Cerebrovascular disease case identification in inpatient electronic medical record data using natural language processing.

BACKGROUND: Abstracting cerebrovascular disease (CeVD) from inpatient electronic medical records (EMRs) through natural language processing (NLP) is pivotal for automated disease surveillance and improving patient outcomes. Existing methods rely on coders' abstraction, which has time delays and under-coding issues. This study sought to develop an NLP-based method to detect CeVD using EMR clinical notes. METHODS: CeVD status was confirmed through a chart review on randomly selected hospitalized patients who were 18 years or older and discharged from 3 hospitals in Calgary, Alberta, Canada, between January 1 and June 30, 2015. These patients' chart data were linked to administrative discharge abstract database (DAD) and Sunrise™ Clinical Manager (SCM) EMR database records by Personal Health Number (a unique lifetime identifier) and admission date. We trained multiple natural language processing (NLP) predictive models by combining two clinical concept extraction methods and two supervised machine learning (ML) methods: random forest and XGBoost. Using chart review as the reference standard, we compared the model performances with those of the commonly applied International Classification of Diseases (ICD-10-CA) codes, on the metrics of sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). RESULT: Of the study sample (n = 3036), the prevalence of CeVD was 11.8% (n = 360); the median patient age was 63; and females accounted for 50.3% (n = 1528) based on chart data. Among 49 extracted clinical documents from the EMR, four document types were identified as the most influential text sources for identifying CeVD disease ("nursing transfer report," "discharge summary," "nursing notes," and "inpatient consultation."). The best performing NLP model was XGBoost, combining the Unified Medical Language System concepts extracted by cTAKES (e.g., top-ranked concepts, "Cerebrovascular accident" and "Transient ischemic attack"), and the term frequency-inverse document frequency vectorizer. Compared with ICD codes, the model achieved higher validity overall, such as sensitivity (25.0% vs 70.0%), specificity (99.3% vs 99.1%), PPV (82.6 vs. 87.8%), and NPV (90.8% vs 97.1%). CONCLUSION: The NLP algorithm developed in this study performed better than the ICD code algorithm in detecting CeVD. The NLP models could result in an automated EMR tool for identifying CeVD cases and be applied for future studies such as surveillance, and longitudinal studies.

Accuracy of Self-Reported COVID-19 Vaccination Status Compared With a Public Health Vaccination Registry in Québec: Observational Diagnostic Study.

BACKGROUND: The accuracy of self-reported vaccination status is important to guide real-world vaccine effectiveness studies and policy making in jurisdictions where access to electronic vaccine registries is restricted. OBJECTIVE: This study aimed to determine the accuracy of self-reported vaccination status and reliability of the self-reported number of doses, brand, and time of vaccine administration. METHODS: This diagnostic accuracy study was completed by the Canadian COVID-19 Emergency Department Rapid Response Network. We enrolled consecutive patients presenting to 4 emergency departments (EDs) in Québec between March 24, 2020, and December 25, 2021. We included adult patients who were able to consent, could speak English or French, and had a proven COVID-19 infection. We compared the self-reported vaccination status of the patients with their vaccination status in the electronic Québec Vaccination Registry. Our primary outcome was the accuracy of the self-reported vaccination status (index test) ascertained during telephone follow-up compared with the Québec Vaccination Registry (reference standard). The accuracy was calculated by dividing all correctly self-reported vaccinated and unvaccinated participants by the sum of all correctly and incorrectly self-reported vaccinated and unvaccinated participants. We also reported interrater agreement with the reference standard as measured by unweighted Cohen κ for self-reported vaccination status at telephone follow-up and at the time of their index ED visit, number of vaccine doses, and brand. RESULTS: During the study period, we included 1361 participants. At the time of the follow-up interview, 932 participants reported at least 1 dose of a COVID-19 vaccine. The accuracy of the self-reported vaccination status was 96% (95% CI 95%-97%). Cohen κ for self-reported vaccination status at phone follow-up was 0.91 (95% CI 0.89-0.93) and 0.85 (95% CI 0.77-0.92) at the time of their index ED visit. Cohen κ was 0.89 (95% CI 0.87-0.91) for the number of doses, 0.80 (95% CI 0.75-0.84) for the brand of the first dose, 0.76 (95% CI 0.70-0.83) for the brand of the second dose, and 0.59 (95% CI 0.34-0.83) for the brand of the third dose. CONCLUSIONS: We reported a high accuracy of self-reported vaccination status for adult patients without cognitive disorders who can express themselves in English or French. Researchers can use self-reported COVID-19 vaccination data on the number of doses received, vaccine brand name, and timing of vaccination to guide future research with patients who are capable of self-reporting their vaccination data. However, access to official electronic vaccine registries is still needed to determine the vaccination status in certain susceptible populations where self-reported vaccination data remain missing or impossible to obtain. TRIAL REGISTRATION: Clinicaltrials.gov NCT04702945; https://clinicaltrials.gov/ct2/show/NCT04702945.

The multicenter impacts of an emergency physician lead on departmental flow and provider experiences.

INTRODUCTION: Emergency department (ED) flow impacts patient safety, quality of care and ED provider satisfaction. Throughput interventions have been shown to improve flow, yet few studies have reported the impact of ED physician leadership roles on patient flow and provider experiences. The study objective was to evaluate the impacts of the emergency physician lead role on ED flow metrics and provider experiences. METHODS: Quantitative data about patient flow metrics were collected from ED information systems in two tertiary hospital EDs and analyzed to compare ED length of stay, EMS hallway length of stay, physician initial assessment time, 72-h readmission and left without being seen rates three months before and following emergency physician lead role implementation. ED flow metrics for adult patients at each site were analyzed independently using descriptive and inferential statistics, t tests and multivariable regression analysis. Qualitative data were collected via surveys from ED providers (physicians, nurses, and EMS) about their experiences working with the emergency physician leads and analyzed for themes about emergency physician leads impact. RESULTS: The number of ED visits was relatively stable pre-post at the Peter Lougheed Centre (Lougheed) but increased pre-post at the Foothills Medical Centre (Foothills). Post-intervention at Lougheed median ED length of stay decreased by 18 min (p < 0.001) and at Foothills ED length of stay increased by 8 min (p < 0.001). EMS length of stay at Lougheed decreased by 20 min (p < 0.001), and at Foothills length of stay increased by 17 min (p < 0.001). Themes in provider feedback were that emergency physician leads (1) facilitated patient flow, (2) impacted provider workload, and (3) supported patient flow and safety with early assessments, treatments and investigations. CONCLUSION: In this study, the emergency physician lead impacted ED flow metrics variably at different sites, but important learnings from provider experiences can guide future emergency physician lead implementation.

RéSUMé: INTRODUCTION: Le flux des services d'urgence a une incidence sur la sécurité des patients, la qualité des soins et la satisfaction des fournisseurs de services d'urgence. Il a été démontré que les interventions de débit améliorent le flux, mais peu d'études ont rapporté l'impact des rôles de leadership des médecins des urgences sur le flux des patients et les expériences des prestataires. L'objectif de l'étude était d'évaluer l'impact du rôle du médecin chef des urgences sur les paramètres de flux des urgences et les expériences des prestataires. MéTHODES: Les données quantitatives sur les paramètres du flux des patients ont été recueillies à partir des systèmes d'information des urgences de deux hôpitaux tertiaires et analysées afin de comparer la durée du séjour aux urgences, la durée du séjour dans le couloir des SMU, le temps d'évaluation initiale par le médecin, les réadmissions dans les 72 heures et les taux de sortie sans consultation trois mois avant et après la mise en œuvre du rôle de chef des urgences. Les paramètres de débit des urgences pour les patients adultes de chaque site ont été analysés indépendamment à l'aide de statistiques descriptives et inférentielles, de tests t et d'une analyse de régression multivariable. Les données qualitatives ont été recueillies par le biais d'enquêtes auprès de fournisseurs de services d'urgence (médecins, infirmières et services médicaux d'urgence) sur leur expérience de travail avec les médecins chefs des services d'urgence et analysées pour en dégager les thèmes concernant l'impact des médecins chefs des services d'urgence. RéSULTATS: Le nombre de visites aux urgences était relativement stable avant et après au Peter Lougheed Centre (Lougheed), mais a augmenté avant et après au Foothills Medical Center (Foothills). Après l'intervention, la durée médiane du séjour aux urgences de Lougheed a diminué de 18 minutes (p < 0.001) et celle des urgences de Foothills a augmenté de 8 minutes (p < 0.001). La durée du séjour en SMU a diminué de 20 minutes à Lougheed (p < 0.001), et a augmenté de 17 minutes à Foothills (p < 0.001). Les thèmes abordés dans les commentaires des fournisseurs étaient les suivants : les responsables des urgences (1) facilitaient le flux des patients, (2) avaient un impact sur la charge de travail des fournisseurs et (3) favorisaient le flux et la sécurité des patients grâce à des évaluations, des traitements et des examens précoces. CONCLUSION: Dans cette étude, le médecin chef des urgences a eu un impact variable sur les paramètres de débit des urgences dans les différents sites, mais les enseignements importants tirés des expériences des fournisseurs peuvent guider la mise en œuvre future du médecin chef des urgences.

Derivation and validation of a clinical decision rule to risk-stratify COVID-19 patients discharged from the emergency department: The CCEDRRN COVID discharge score.

Objective: To risk-stratify COVID-19 patients being considered for discharge from the emergency department (ED). Methods: We conducted an observational study to derive and validate a clinical decision rule to identify COVID-19 patients at risk for hospital admission or death within 72 hours of ED discharge. We used data from 49 sites in the Canadian COVID-19 Emergency Department Rapid Response Network (CCEDRRN) between March 1, 2020, and September 8, 2021. We randomly assigned hospitals to derivation or validation and prespecified clinical variables as candidate predictors. We used logistic regression to develop the score in a derivation cohort and examined its performance in predicting short-term adverse outcomes in a validation cohort. Results: Of 15,305 eligible patient visits, 535 (3.6%) experienced the outcome. The score included age, sex, pregnancy status, temperature, arrival mode, respiratory rate, and respiratory distress. The area under the curve was 0.70 (95% confidence interval [CI] 0.68-0.73) in derivation and 0.71 (95% CI 0.68-0.73) in combined derivation and validation cohorts. Among those with a score of 3 or less, the risk for the primary outcome was 1.9% or less, and the sensitivity of using 3 as a rule-out score was 89.3% (95% CI 82.7-94.0). Among those with a score of ≥9, the risk for the primary outcome was as high as 12.2% and the specificity of using 9 as a rule-in score was 95.6% (95% CI 94.9-96.2). Conclusion: The CCEDRRN COVID discharge score can identify patients at risk of short-term adverse outcomes after ED discharge with variables that are readily available on patient arrival.

CCEDRRN COVID-19 Infection Score (CCIS): development and validation in a Canadian cohort of a clinical risk score to predict SARS-CoV-2 infection in patients presenting to the emergency department with suspected COVID-19.

OBJECTIVES: To develop and validate a clinical risk score that can accurately quantify the probability of SARS-CoV-2 infection in patients presenting to an emergency department without the need for laboratory testing. DESIGN: Cohort study of participants in the Canadian COVID-19 Emergency Department Rapid Response Network (CCEDRRN) registry. Regression models were fitted to predict a positive SARS-CoV-2 test result using clinical and demographic predictors, as well as an indicator of local SARS-CoV-2 incidence. SETTING: 32 emergency departments in eight Canadian provinces. PARTICIPANTS: 27 665 consecutively enrolled patients who were tested for SARS-CoV-2 in participating emergency departments between 1 March and 30 October 2020. MAIN OUTCOME MEASURES: Positive SARS-CoV-2 nucleic acid test result within 14 days of an index emergency department encounter for suspected COVID-19 disease. RESULTS: We derived a 10-item CCEDRRN COVID-19 Infection Score using data from 21 743 patients. This score included variables from history and physical examination and an indicator of local disease incidence. The score had a c-statistic of 0.838 with excellent calibration. We externally validated the rule in 5295 patients. The score maintained excellent discrimination and calibration and had superior performance compared with another previously published risk score. Score cut-offs were identified that can rule-in or rule-out SARS-CoV-2 infection without the need for nucleic acid testing with 97.4% sensitivity (95% CI 96.4 to 98.3) and 95.9% specificity (95% CI 95.5 to 96.0). CONCLUSIONS: The CCEDRRN COVID-19 Infection Score uses clinical characteristics and publicly available indicators of disease incidence to quantify a patient's probability of SARS-CoV-2 infection. The score can identify patients at sufficiently high risk of SARS-CoV-2 infection to warrant isolation and empirical therapy prior to test confirmation while also identifying patients at sufficiently low risk of infection that they may not need testing. TRIAL REGISTRATION NUMBER: NCT04702945.

Decision support for computed tomography in the emergency department: a multicenter cluster-randomized controlled trial.

OBJECTIVES: Clinical decision support may facilitate evidence-based imaging, but most studies to date examining the impact of decision support have used non-randomized designs which limit the conclusions that can be drawn from them. This randomized trial examines if decision support can reduce computed tomography (CT) utilization for patients with mild traumatic brain injuries and suspected pulmonary embolism in the emergency department. This study was funded by a competitive public research grant and registered on ClinicalTrials.gov (NCT02410941). METHODS: Emergency physicians at five urban sites were assigned to voluntary decision support for CT imaging of patients with either head injuries or suspected pulmonary embolism using a cluster-randomized design over a 1-year intervention period. The co-primary outcomes were CT head and CT pulmonary angiography utilization. CT pulmonary angiography diagnostic yield (proportion of studies diagnostic for acute pulmonary embolism) was a secondary outcome. RESULTS: A total of 225 physicians were randomized and studied over a 2-year baseline and 1-year intervention period. Physicians interacted with the decision support in 38.0% and 45.0% of eligible head injury and suspected pulmonary embolism cases, respectively. A mixed effects logistic regression model demonstrated no significant impact of decision support on head CT utilization (OR 0.93, 95% CI 0.79-1.10, p = 0.31), CT pulmonary angiography utilization (OR 0.98, 95% CI 0.88-1.11, p = 0.74) or diagnostic yield (OR 1.23, 95% CI 0.96-1.65, p = 0.10). However, overall CT pulmonary diagnostic yield (17.7%) was almost three times higher than that reported by a recent large US study, suggesting that selective imaging was already being employed. CONCLUSION: Voluntary decision support addressing many commonly cited barriers to evidence-based imaging did not significantly reduce CT utilization or improve diagnostic yield but was limited by low rates of participation and high baseline rates of selective imaging. Demonstrating value to clinicians through interventions that improve workflow is likely necessary to meaningfully change imaging practices.

RéSUMé: OBJECTIFS: Le soutien à la décision clinique peut faciliter l'imagerie fondée sur des données probantes, mais la plupart des études à ce jour examinant l'impact du soutien à la décision ont utilisé des modèles non randomisés qui limitent les conclusions qui peuvent en être tirées. Cet essai randomisé examine si l'aide à la décision peut réduire l'utilisation de la tomodensitométrie chez les patients présentant des lésions cérébrales traumatiques légères et une embolie pulmonaire présumée au service des urgences. Cette étude a été financée par une subvention de recherche publique compétitive et enregistrée sur ClinicalTrials.gov (NCT02410941). MéTHODES: Les médecins urgentistes de cinq sites urbains ont été assignés à une aide à la décision volontaire pour l'imagerie par tomodensitométrie des patients présentant soit un traumatisme crânien, soit une suspicion d'embolie pulmonaire, selon une conception randomisée en grappes sur une période d'intervention d'un an. Les résultats co-primaires étaient l'utilisation de la tomodensitométrie de la tête et de la tomodensitométrie par angiographie pulmonaire. Le rendement diagnostique de l'angiographie pulmonaire par TDM (proportion d'études diagnostiquant une embolie pulmonaire aiguë) était un résultat secondaire. RéSULTATS: Au total, 225 médecins ont été randomisés et étudiés au cours d'une période de référence de deux ans et d'une période d'intervention d'un an. Les médecins ont interagi avec l'aide à la décision dans 38,0 % et 45,0 % des cas admissibles de blessure à la tête et d'embolie pulmonaire soupçonnée, respectivement. Un modèle de régression logistique à effets mixtes n'a démontré aucun impact significatif de l'aide à la décision sur l'utilisation de la tomodensitométrie de la tête (OR 0,93, IC 95 % 0,79-1,10, p = 0,31), l'utilisation de l'angiographie pulmonaire par tomodensitométrie (OR 0,98, IC 95 % 0,88-1,11, p = 0,74) ou le rendement diagnostique (OR 1,23, IC 95 % 0,96-1,65, p = 0,10). Toutefois, le rendement global du diagnostic pulmonaire par TDM (17,7 %) était près de trois fois supérieur à celui rapporté par une étude récente aux États-Unis, ce qui laisse supposer que l'imagerie sélective était déjà utilisée. CONCLUSIONS: L'aide à la décision volontaire visant à éliminer de nombreux obstacles fréquemment cités à l'imagerie fondée sur des données probantes n'a pas réduit de façon significative l'utilisation de la tomodensitométrie ni amélioré le rendement diagnostique, mais a été limitée par de faibles taux de participation et des taux de base élevés d'imagerie sélective. La démonstration de la valeur pour les cliniciens par des interventions qui améliorent le flux de travail est probablement nécessaire pour changer de manière significative les pratiques d'imagerie.

Sex-related differences in opioid administration in the emergency department: a population-based study.

BACKGROUND: Sex differences in pain experience and expression may influence ED pain management. Our objective was to evaluate the effect of sex on ED opioid administration. METHODS: We conducted a multicentre population-based observational cohort study using administrative data from Calgary's four EDs between 2017 and 2018. Eligible patients had a presenting complaint belonging to one of nine pain categories or an arrival pain score >3. We performed multivariable analyses to identify predictors of opioid administration and stratified analyses by age, pain severity and pain category. RESULTS: We studied 119 510 patients (mean age 47.4 years; 55.4% female). Opioid administration rates were similar for men and women. After adjusting for age, hospital site, pain category, ED length of stay and pain severity, male sex was not a predictor of opioid treatment (adjusted OR (aOR)=0.93; 95% CI 0.85 to 1.02). However, men were more likely to receive opioids in the categories of trauma (aOR=1.58, 95% CI 1.40 to 1.78), flank pain (aOR=1.24, 95% CI 1.11 to 1.38), headache (aOR=1.18, 95% CI 1.03 to 1.34) and abdominal pain (aOR=1.11, 95% CI 1.08 to 1.18). Pain category appears to be a strong determinant of opioid administration, especially back pain (aOR=6.56, 95% CI 5.99 to 7.19) and flank pain (aOR=6.04, 95% CI 5.48 to 6.65). There was significant variability in opioid provision by ED site (aOR 0.76 to 1.24). CONCLUSIONS: This population-based study demonstrated high variability in opioid use across different settings. Overall, men and women had similar likelihood of receiving opioids; however men with trauma, flank pain, headache and abdominal pain were much more likely to receive opioids. ED physicians should self-examine their analgesic practices with respect to possible sex biases, and departments should introduce evidence-based, indication-specific analgesic protocols to reduce practice variability and optimise opioid analgesia.

Transfusions in patients with iron deficiency anemia following release of Choosing Wisely Guidelines.

BACKGROUND: In 2016, based on recommendations of the American Association of Blood Banks (AABB), Choosing Wisely Canada released transfusion guidelines for patients with Iron Deficiency Anemia. The goal of the present study was to examine the number of transfusions given in Calgary emergency departments (EDs) before and after the release of these guidelines. METHODS: We analyzed 11,786 anemia encounters from January 2014 to December 2019. A transfusion was considered potentially avoidable if the patient's hemoglobin was > 70 g/L and if the patient was hemodynamically stable. We used time-series analyses to examine change in rate of total and potentially avoidable transfusions quarterly over the total and pre and post intervention periods. RESULTS: In total, 1409/11,786 (12.0%) of the encounters received transfusions; 80.0% (1127/1409) were indicated while 19.9% (281/1409) were potentially avoidable. In the pre-intervention period, the rate of potentially avoidable transfusions was 21.5% (133/618) and in the post-intervention period, the rate of potentially avoidable transfusions was 18.7% (148/791). The rate of potentially avoidable transfusions decreased quarterly at a rate of 0.3% which did not reach statistical significance (p = 0.06). DISCUSSION: Our data suggest that the number of potentially avoidable transfusions has not decreased since the release of Choosing Wisely Canada guidelines and local educational initiatives. This may be due to the fact that there is a pre-existing down trend in the number of transfusions provided.

RéSUMé: CONTEXTE: En 2016, sur la base des recommandations de l'AABB (Association américaine des banques de sang) Choisir avec soin Canada a publié des directives sur la pratique transfusionnelle pour les patients atteints d'anémie ferriprive. Le but de la présente étude était d'examiner le nombre de transfusions administrées dans les services d'urgence (SU) de Calgary avant et après la publication de ces directives. MéTHODES: Nous avons analysé 11 786 cas d'anémie entre janvier 2014 et décembre 2019. Une transfusion était jugée comme potentiellement évitable si l'hémoglobine du patient était > 70 g/L et si le patient était stable sur le plan hémodynamique. Nous avons utilisé des analyses de séries chronologiques pour examiner trimestriellement le changement du taux de transfusions totales et potentiellement évitables au cours de l'ensemble des périodes, y compris avant et après l'intervention. RéSULTATS: Au total, 1409/11786 (12.0%) des cas ont reçu des transfusions ; 80.0% (1127/1409) ont été indiqués tandis que 19.9 % (281/1409) étaient potentiellement évitables. Pendant la période précédant l'intervention, le taux de transfusions potentiellement évitables était de 21.5 % (133/618) et dans la période postérieur à l'intervention, le taux de transfusions potentiellement évitables était de 18,7 % (148/791). Le taux de transfusions potentiellement évitables a diminué chaque trimestre à un taux de 0,3 % qui n'a pas atteint la signification statistique (p = 0,06). DISCUSSION: Nos données suggèrent que le nombre de transfusions potentiellement évitables n'a pas diminué depuis la publication des directives de Choisir avec soin Canada et des initiatives éducatives locales. Cela peut être dû au fait qu'il existe une tendance à la baisse préexistante du nombre de transfusions fournies.

Prospective comparative evaluation of the European Society of Cardiology (ESC) 1-hour and a 2-hour rapid diagnostic algorithm for myocardial infarction using high-sensitivity troponin-T.

OBJECTIVE: Both 1- and 2-hour rapid diagnostic algorithms using high-sensitivity troponin (hs-cTn) have been validated to diagnose acute myocardial infarction (MI), leaving physicians uncertain which algorithm is preferable. The objective of this study was to prospectively evaluate the diagnostic performance of 1- and 2-hour algorithms in clinical practice in a Canadian emergency department (ED). METHODS: ED patients with chest pain had high-sensitivity cardiac troponin-T (hs-cTnT) collected on presentation and 1- and 2-hours later at a single academic centre over a 2-year period. The primary outcome was index MI, and the secondary outcome was 30-day major adverse cardiac events (MACE). All outcomes were adjudicated. RESULTS: We enrolled 608 patients undergoing serial hs-cTnT sampling. Of these, 350 had a valid 1-hour and 550 had a 2-hour hs-cTnT sample. Index MI and 30-day MACE prevalence was ~12% and 14%. Sensitivity of the 1- and 2-hour algorithms was similar for index MI 97.3% (95% CI: 85.8-99.9%) and 100% (95% CI: 91.6-100%) and 30-day MACE: 80.9% (95% CI: 66.7-90.9%) and 83.3% (95% CI: 73.2-90.8%), respectively. Both algorithms accurately identified about 10% of patients as high risk. CONCLUSIONS: Both algorithms were able to classify almost two-thirds of patients as low risk, effectively ruling out MI and conferring a low risk of 30-day MACE for this group, while reliably identifying high-risk patients. While both algorithms had equivalent diagnostic performance, the 2-hour algorithm offers several practical advantages, which may make it preferable to implement. Broad implementation of similar algorithms across Canada can expedite patient disposition and lead to resource savings.

A Multicenter Assessment of the Sensitivity and Specificity for a Single High-Sensitivity Cardiac Troponin Test at Emergency Department Presentation for Hospital Admission.

BACKGROUND: Studies have illustrated how a low or undetectable high-sensitivity cardiac troponin (hs-cTn) concentration at emergency department (ED) presentation can rule out myocardial infarction (MI). A problem with using an undetectable hs-cTn cutoff is that this value may be defined differently among hospitals and is also difficult to monitor. In the present study, we assess the diagnostic performance of a clinical chemistry score (CCS) vs hs-cTn alone in the presentation blood sample in the ED for patient hospital admission in a multicenter setting. METHODS: From January 1 to June 30, 2018, consecutive patients with random glucose, creatinine (for an estimated glomerular filtration rate calculation), and hs-cTnI (Abbott, 2 hospitals, Hamilton, Ontario, n = 10496) or hs-cTnT (Roche, 4 hospitals, Calgary, Alberta, n = 25177) were assessed for hospital admission with the CCS (range of scores, 0-5) or hs-cTn alone. Sensitivity, specificity, predicative values, and likelihood ratios were calculated for a CCS of 0 and 5 and for hs-cTn alone (hs-cTnI cutoffs, 5 and 26 ng/L; hs-cTnT cutoffs, 6 and 14 ng/L). RESULTS: The CCS of 0 (CCS <1) identified approximately 10% of all patients as low risk and had a sensitivity for hospital admission of nearly 98% as compared to <93% when hs-cTnT (<6 ng/L) or hs-cTnI (<5 ng/L) cutoffs alone were used. A CCS ≥5 had a specificity for hospital admission >95%, with approximately 14% of patients at high risk. CONCLUSIONS: An ED disposition (admit or send home) using the presentation blood sample could occur in nearly 25% of all patients by use of the CCS.

Gene set analysis and reduction for a continuous phenotype: Identifying markers of birth weight variation based on embryonic stem cells and immunologic signatures.

BACKGROUND: Gene set analysis is a popular approach to examine the association between a predefined gene set and a phenotype. Few methods have been developed for a continuous phenotype. However, often not all the genes within a significant gene set contribute to its significance. There is no gene set reduction method developed for continuous phenotype. We developed a computationally efficient analytical tool, called linear combination test for gene set reduction (LCT-GSR) to identify core subsets of gene sets associated with a continuous phenotype. Identifying the core subset enhances our understanding of the biological mechanism and reduces costs of disease risk assessment, diagnosis and treatment. RESULTS: We evaluated the performance of our analytical tool by applying it to two real microarray studies. In the first application, we analyzed pathway expression measurements in newborns' blood to discover core genes contributing to the variation in birth weight. On average, we were able to reduce the number of genes in the 33 significant gene sets of embryonic stem cell signatures by 84.3% resulting in 229 unique genes. Using immunologic signatures, on average we reduced the number of genes in the 210 significant gene sets by 89% leading to 1603 unique genes. There were 180 unique core genes overlapping across the two databases. In the second application, we analyzed pathway expression measurements in a cohort of lethal prostate cancer patients from Swedish Watchful Waiting cohort to identify main genes associated with tumor volume. On average, we were able to reduce the number of genes in the 17 gene sets by 90% resulting in 47 unique genes. CONCLUSIONS: We conclude that LCT-GSR is a statistically sound analytical tool that can be used to extract core genes associated with a continuous phenotype. It can be applied to a wide range of studies in which dichotomizing the continuous phenotype is neither easy nor meaningful. Reduction to the most predictive genes is crucial in advancing our understanding of issues such as disease prevention, faster and more efficient diagnosis, intervention strategies and personalized medicine.

Low High-Sensitivity Troponin Thresholds Identify Low-Risk Patients With Chest Pain Unlikely to Benefit From Further Risk Stratification.

BACKGROUND: Very low high-sensitivity cardiac troponin T (hs-cTnT) thresholds on presentation can rule out acute myocardial infarction (AMI), but the ability to identify patients at low risk of 30-day major adverse cardiac events (MACE) is less clear. This study examines the sensitivity of low concentrations of hs-cTnT on presentation to rule out 30-day MACE. METHODS: This prospective cohort study enrolled patients with chest pain presenting to the emergency department with nonischemic electrocardiograms who underwent AMI rule-out with an hs-cTnT assay. The primary outcome was 30-day MACE; secondary outcomes were individual MACE components. Because guidelines recommend using a single hs-cTnT strategy only for patients with more than 3 hours since symptom onset, a subgroup analysis was performed for this population. Outcomes were adjudicated on the basis of review of medical records and telephone follow-up. RESULTS: Of 1167 patients enrolled, 125 (10.7%) experienced 30-day MACE and 97 (8.3%) had AMI on the index visit. More than one-third of patients (35.6%) had presenting hs-cTnT concentrations below the limit of detection (5 ng/L), which was 94.4% (95% confidence interval [CI], 88.8-97.7) sensitive for 30-day MACE and 99.0% (95% CI, 94.5-100) sensitive for index AMI. Of 292 patients (25.0%) with hs-cTnT < 5 ng/L and at least 3 hours since symptom onset, only 3 experienced 30-day MACE (sensitivity 97.6%; 95% CI, 93.2-100) and none had AMI within 30 days (sensitivity 100%; 95% CI, 96.3-100). CONCLUSIONS: Among patients with nonischemic electrocardiograms and > 3 hours since symptom onset, low hs-cTnT thresholds on presentation confer a very low risk of 30-day MACE. In the absence of a high-risk clinical presentation, further risk stratification is likely to be low yield.

CONTEXTE: Un seuil de troponine T cardiaque hypersensible (TnTc-hs) très bas au moment de la consultation permet d'écarter le diagnostic d'infarctus aigu du myocarde (IAM), mais l'utilité de ce paramètre pour reconnaître les patients exposés à un faible risque d'événement cardiaque indésirable majeur (ECIM) à 30 jours est moins bien établie. Les auteurs examinent la sensibilité de la présence d'une faible concentration de TnTc-hs à la consultation comme critère pour écarter la possibilité d'un ECIM à 30 jours. MÉTHODE: Ont été admis dans cette étude de cohorte prospective les patients qui se sont présentés à l'urgence en raison d'une douleur à la poitrine, dont l'électrocardiogramme n'a pas révélé d'ischémie et chez qui le diagnostic d'IAM a été écarté au moyen d'un dosage de la TnTc-hs. Le critère d'évaluation principal était la survenue d'un ECIM à 30 jours; les critères d'évaluation secondaires étaient les composantes individuelles de l'ECIM. Comme les lignes directrices recommandent le recours à un simple dosage de la TnTc-hs seulement pour les patients présentant des symptômes depuis plus de 3 heures, une analyse a été réalisée dans ce sous-groupe de la population à l'étude. Les critères d'évaluation ont été confirmés par un examen des dossiers médicaux et par un suivi téléphonique. RÉSULTATS: Des 1167 patients retenus, 125 (10,7 %) ont présenté un ECIM à 30 jours et 97 (8,3 %) avaient reçu un diagnostic d'IAM à la visite de référence. Au moment de la consultation, plus du tiers des patients (35,6 %) présentaient une concentration de TnTc-hs sous le seuil de détection (5 ng/l), ce qui représente une sensibilité de 94,4 % (intervalle de confiance [IC] à 95 % : de 88,8 à 97,7) dans le cas de l'ECIM à 30 jours et de 99,0 % (IC à 95 % : de 94,5 à 100) dans le cas de l'IAM de référence. Des 292 patients (25,0 %) présentant un taux de TnTc-hs < 5 ng/l et des symptômes apparus depuis au moins 3 heures, seulement 3 ont subi un ECIM à 30 jours (sensibilité de 97,6 %; IC à 95 % : de 93,2 à 100) et aucun n'a subi d'IAM dans les 30 jours (sensibilité de 100 %; IC à 95 % : de 96,3 à 100). CONCLUSIONS: Chez les patients dont l'électrocardiogramme ne révèle pas d'ischémie et qui présentent des symptômes depuis au moins 3 heures, un seuil de TnTc-hs faible au moment de la consultation est associé à un très faible risque d'ECIM à 30 jours. En l'absence d'un tableau clinique associé à un risque élevé, il est peu probable qu'une stratification du risque plus poussée soit utile.

Comparative Evaluation of 2-Hour Rapid Diagnostic Algorithms for Acute Myocardial Infarction Using High-Sensitivity Cardiac Troponin T.

BACKGROUND: Symptoms of acute coronary syndrome account for a large proportion of emergency department (ED) visits and hospitalizations. High-sensitivity troponin can rapidly rule out or rule in acute myocardial infarction (AMI) within a short time of ED arrival. We sought to validate test characteristics and classification performance of 2-hour high-sensitivity troponin T (hsTnT) algorithms for the rapid diagnosis of AMI. METHODS: We included consecutive patients from 4 academic EDs with suspected cardiac chest pain who had hsTnT assays performed 2 hours apart (± 30 minutes) as part of routine care. The primary outcome was AMI at 7 days. Secondary outcomes included major adverse cardiac events (mortality, AMI, and revascularization). Test characteristics and classification performance for multiple 2-hour algorithms were quantified. RESULTS: Seven hundred twenty-two patients met inclusion criteria. Seven-day AMI incidence was 10.9% and major adverse cardiac event incidence was 13.7%. A 2-hour rule-out algorithm proposed by Reichlin and colleagues ruled out AMI in 59.4% of patients with 98.7% sensitivity and 99.8% negative predictive value (NPV). The 2-hour rule-out algorithm proposed by the United Kingdom National Institute for Health and Care Excellence ruled out AMI in 50.3% of patients with similar sensitivity and NPV. Other exploratory algorithms had similar sensitivity but marginally better classification performance. According to Reichlin et al., the 2-hour rule-in algorithm ruled in AMI in 16.5% of patients with 92.4% specificity and 58.5% positive predictive value. CONCLUSIONS: Two-hour hsTnT algorithms can rule out AMI with very high sensitivity and NPV. The algorithm developed by Reichlin et al. had superior classification performance. Reichlin and colleagues' 2-hour rule-in algorithm had poor positive predictive value and might not be suitable for early rule-in decision-making.

Undetectable Concentrations of a Food and Drug Administration-approved High-sensitivity Cardiac Troponin T Assay to Rule Out Acute Myocardial Infarction at Emergency Department Arrival.

BACKGROUND: The objective of this study was to quantify the sensitivity of very low concentrations of high-sensitivity cardiac troponin T (hsTnT) at ED arrival for acute myocardial infarction (AMI) in a large cohort of chest pain patients evaluated in real-world clinical practice. METHODS: This retrospective study included consecutive ED patients with suspected cardiac chest pain evaluated in four urban EDs, excluding those with ST-elevation AMI, cardiac arrest or abnormal kidney function. The primary outcomes were AMI at 7, 30, and 90 days. Secondary outcomes included major adverse cardiac events (MACE; all-cause mortality, AMI, and revascularization) and the individual MACE components. Test characteristics were calculated for hsTnT values from 3 to 200 ng/L . RESULTS: A total of 7,130 patients met inclusion criteria. AMI incidences at 7, 30, and 90 days were 5.8, 6.0, and 6.2%. When the hsTnT assay was performed at ED arrival, the limit of blank of the assay (3 ng/L) ruled out 7-day AMI in 15.5% of patients with 100% sensitivity and negative predictive value (NPV). The limit of detection of the assay (5 ng/L) ruled out AMI in 33.6% of patients with 99.8% sensitivity and 99.95% NPV for 7-day AMI. The limit of quantification (the Food and Drug Administration [FDA]-approved cutoff for lower the reportable limit) of 6 ng/L ruled out AMI in 42.2% of patients with 99.8% sensitivity and 99.95% NPV. The sensitivities of the cutoffs of <3, <5, and <6 ng/L for 7-day MACE were 99.6, 97.4, and 96.6%, respectively. The NPVs of the cutoffs of <3, <5, and <6 ng/L for 7-day MACE were 99.8, 99.5, and 99.4%, respectively. A secondary analysis was performed in a subgroup of 3,549 higher-risk patients who underwent serial troponin testing. In this subgroup, a cutoff of 3 ng/L ruled out 7-day AMI in 9.6% of patients with 100% sensitivity and NPV, a cutoff of 5 ng/L ruled out 7-day AMI in 23.3% of patients with 99.7% sensitivity and 99.9% NPV, and a cutoff of 6 ng/L ruled out 7-day AMI in 29.8% of patients with 99.7 and 99.9% NPV. In the higher-risk subgroup, the sensitivities of cutoffs of <3, <5, and <6 ng/L for 7-day MACE were 99.8, 97.4, and 96.6%, respectively. In this higher-risk subgroup, the NPV of cutoffs of <3, <5, and <6 ng/L for 7-day MACE were 99.7, 98.5, and 98.4%, respectively. CONCLUSIONS: When used in real-world clinical practice conditions, hsTnT concentrations < 6 ng/L (below the lower reportable limit for an FDA-approved assay) at the time of ED arrival can rule out AMI with very high sensitivity and NPV. The sensitivity for MACE is unacceptably low, and thus a single-troponin rule-out strategy should only be used in the context of a structured risk evaluation.

Can Public Health Risk Assessment Using Risk Matrices Be Misleading?

The risk assessment matrix is a widely accepted, semi-quantitative tool for assessing risks, and setting priorities in risk management. Although the method can be useful to promote discussion to distinguish high risks from low risks, a published critique described a problem when the frequency and severity of risks are negatively correlated. A theoretical analysis showed that risk predictions could be misleading. We evaluated a practical public health example because it provided experiential risk data that allowed us to assess the practical implications of the published concern that risk matrices would make predictions that are worse than random. We explored this predicted problem by constructing a risk assessment matrix using a public health risk scenario-Tainted blood transfusion infection risk-That provides negative correlation between harm frequency and severity. We estimated the risk from the experiential data and compared these estimates with those provided by the risk assessment matrix. Although we validated the theoretical concern, for these authentic experiential data, the practical scope of the problem was limited. The risk matrix has been widely used in risk assessment. This method should not be abandoned wholesale, but users must address the source of the problem, apply the risk matrix with a full understanding of this problem and use matrix predictions to inform, but not drive decision-making.

Sepsis After Cardiac Surgery Early in Infancy and Adverse 4.5-Year Neurocognitive Outcomes.

BACKGROUND: We aimed to determine whether sepsis is associated with neurocognitive outcomes 4.5 years after congenital heart disease surgery in early infancy. METHODS AND RESULTS: A secondary analysis from a prospective inception cohort included all children having congenital heart disease surgery done at ≤6 weeks of age with cardiopulmonary bypass at the Western Canadian referral center from 1996 to 2009. Follow-up at the referral center determined the primary outcomes at 4.5 years with full-scale, performance, and verbal intelligence quotients on the Wechsler Preschool and Primary Scale of Intelligence. Perioperative variables were collected prospectively, and confirmation of blood culture-positive sepsis was done retrospectively. Multiple linear regression models for neurocognitive outcomes and multiple Cox proportional hazards regression for mortality were determined. Sepsis occurred in 97 of 502 patients (19%) overall and in 76 of 396 survivors (19%) with 4.5-year follow-up. By 4.5 years, there were 91 (18%) deaths, and 396 of 411 survivors (96%) had follow-up completed. Extracorporeal membrane oxygenation was associated with worse scores on all neurocognitive outcomes on multivariable regression; the association between extracorporeal membrane oxygenation and full-scale intelligence quotient had a regression coefficient of -13.6 (95% CI -21.3 to -5.9; P=0.001). Sepsis perioperatively was associated with performance and verbal intelligence quotients, with a trend for full-scale intelligence quotient (P=0.058) on multivariable regression. The regression coefficient for sepsis was strongest for performance intelligence quotient (-5.31; 95% CI -9.84 to -0.78; P=0.022). Sepsis was not but extracorporeal membrane oxygenation was associated with mortality by 4.5 years. CONCLUSIONS: Perioperative sepsis and extracorporeal membrane oxygenation were associated with adverse neurocognitive outcomes on multivariable regression. Quality improvement to prevent sepsis has the potential to improve long-term neurocognitive outcomes in infants after surgery for congenital heart disease.

Clinical Outcome Score Predicts Adverse Neurodevelopmental Outcome After Infant Heart Surgery.

BACKGROUND: The purpose of this study was to determine whether a clinical outcome score derived from early postoperative events is associated with Bayley-III scores at 18 to 24 months among infants undergoing cardiopulmonary bypass surgery. METHODS: Included were infants aged 6 weeks or less who underwent surgery between 2005 and 2009, all of whom were referred for neurodevelopmental evaluation at 18 to 24 months. We excluded children with chromosomal abnormalities. The prespecified clinical outcome score had a range of 0 to 7. Lower scores indicated a more rapid postoperative recovery. Patients requiring extracorporeal life support were assigned a score of 7. RESULTS: One hundred and ninety-nine subjects were included. Surgical procedures were arterial switch (72), Norwood (60), repair of total anomalous pulmonary venous connection (29), and other (38). Nine subjects had postoperative extracorporeal life support. Mean clinical outcome score in the Norwood group was 4.0 ± 1.4 versus the arterial switch group (2.6 ± 1.5, p < 0.001), total anomalous pulmonary venous connection group (2.8 ± 1.8, p < 0.01), and other group (4.0 ± 1.8, p = not significant). Among children who had a clinical outcome score of 4 or greater, there was a decrease in Bayley-III cognitive score of 5.7 (95% confidence interval: 1.5 to 9.9, p = 0.009), a decrease in language score of 10.0 (95% confidence interval: 4.9 to 15.1, p < 0.001), and a decrease in motor score of 9.7 (95% confidence interval: 4.8 to 14.5, p < 0.001). Time until lactate of 2.0 mmol/L or less and highest 24-hour inotrope score increased with increasing clinical outcome score (p < 0.0001). CONCLUSIONS: Clinical outcome scores of 4 or greater were associated with significantly lower Bayley-III scores at 18 to 24 months. This score may be valuable as an endpoint when evaluating novel potential therapies for this high-risk population.

Neurocognitive outcomes at kindergarten entry after surgical repair of total anomalous pulmonary venous connection in early infancy.

The objective of this study was to determine neurocognitive outcomes 4.5 years after surgery for TAPVC in infancy and predictors of these outcomes. A cohort having TAPVC repair at age ≤6 weeks between 1998 and 2007 were followed by the Complex Pediatric Therapies Follow-up Program at 4.5 years. Outcomes include mortality, full-scale intelligence quotient (FSIQ), verbal IQ (VIQ), performance IQ (PIQ), visual motor integration (VMI), and general adaptive composite of the Adaptive Behavior Assessment System (GAC). There were 51 infants with simple TAPVC [4 year mortality 4 (8%)], and 16 with complex TAPVC [4 year mortality 7 (44%)], hazard ratio (HR) 7.02 (95% CI 2.05-24.07, p = 0.002). Of the 47 survivors after simple TAPVC, FSIQ (SD) was 92 (17), VIQ 92 (17), PIQ 94 (15), VMI 92 (15), and GAC 92 (15). Independent predictors of neurocognitive outcome included father's socioeconomic status, mother's years of schooling, gender, post-operative base deficit, and deep hypothermic circulatory arrest (DHCA) time. Complex TAPVC was associated on univariate analysis only with PIQ [81.9 (10.2) vs. 93.6 (15.4); p = 0.012] and FSIQ [80.7 (10.1) vs. 92.0 (17.7); p = 0.017]. Original peoples accounted for 25/51 (49%) of simple and 3/16 (19%) of complex TAPVC. Original peoples race was associated with 4-year mortality [HR 6.85 (95% CI 2.15, 21.76, p = 0.001)]. Survivors of TAPVC repair in early infancy have encouraging neurocognitive outcomes. Few independent predictors of neurocognitive outcome were found, with post-operative acidosis and DHCA time being potentially modifiable. Original peoples account for an unexpected proportion of patients (42%) and have a higher mortality.

Linear combination test for gene set analysis of a continuous phenotype.

BACKGROUND: Gene set analysis (GSA) methods test the association of sets of genes with a phenotype in gene expression microarray studies. Many GSA methods have been proposed, especially methods for use with a binary phenotype. Equally, if not more importantly however, is the ability to test the enrichment of a gene signature or pathway against the continuous phenotypes which are routinely and commonly observed in, for example, clinicopathological measurements. It is not always easy or meaningful to dichotomize continuous phenotypes into two classes, and attempting to do this may lead to the inaccurate classification of samples, which would affect the downstream enrichment analysis. In the present study, we have build on recent efforts to incorporate correlation structure within gene sets and pathways into the GSA test statistic. To address the issue of continuous phenotypes directly without the need for artificial discrete classification and thus increase the power of the test while ensuring computational efficiency and rigor, new GSA methods that can incorporate a covariance matrix estimator for a continuous phenotype may present an effective approach. RESULTS: We have designed a new method by extending the GSA approach called Linear Combination Test (LCT) from a binary to a continuous phenotype. Simulation studies and a real microarray dataset were used to compare the proposed LCT for a continuous phenotype, a modification of LCT (referred to as LCT2), and two publicly available GSA methods for continuous phenotypes. CONCLUSIONS: We found that the LCT methods performed better than the other two GSA methods; however, this finding should be understood in the context of our specific simulation studies and the real microarray dataset that were used to compare the methods. Free R-codes to perform LCT for binary and continuous phenotypes are available at http://www.ualberta.ca/~yyasui/homepage.html. The R-code to perform LCT for a continuous phenotype is available as Additional file 1.