rate for individuals without influenza was estimated to adjust for unrelated complications. Shapley values were used to examine the model's correctness and demonstrate its dependence on different features. Bias was assessed for race and sex. Inverse propensity weighting was used in the derivation stage to correct for biases. The GFlu-CxFlag model was compared to the pre-existing Medial EarlySign Flu Algomarker and existing risk guidelines that describe high-risk patients who would benefit from influenza vaccination. The GFlu-CxFlag outperformed other traditional risk-based models; the area under curve (AUC) was 0.786 [0.783−0.789], compared with 0.694 [0.690−0.698] (p-value < 0.00001). The presence of acute and chronic respiratory diseases, age, and previous emergency department visits contributed most to the GFlu-CxFlag model's prediction. When higher numerical scores were assigned to more severe complications, the GFlu-CxFlag AUC increased to 0.828 [0.823−0.833], with excellent discrimination in the final model used to perform the risk stratification of the population. The GFlu-CxFlag can better identify high-risk individuals than existing models based on vaccination guidelines, thus creating a population-based risk stratification for individual risk assessment and deployment in vaccine hesitancy reduction programs in our health system.

At the Intersection of Gut Microbiome and Stroke: A Systematic Review of the Literature.

Background: Ischemic and hemorrhagic stroke are associated with a high rate of long-term disability and death. Recent investigations focus efforts to better understand how alterations in gut microbiota composition influence clinical outcomes. A key metabolite, trimethylamine N-oxide (TMAO), is linked to multiple inflammatory, vascular, and oxidative pathways. The current biochemical underpinnings of microbial effects on stroke remain largely understudied. The goal of our study is to explore the current literature to explain the interactions between the human gut microbiome and stroke progression, recovery, and outcome. We also provide a descriptive review of TMAO. Methods: A systematic literature search of published articles between January 1, 1990, and March 22, 2020, was performed on the PubMed database to identify studies addressing the role of the microbiome and TMAO in the pathogenesis and recovery of acute stroke. Our initial investigation focused on human subject studies and was further expanded to include animal studies. Relevant articles were included, regardless of study design. The analysis included reviewers classifying and presenting selected articles by study design and sample size in a chart format. Results: A total of 222 titles and abstracts were screened. A review of the 68 original human subject articles resulted in the inclusion of 24 studies in this review. To provide further insight into TMAO as a key player, an additional 40 articles were also reviewed and included. Our findings highlighted that alterations in richness and abundance of gut microbes and increased plasma TMAO play an important role in vascular events and outcomes. Our analysis revealed that restoration of a healthy gut, through targeted TMAO-reducing therapies, could provide alternative secondary prevention for at-risk patients. Discussion: Biochemical interactions between the gut microbiome and inflammation, resulting in metabolic derangements, can affect stroke progression and outcomes. Clinical evidence supports the importance of TMAO in modulating underlying stroke risk factors. Lack of standardization and distinct differences in sample sizes among studies are major limitations.

A Multicenter Clinical Study To Demonstrate the Diagnostic Accuracy of the GenMark Dx ePlex Blood Culture Identification Gram-Negative Panel.

Bacteremia can progress to septic shock and death without appropriate medical intervention. Increasing evidence supports the role of molecular diagnostic panels in reducing the clinical impact of these infections through rapid identification of the infecting organism and associated antimicrobial resistance genes. We report the results of a multicenter clinical study assessing the performance of the GenMark Dx ePlex investigational-use-only blood culture identification Gram-negative panel (BCID-GN), a rapid diagnostic assay for detection of bloodstream pathogens in positive blood culture (PBC) bottles. Prospective, retrospective, and contrived samples were tested. Results from the BCID-GN were compared to standard-of-care bacterial identification methods. Antimicrobial resistance genes (ARGs) were identified using PCR and sequence analysis. The final BCID-GN analysis included 2,444 PBC samples, of which 926 were clinical samples with negative Gram stain results. Of these, 109 samples had false-negative and/or -positive results, resulting in an overall sample accuracy of 88.2% (817/926). After discordant resolution, overall sample accuracy increased to 92.9% (860/926). Pre- and postdiscordant resolution sample accuracy excludes 37 Gram-negative organisms representing 20 uncommon genera, 10 Gram-positive organisms, and 1 Candida species present in 5% of samples that are not targeted by the BCID-GN. The overall weighted positive percent agreement (PPA), which averages the individual PPAs from the 27 targets (Gram-negative and ARG), was 94.9%. The limit of detection ranged from 104 to 107 CFU/ml, except for one strain of Fusobacterium necrophorum at 108 CFU/ml.

Mapping the Evidence on Rapid Diagnosis of Bloodstream Infections: A Scoping Review.

BACKGROUND: Laboratory and other healthcare professionals participate in developing clinical practice guidelines through systematic review of the evidence. A significant challenge is the identification of areas for analytic focus when the evidence consists of several categories of interventions and outcomes that span both laboratory and clinical processes. The challenge increases when these interventions present as sets of combined interventions. A scoping review may provide a transparent and defensible analytic route forward for systematic reviews challenged in this manner. CONTENT: A scoping review was carried out to characterize the evidence on rapid identification of bloodstream infections. Fifty-five studies previously identified by the supported systematic review were charted in duplicate. Charted records were analyzed using descriptive content analysis and evidence mapping with a 5-step process. SUMMARY: The 5-step analysis culminated in the characterization of 9 different intervention chain configurations that will facilitate the comparison of complex intervention practices across studies. Furthermore, our evidence map indicates that the current evidence base is strongly centered on 3 specific clinical outcomes, and it links these outcomes to the most represented intervention chain configurations. The scoping review effort generated a route forward for the supported systematic review and meta-analysis.

Stroke in SARS-CoV-2 Infection: A Pictorial Overview of the Pathoetiology.

Since the early days of the pandemic, there have been several reports of cerebrovascular complications during the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. Numerous studies proposed a role for SARS-CoV-2 in igniting stroke. In this review, we focused on the pathoetiology of stroke among the infected patients. We pictured the results of the SARS-CoV-2 invasion to the central nervous system (CNS) via neuronal and hematogenous routes, in addition to viral infection in peripheral tissues with extensive crosstalk with the CNS. SARS-CoV-2 infection results in pro-inflammatory cytokine and chemokine release and activation of the immune system, COVID-19-associated coagulopathy, endotheliitis and vasculitis, hypoxia, imbalance in the renin-angiotensin system, and cardiovascular complications that all may lead to the incidence of stroke. Critically ill patients, those with pre-existing comorbidities and patients taking certain medications, such as drugs with elevated risk for arrhythmia or thrombophilia, are more susceptible to a stroke after SARS-CoV-2 infection. By providing a pictorial narrative review, we illustrated these associations in detail to broaden the scope of our understanding of stroke in SARS-CoV-2-infected patients. We also discussed the role of antiplatelets and anticoagulants for stroke prevention and the need for a personalized approach among patients with SARS-CoV-2 infection.

Variants at the MHC Region Associate With Susceptibility to Clostridioides difficile Infection: A Genome-Wide Association Study Using Comprehensive Electronic Health Records.

Background: Clostridioides difficile is a major cause of healthcare-associated and community-acquired diarrhea. Host genetic susceptibility to Clostridioides difficile infection has not been studied on a large-scale. Methods: A total of 1,160 Clostridioides difficile infection cases and 15,304 controls were identified by applying the eMERGE Clostridioides difficile infection algorithm to electronic health record data. A genome-wide association study was performed using a linear mixed model, adjusted for significant covariates in the full dataset and the antibiotic subgroup. Colocalization and MetaXcan were performed to identify potential target genes in Clostridioides difficile infection - relevant tissue types. Results: No significant genome-wide association was found in the meta-analyses of the full Clostridioides difficile infection dataset. One genome-wide significant variant, rs114751021, was identified (OR = 2.42; 95%CI = 1.84-3.11; p=4.50 x 10-8) at the major histocompatibility complex region associated with Clostridioides difficile infection in the antibiotic group. Colocalization and MetaXcan identified MICA, C4A/C4B, and NOTCH4 as potential target genes. Down-regulation of MICA, upregulation of C4A and NOTCH4 was associated with a higher risk for Clostridioides difficile infection. Conclusions: Leveraging the EHR and genetic data, genome-wide association, and fine-mapping techniques, this study identified variants and genes associated with Clostridioides difficile infection, provided insights into host immune mechanisms, and described the potential for novel treatment strategies for Clostridioides difficile infection. Future replication and functional validation are needed.

Clinical evaluation of the acuitas® AMR gene panel for rapid detection of bacteria and genotypic antibiotic resistance determinants.

Urinary tract infections are leading causes of hospital admissions. Accurate and timely diagnosis is important due to increasing morbidity and mortality from antimicrobial resistance. We evaluated a polymerase chain reaction test (Acuitas AMR Gene Panel with the Acuitas Lighthouse Software) for detection of 5 common uropathogens (Escherichia coli, Klebsiella pneumoniae, Proteus mirabilis, Pseudomonas aeruginosa, Enterococcus faecalis) and antibiotic resistance genes directly from urine for prediction of phenotypic resistance. Overall percent agreement was 97% for semiquantitative detection of uropathogens versus urine culture using a cut-off of 104 colony forming units per mL urine. Overall accuracy was 91% to 93% for genotypic prediction of common antibiotic resistance harbored by E. coli, K. pneumoniae, and P. mirabilis.

Informing patients that they are at high risk for serious complications of viral infection increases vaccination rates.

For many vaccine-preventable diseases like influenza, vaccination rates are lower than optimal to achieve community protection. Those at high risk for infection and serious complications are especially advised to be vaccinated to protect themselves. Using influenza as a model, we studied one method of increasing vaccine uptake: informing high-risk patients, identified by a machine learning model, about their risk status. Patients (N=39,717) were evenly randomized to (1) a control condition (exposure only to standard direct mail or patient portal vaccine promotion efforts) or to be told via direct mail, patient portal, and/or SMS that they were (2) at high risk for influenza and its complications if not vaccinated; (3) at high risk according to a review of their medical records; or (4) at high risk according to a computer algorithm analysis of their medical records. Patients in the three treatment conditions were 5.7% more likely to get vaccinated during the 112 days post-intervention (p < .001), and did so 1.4 days earlier (p < .001), on average, than those in the control group. There were no significant differences among risk messages, suggesting that patients are neither especially averse to nor uniquely appreciative of learning their records had been reviewed or that computer algorithms were involved. Similar approaches should be considered for COVID-19 vaccination campaigns.

Early Detection of Septic Shock Onset Using Interpretable Machine Learners.

BACKGROUND: Developing a decision support system based on advances in machine learning is one area for strategic innovation in healthcare. Predicting a patient's progression to septic shock is an active field of translational research. The goal of this study was to develop a working model of a clinical decision support system for predicting septic shock in an acute care setting for up to 6 h from the time of admission in an integrated healthcare setting. METHOD: Clinical data from Electronic Health Record (EHR), at encounter level, were used to build a predictive model for progression from sepsis to septic shock up to 6 h from the time of admission; that is, T = 1, 3, and 6 h from admission. Eight different machine learning algorithms (Random Forest, XGBoost, C5.0, Decision Trees, Boosted Logistic Regression, Support Vector Machine, Logistic Regression, Regularized Logistic, and Bayes Generalized Linear Model) were used for model development. Two adaptive sampling strategies were used to address the class imbalance. Data from two sources (clinical and billing codes) were used to define the case definition (septic shock) using the Centers for Medicare & Medicaid Services (CMS) Sepsis criteria. The model assessment was performed using Area under Receiving Operator Characteristics (AUROC), sensitivity, and specificity. Model predictions for each feature window (1, 3 and 6 h from admission) were consolidated. RESULTS: Retrospective data from April 2005 to September 2018 were extracted from the EHR, Insurance Claims, Billing, and Laboratory Systems to create a dataset for septic shock detection. The clinical criteria and billing information were used to label patients into two classes-septic shock patients and sepsis patients at three different time points from admission, creating two different case-control cohorts. Data from 45,425 unique in-patient visits were used to build 96 prediction models comparing clinical-based definition versus billing-based information as the gold standard. Of the 24 consolidated models (based on eight machine learning algorithms and three feature windows), four models reached an AUROC greater than 0.9. Overall, all the consolidated models reached an AUROC of at least 0.8820 or higher. Based on the AUROC of 0.9483, the best model was based on Random Forest, with a sensitivity of 83.9% and specificity of 88.1%. The sepsis detection window at 6 h outperformed the 1 and 3-h windows. The sepsis definition based on clinical variables had improved performance when compared to the sepsis definition based on only billing information. CONCLUSION: This study corroborated that machine learning models can be developed to predict septic shock using clinical and administrative data. However, the use of clinical information to define septic shock outperformed models developed based on only administrative data. Intelligent decision support tools can be developed and integrated into the EHR and improve clinical outcomes and facilitate the optimization of resources in real-time.

Comparing Clinical Characteristics of Influenza and Common Coronavirus Infections Using Electronic Health Records.

BACKGROUND: We compared outcomes in inpatients and outpatients, pre-COVID-19, who were infected with either coronavirus or influenza. METHODS: Using deidentified electronic health records data from the Geisinger-Regeneron partnership, we compared patients with RT-PCR-positive tests for the 4 common coronaviruses (229E, HKU1, NL63, OC43) or influenza (A and B) from June 2016 to February 2019. RESULTS: Overall, 52 833 patients were tested for coronaviruses and influenza. For patients ≥21 years old, 1555 and 3991 patient encounters had confirmed positive coronavirus and influenza tests, respectively. Both groups had similar intensive care unit (ICU) admission rates (7.2% vs 6.1%, P = .12), although patients with coronavirus had significantly more pneumonia (15% vs 7.4%, P < .001) and higher death rate within 30 days (4.9% vs 3.0%, P < .001). After controlling for other covariates, coronavirus infection still had a higher risk of death and pneumonia than influenza (odds ratio, 1.64 and 2.05, P < .001), with no significant difference in ICU admission rates. CONCLUSIONS: Common coronaviruses cause significant morbidity, with potentially worse outcomes than influenza. Identifying a subset of patients who are more susceptible to poor outcomes from common coronavirus infections may help plan clinical interventions in patients with suspected infections.

Increasing the Density of Laboratory Measures for Machine Learning Applications.

BACKGROUND: The imputation of missingness is a key step in Electronic Health Records (EHR) mining, as it can significantly affect the conclusions derived from the downstream analysis in translational medicine. The missingness of laboratory values in EHR is not at random, yet imputation techniques tend to disregard this key distinction. Consequently, the development of an adaptive imputation strategy designed specifically for EHR is an important step in improving the data imbalance and enhancing the predictive power of modeling tools for healthcare applications. METHOD: We analyzed the laboratory measures derived from Geisinger's EHR on patients in three distinct cohorts-patients tested for Clostridioides difficile (Cdiff) infection, patients with a diagnosis of inflammatory bowel disease (IBD), and patients with a diagnosis of hip or knee osteoarthritis (OA). We extracted Logical Observation Identifiers Names and Codes (LOINC) from which we excluded those with 75% or more missingness. The comorbidities, primary or secondary diagnosis, as well as active problem lists, were also extracted. The adaptive imputation strategy was designed based on a hybrid approach. The comorbidity patterns of patients were transformed into latent patterns and then clustered. Imputation was performed on a cluster of patients for each cohort independently to show the generalizability of the method. The results were compared with imputation applied to the complete dataset without incorporating the information from comorbidity patterns. RESULTS: We analyzed a total of 67,445 patients (11,230 IBD patients, 10,000 OA patients, and 46,215 patients tested for C. difficile infection). We extracted 495 LOINC and 11,230 diagnosis codes for the IBD cohort, 8160 diagnosis codes for the Cdiff cohort, and 2042 diagnosis codes for the OA cohort based on the primary/secondary diagnosis and active problem list in the EHR. Overall, the most improvement from this strategy was observed when the laboratory measures had a higher level of missingness. The best root mean square error (RMSE) difference for each dataset was recorded as -35.5 for the Cdiff, -8.3 for the IBD, and -11.3 for the OA dataset. CONCLUSIONS: An adaptive imputation strategy designed specifically for EHR that uses complementary information from the clinical profile of the patient can be used to improve the imputation of missing laboratory values, especially when laboratory codes with high levels of missingness are included in the analysis.

Rapid Diagnostics for Blood Cultures: Supporting Decisions for Antimicrobial Therapy and Value-Based Care.

Bacteremia and sepsis are critically important syndromes with high mortality, morbidity, and associated costs. Bloodstream infections and sepsis are among the top causes of mortality in the US, with >600 deaths each day. Most septic patients can be found in emergency medicine departments or critical care units, settings in which rapid administration of targeted antibiotic therapy can reduce mortality. Unfortunately, routine blood cultures are not rapid enough to aid in the decision of therapeutic intervention at the onset of bacteremia. As a result, empiric, broad-spectrum treatment is common-a costly approach that may fail to target the correct microbe effectively, may inadvertently harm patients via antimicrobial toxicity, and may contribute to the evolution of drug-resistant microbes. To overcome these challenges, laboratorians must understand the complexity of diagnosing and treating septic patients, focus on creating algorithms that rapidly support decisions for targeted antibiotic therapy, and synergize with existing emergency department and critical care clinical practices put forth in the Surviving Sepsis Guidelines.

Performance of the T2Bacteria Panel for Diagnosing Bloodstream Infections: A Diagnostic Accuracy Study.

Background: Blood cultures, the gold standard for diagnosing bloodstream infections (BSIs), are insensitive and limited by prolonged time to results. The T2Bacteria Panel (T2 Biosystems) is a direct-from-blood, nonculture test that identifies the most common ESKAPE bacteria (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Pseudomonas aeruginosa, and Escherichia coli). Objective: To assess performance of the T2Bacteria Panel in diagnosing suspected BSIs in adults. Design: Prospective patient enrollment (8 December 2015 through 4 August 2017). Setting: Eleven U.S. hospitals. Patients: 1427 patients for whom blood cultures were ordered as standard of care. Intervention: Paired blood culture and T2Bacteria testing. Measurements: Performance of T2Bacteria compared with a single set of blood cultures in diagnosing proven, probable, and possible BSIs caused by T2Bacteria-targeted organisms. Results: Blood culture and T2Bacteria results were positive for targeted bacteria in 3% (39 of 1427) and 13% (181 of 1427) of patients, respectively. Mean times from start of blood culture incubation to positivity and species identification were 38.5 (SD, 32.8) and 71.7 (SD, 39.3) hours, respectively. Mean times to species identification with T2Bacteria were 3.61 (SD, 0.2) to 7.70 (SD, 1.38) hours, depending on the number of samples tested. Per-patient sensitivity and specificity of T2Bacteria for proven BSIs were 90% (95% CI, 76% to 96%) and 90% (CI, 88% to 91%), respectively; the negative predictive value was 99.7% (1242 of 1246). The rate of negative blood cultures with a positive T2Bacteria result was 10% (146 of 1427); 60% (88 of 146) of such results were associated with probable (n = 62) or possible (n = 26) BSIs. If probable BSIs and both probable and possible BSIs were assumed to be true positives missed by blood culture, per-patient specificity of T2Bacteria was 94% and 96%, respectively. Limitation: Low prevalence of positive blood cultures, collection of a single set of culture specimens, and inability of T2Bacteria to detect nontargeted pathogens. Conclusion: The T2Bacteria Panel rapidly and accurately diagnoses BSIs caused by 5 common bacteria. Primary Funding Source: T2 Biosystems.

Near point-of-care adoption of Cepheid Xpert® Flu/RSV XC testing within an integrated healthcare delivery network.

Flu/RSV testing was implemented in out-patient clinic-based physician office laboratories throughout Pennsylvania. On-site testing reduced the collect-to-result time by 70% when compared to testing in a centralized core laboratory; over- or under-treatment for influenza A and B (measured by anti-viral prescription) was reduced by 15% (P < 0.0001). Antimicrobial prescription was not affected by on-site testing.

Matrix-Assisted Laser Desorption Time of Flight Mass Spectrometry.

Matrix-assisted laser desorption time of flight mass spectrometry (MALDI-TOF MS), adapted for use in clinical microbiology laboratories, challenges current standards of microbial detection and identification. This article summarizes the capabilities of MALDI-TOF MS in diagnostic clinical microbiology laboratories and describes the underpinnings of the technology, highlighting topics such as sample preparation, spectral analysis, and accuracy. The use of MALDI-TOF MS in the clinical microbiology laboratory is growing, and, when properly deployed, can accelerate diagnosis and improve patient care.