Introduction: Emergency department (ED) crowding, long waits for care, and paramedic offload delay are of increasing concern. Older adults living in long-term care (LTC) are more likely to utilize the ED and are vulnerable to adverse events. We sought to identify existing programs that seek to avoid ED visits from LTC facilities where allied health professionals are the primary providers of the intervention and, to evaluate their efficacy and safety. Methods: We completed this systematic review based on a protocol we published apriori and following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement. We systematically searched Medline, CINAHL and EMBASE with terms relating to long-term care, emergency services, hospitalization and allied health personnel. Two investigators independently selected studies and extracted data using a piloted standardized form and evaluated the risk of bias of included studies. We report a narrative synthesis grouped by intervention categories. Results: We reviewed 11,176 abstracts and included 22 studies. Most studies were observational and few assessed patient safety. We found five categories of interventions including: 1) use of advanced practice nursing; 2) a program called Interventions to Reduce Acute Care Transfers (INTERACT); 3) end-of-life care; 4) condition specific interventions; and 5) use of extended care paramedics. Of the 13 studies that reported ED visits, all (100%) reported a decrease, and of the 16/17 that reported hospitalization, 94.1% reported a decrease. Patient adverse events such as functional status and relapse were seldom reported (6/22) as were measures of emergency system function such as crowding/inability of paramedics to transfer care to the ED (1/22). Only 4/22 studies evaluated patient mortality and 3/4 found a non-statistically significant worsening. When measured, studies reported decreased hospital length of stay, more time spent with patients by allied health professionals and cost savings. Conclusion: We found five types of programs/interventions which all demonstrated a decrease in ED visits or hospitalization. Many identified programs focused on improved primary care for patients. Interventions addressing acute care issues such as those provided by community paramedics, patient preferences, and quality of life indicators all deserve more study.

Introduction: Prognostication is a significant challenge early in the post-cardiac arrest period. Common prognostic factors for neurological survival are unreliable (high false positive rates) until 72 hours post-cardiac arrest. It is not known whether there are a combination of factors that can be utilized earlier in the post-cardiac arrest period to accurately predict patient outcome. Our objective was to predict neurological outcome utilizing a novel combination of patient factors early in the post-cardiac arrest period. Methods: We conducted a retrospective cohort study using data from our local cardiac arrest registry. We included adult patients who obtained a return of spontaneous circulation (ROSC) after out-of-hospital cardiac arrest (OHCA). We excluded patients who did not survive for at least 24 hours post-ROSC and those who had a do not resuscitate (DNR) order within 2 hours of ROSC. We performed an ordinal regression analysis using the proportional odds model to predict neurological outcome (modified rankin score (mRS)). We included a good neurological outcome (mRS 0-2), poor neurological outcome (mRS 3-5), and dead (mRS 6) as an ordinal outcome. We included a number of patient demographics, intra- and post-cardiac arrest factors as covariates in our model. The predictive performance of our model was analyzed using receiver operating characteristic (ROC) curves for discrimination and Brier statistic for calibration. Results: We included 3448 patients in our analysis. We found that an initial shockable rhythm (odds ratio (OR) 4.1; 95% confidence interval (CI) 3.6, 5.4), the absence of pupillary reflexes (OR 3.5; 95% CI 2.4,4.8) and maximum motor score on the Glasgow Coma Scale (GCS) (OR 1.5; 95% CI 1.4,1.6) had the greatest association with improved neurologic outcome. Longer duration of resuscitation was associate with worse outcomes (OR 0.84, 95% CI 0.82,0.87). The overall performance of our model was excellent with an area under the ROC curve of 0.89 and a Brier statistic of 0.13. Conclusion: Our model predicted good neurological outcome with a high rate of accuracy, however external validation of the model is required. This model may be useful in providing initial risk stratification of patients in clinical practice and future research on post-cardiac arrest care.

Introduction: Chest pain and symptoms of acute coronary syndrome are a leading cause of emergency department (ED) visits in Canada. Validated 2-hour high-sensitivity troponin algorithms can rapidly and accurately rule-in or rule-out myocardial infarction (MI) in most patients. The objective of this study was to quantify the incidence and timing of major adverse cardiac events (MACE: MI, death, or urgent revascularization) in the 30-days following the index ED encounter among patients who had MI ruled out using a 2-hour high-sensitivity troponin T (hs-cTnT) algorithm. We also sought to identify patient characteristics associated with very low risk of MACE. Methods: This was a secondary analysis of data prospectively collected from adult patients presenting with a primary complaint of chest pain or symptoms of ACS. This analysis focused on patients who had an MI ruled out using a validated 2-hour serial hs-cTnT diagnostic algorithm. Incidence of 30-day MACE was quantified. Sex-specific Kaplan-Meier curves were constructed to describe timing of MACE events after MI rule-out. Demographic and clinical variables of patients who did or did not have MACE were compared using simple bivariable analyses. Results: This analysis included 550 patients with serial 2h hs-cTnT testing. Of these, MI was ruled out in 344 (62.5% of patients), ruled in 67 (12.2%), and 139 (25.3%) had nondiagnostic hs-cTnT results. Among the 344 patients who had MI ruled out, 11 (3.2%) experienced a MACE in the 30 days following their index ED encounter. These included 10 (2.9%) unplanned revascularizations and 1 (0.3%) fatal MI. MACE occurred at a median of 5 days (range: 0-23 days) after the index ED encounter. Of the 11 patients experiencing MACE, 9 (81.8%) had a normal ECG at their index ED encounter. None of the 93 (27.0%) ruled-out patients under the age of 50 experienced a MACE in the follow-up period. Patients experiencing MACE were more likely to have a history of coronary disease and multiple vascular risk factors compared to those not experiencing MACE. Conclusion: The validated 2h hs-cTnT AMI algorithm ruled-out MI in a large proportion of patients. The 30-day MACE incidence after MI rule-out was 3%. Most MACE events were unplanned revascularizations. We determined that age < 50 was associated with event-free survival and may be of value in identifying patients who do not need additional cardiac testing after MI has been ruled out using high-sensitivity troponin testing.

Introduction: The Prehospital Evidence-based Practice (PEP) program is an online, freely accessible, continuously updated repository of appraised EMS research evidence. This report is an analysis of published evidence for EMS interventions used to assess and treat patients suffering from hypoglycemia. Methods: PubMed was systematically searched in June 2019. One author screened titles, abstracts and full-texts for relevance. Trained appraisers reviewed full text articles, scored each on a three-point Level of Evidence (LOE) scale (based on study design and quality) and three-point Direction of Evidence (DOE) scale (supportive, neutral, or opposing findings for each intervention's primary outcome), abstracted the primary outcome, setting and assigned an outcome category (patient or process). Second party appraisal was conducted for all included studies. The level and direction of each intervention was plotted in an evidence matrix, based on appraisals. Results: Twenty-nine studies were included and appraised for seven interventions: 5 drugs (Dextrose 50% (D50), Dextrose 10% (D10), glucagon, oral glucose and thiamine), one assessment tool (point-of-care (POC) glucose testing) and one call disposition (treat-and-release). The most frequently reported study primary outcomes were related to: clinical improvement (n = 15, 51.7%), feasibility/safety (n = 8, 27.6%), and diagnostics (n = 6, 20.7%). The majority of outcomes were patient focused (n = 18, 62.0%). Conclusion: EMS interventions for treating hypoglycemia are informed by high-quality supportive evidence. Both D50 and D10 are supported by high-quality evidence; suggesting D10 may be an effective alternative to the standard D50. “Treat-and-release” practices for hypoglycemia are supported by moderate-quality evidence for the patient related outcomes of relapse, patient preference and complications. This body of evidence is high-quality, patient-focused and conducted in the prehospital setting thus generalizable paramedic practice.

Introduction: Time-to-treatment plays a pivotal role in survival from sudden cardiac arrest (SCA). Every minute delay in defibrillation results in a 7-10% reduction in survival. This is particularly problematic in rural and remote regions, where bystander and EMS response is often prolonged and automated external defibrillators (AED) are often not available. Our objective was to examine the feasibility of a novel AED drone delivery method for rural and remote SCA. A secondary objective was to compare times between AED drone delivery and ambulance response to various mock SCA resuscitations. Methods: We conducted 6 simulations in two different rural communities in southern Ontario. During phase 1 (4 simulations) a “mock” call was placed to 911 and a single AED drone and an ambulance were simultaneously dispatched from the same location to a pre-determined destination. Once on scene, trained first responders retrieved the AED from the drone and initiated resuscitative efforts on a manikin. The second phase (2 scenarios) were done in a similar manner save for the drone being dispatched from a regionally optimized location for drone response. Results: Phase 1: The distance from dispatch location to scene varied from 6.6 km to 8.8 km. Mean (SD) response time from 911 call to scene arrival was 11.2 (+/- 1.0) minutes for EMS compared to 8.1 (+/- 0.1) for AED drone delivery. In all four simulations, the AED drone arrived before EMS, ranging from 2.1 to 4.4 minutes faster. The mean time for trained responders to retrieve the AED and apply it to the manikin was 35 (+/- 5) sec. No difficulties were encountered in drone activation by dispatch, drone lift off, landing or removal of the AED from the drone by responders. Phase 2: The ambulance response distance was 20km compared to 9km for the drone. Drones were faster to arrival at the scene by 7 minutes and 8 minutes with AED application 6 and 7 minutes prior to ambulance respectively. Conclusion: This implementation study suggests AED drone delivery is feasible with improvements in response time during a simulated SCA scenario. These results suggest the potential for AED drone delivery to decrease time to first defibrillation in rural and remote communities. Further research is required to determine the appropriate distance for drone delivery of an AED in an integrated EMS system as well as optimal strategies to simplify bystander application of a drone delivered AED.

Introduction: An increasing number of Canadian paramedic services are creating Community Paramedic programs targeting treatment of long-term care (LTC) patients on-site. We explored the characteristics, clinical course and disposition of LTC patients cared for by paramedics during an emergency call, and the possible impact of Community Paramedic programs. Methods: We completed a health records review of paramedic call reports and emergency department (ED) records between April 1, 2016 and March 31, 2017. We utilized paramedic dispatch data to identify emergency calls originating from LTC centers resulting in transport to one of the two EDs of the Ottawa Hospital. We excluded patients with absent vital signs, a Canadian Triage and Acuity Scale (CTAS) score of 1, and whose transfer to hospital were deferrable or scheduled. We stratified remaining cases by month and selected cases using a random number generator to meet our apriori sample size. We collected data using a piloted standardized form. We used descriptive statistics and categorized patients into groups based on the ED care received and if the treatment received fit into current paramedic medical directives. Results: Characteristics of the 381 included patients were mean age 82.5 years, 58.5% female, 59.7% hypertension, 52.6% dementia and 52.1% cardiovascular disease. On arrival at hospital, 57.7% of patients waited in offload delay for a median time of 45 minutes (IQR 33.5-78.0). We could identify 4 groups: 1) Patients requiring no treatment or diagnostics in the ED (7.9%); 2) Patients receiving ED treatment within current paramedic medical directives and no diagnostics (3.2%); 3) Patients requiring diagnostics or ED care outside current paramedic directives (54.9%); and 4) patients requiring admission (34.1%). Most patients were discharged from the ED (65.6%), and 1.1% died. The main ED diagnoses were infection (18.6%) and musculoskeletal injury (17.9%). Of the patients that required ED care but were discharged, 64.1% required x-rays, 42.1% CT, and 3.4% ultrasound. ED care included intravenous fluids (35.7%), medication (67.5%), antibiotics (29.4%), non-opioid analgesics (29.4%) and opioids (20.7%). Overall, 11.1% of patients didn't need management beyond current paramedic capabilities. Conclusion: Many LTC patients could receive care by paramedics on-site within current medical directives and avoid a transfer to the ED. This group could potentially grow using Community Paramedics with an expanded scope of practice.

Introduction: eCTAS is a real time electronic triage decision-support tool designed to improve patient safety and quality of care by standardizing the application of the Canadian Triage and Acuity Scale (CTAS). The tool dynamically calculates a recommended CTAS score based on the presenting complaint, vital signs and selected clinical modifiers. The primary objective was to assess consistency of CTAS score distributions across 35 emergency departments (EDs) by 16 presenting complaints pre and post eCTAS implementation. Methods: This retrospective cohort study used population-based administrative data from January 2016 to December 2018 from all hospital EDs in Ontario that had implemented eCTAS with at least 9 months of data. Following a 3-month stabilization period, we compared data for 6 months post-eCTAS implementation to the same 6-month period the previous year (pre-implementation) to account for potential seasonal variation, patient volume and case-mix. We included triage encounters of adult (≥18 years) patients if they had one of 16 pre-specified high-volume, presenting complaints. A paired-samples t-test was used to determine consistency by estimating the absolute difference in CTAS distribution for each presenting complaint, by each hospital, pre and post eCTAS implementation, compared to the overall average of the 35 EDs. Results: There were 183,231 triage encounters in the pre-eCTAS cohort and 179,983 in the post-eCTAS cohort from 35 EDs across the province. Triage scores were more consistent with the overall average after eCTAS implementation in 6 (37.5%) presenting complaints: chest pain (cardiac features) (p < 0.001), extremity weakness/symptoms of cerebrovascular accident (p < 0.001), fever (p < 0.001), shortness of breath (p < 0.001), syncope (p = 0.02), and hyperglycemia (p = 0.03). Triage consistency was similar pre and post eCTAS implementation for the presenting complaints of altered level of consciousness, anxiety/situational crisis, confusion, depression/suicidal/deliberate self-harm, general weakness, head injury, palpitations, seizure, substance misuse/intoxication or vertigo. Conclusion: A standardized, electronic approach to performing triage assessments increased consistency in CTAS scores across many, but not all, high-volume CEDIS complaints. This does not reflect triage accuracy, as there are no known benchmarks for triage accuracy. Improvements in consistency were greatest for sentinel presenting complaints with a minimum allowable CTAS score.

Introduction: The Canadian Syncope Risk Score (CSRS) is a validated risk tool developed using the best practices of conventional biostatistics, for predicting 30-day serious adverse events (SAE) after an Emergency Department (ED) visit for syncope. We sought to improve on the prediction ability of the CSRS and compared it to physician judgement using artificial intelligence (AI) research with modern machine learning (ML) methods. Methods: We used the prospective multicenter cohort data collected for the CSRS derivation and validation at 11 EDs across Canada over an 8-year period. The same 43 candidate variables considered for CSRS development were used to train and validate the four classes of ML models to predict 30-day SAE (death, arrhythmias, MI, structural heart disease, pulmonary embolism, hemorrhage) after ED disposition. Physician judgement was modeled using the two variables, referral for consultation and hospitalization. We compared the area under the curve (AUC) for the three models. Results: The proportion of patients who suffered 30-day SAE in the derivation cohort (N = 4030) was 3.6% and in validation phase (N = 2290) was 3.4%. Characteristics of the both cohorts were similar with no shift. The best performing ML model, a gradient boosting tree-based model used all 43 variables as predictors as opposed to the 9 final CSRS predictors. The AUC for the three models on the validation data were: best ML model 0.91 (95% CI 0.87–0.93), CSRS 0.87 (95% CI 0.83–0.90) and physician judgment 0.79 (95% CI 0.74 - 0.84). The most important predictors in the ML model were the same as the CSRS predictors. Conclusion: A ML model developed using AI method for risk-stratification of ED syncope performed with slightly better discrimination ability though not significantly different when compared to the CSRS. Both the ML model and the CSRS were better predictors of poor outcomes after syncope than physician judgement. ML models can perform with similar discrimination abilities when compared to traditional statistical models and outperform physician judgement given their ability to use all candidate variables.

Introduction: Patients presenting to the emergency department (ED) with hypotension have a high mortality rate and require careful yet rapid resuscitation. The use of cardiac point of care ultrasound (PoCUS) in the ED has progressed beyond the basic indications of detecting pericardial fluid and activity in cardiac arrest. We examine if finding left ventricular dysfunction (LVD) on emergency physician performed PoCUS reliably predicts the presence of cardiogenic shock in hypotensive ED patients. Methods: We prospectively collected PoCUS findings performed in 135 ED patients with undifferentiated hypotension as part of an international study. Patients with clearly identified etiologies for hypotension were excluded, along with other specific presumptive diagnoses. LVD was defined as identification of a generally hypodynamic LV in the setting of shock. PoCUS findings were collected using a standardized protocol and data collection form. All scans were performed by PoCUS-trained emergency physicians. Final shock type was defined as cardiogenic or non-cardiogenic by independent specialist blinded chart review. Results: All 135 patients had complete follow up. Median age was 56 years, 53% of patients were male. Disease prevalence for cardiogenic shock was 12% and the mortality rate was 24%. The presence of LVD on PoCUS had a sensitivity of 62.50% (95%CI 35.43% to 84.80%), specificity of 94.12% (88.26% to 97.60%), positive-LR 10.62 (4.71 to 23.95), negative-LR 0.40 (0.21 to 0.75) and accuracy of 90.37% (84.10% to 94.77%) for detecting cardiogenic shock. Conclusion: Detecting left ventricular dysfunction on PoCUS in the ED may be useful in confirming the underlying shock type as cardiogenic in otherwise undifferentiated hypotensive patients.

Introduction: Distributing take-home naloxone (THN) kits from Emergency Departments (EDs) is an important strategy for preventing opioid overdose deaths. However, there is a lack of clear operational guidance for implementing ED-based THN programs. This scoping review had two objectives: 1) identify key strategies for THN distribution in EDs, and 2) develop a theory-informed implementation model that can be used to optimize the effectiveness of ED-based THN programs. Methods: We systematically searched health science databases through April 18, 2019. The search strategy combined terms representing the ED, naloxone, and take-home kits/bystander administration. Two reviewers independently screened the search results. We included all peer-reviewed articles that described THN distribution within EDs. A standardized form was used for data extraction. Included studies were coded by two reviewers and mapped to domains of the Consolidated Framework for Implementation Research (CFIR). A third reviewer with content expertise adjudicated disagreements in record screening and data coding. Results: Database searching retrieved 717 records after duplicates were removed. 87 full-text studies were assessed for eligibility. Two studies were added through other sources, resulting in a total of 21 studies included in the final review. Of note, 14 studies evaluated existing ED-based THN programs. We synthesized themes that emerged within each CFIR domain and identified four key implementation strategies: 1) develop ED policies on opioid harm reduction; 2) collaborate with community and government partners to ensure programs meet patient needs; 3) address provider attitudes and knowledge gaps through dedicated training; and 4) establish guidelines to identify patients who are at risk of opioid overdose, and engage at-risk patients to maximize THN acceptance. Conclusion: ED-based THN programs must be tailored to local community needs and available hospital resources. Innovative implementation strategies are needed to promote ED provider engagement, and reduce barriers to patient acceptance of THN in the ED. This scoping review highlights key considerations for ED-THN implementation that can guide EDs to establish new programs, or refine existing programs to maximize their effectiveness.

Introduction: Frequent users to emergency departments (EDs) are a diverse group of patients with a disproportionate number of ED presentations. This study aimed to compare sociodemographic and clinical characteristics of adult high system users (HSUs) and control groups in two provinces. Methods: Cohorts of HSUs were created for Alberta and Ontario by identifying the patients with the top 10% of ED presentations in the National Ambulatory Care Reporting System during April 2015 to March 2016. Random samples of patients not in the HSU groups were selected in each province as controls (4:1 ratio). Sociodemographic and presentation data (e.g., Canadian Triage and Acuity Scale [CTAS], disposition) were extracted and compared using separate logistic regression models. Results: In Alberta, 101,250 HSU patients made 686,918 ED presentations (median [med] = 5 interquartile range [IQR] 4,7 presentations per patient), compared with 401,923 controls who made 560,765 ED presentations (med = 1 IQR 1,2 per patient). HSUs were more likely to be female (odds ratio (OR) = 1.20 95% confidence interval (CI) 1.18,1.22), older (OR = 1.03 per 5y 95%CI 1.03,1.03), live closer to hospital (OR = 1.02 per 100km 95%CI 1.00,1.03), and be from the lowest income quintile (OR = 1.39 95%CI 1.37,1.42) than controls. In Ontario, 478,424 HSUs made 2,222,487 ED presentations (med = 4 IQR 3,5 per patient) and 1,714,037 controls made 2,114,070 ED presentations (med = 1 IQR 1,1 per patient). Ontario HSUs were also more likely to be female (OR = 1.13 95%CI 1.12,1.14), older (OR = 1.03 per 5y 95%CI 1.03,1.03), and from the lowest income quintile (OR = 1.41 95%CI 1.40,1.42) than controls, but were less likely to live closer to hospital (OR = 0.93 per 100km 95%CI 0.92, 0.93). Higher acuity was seen in Ontario (CTAS 1/2 vs. others OR = 1.05 95%CI 1.04,1.06) but not for Alberta (CTAS 1/2 vs others OR = 0.75, 95%CI 0.74,0.76). Discharges were less likely in the HSUs compared to controls (Alberta OR = 0.89 95%CI 0.88,0,90; Ontario OR = 0.65 95%CI 0.65,0.66). HSUs were more likely to leave without being seen (Alberta OR = 1.10 95%CI 1.07,1.13; Ontario OR = 1.37 95%CI 1.35,1.40) and against medical advice (Alberta OR = 1.47 95%CI 1.41,1.53; Ontario OR = 1.67 95%CI 1.63,1.71). Conclusion: HSUs were more likely to be female, older, and poorer than controls. Ontario HSUs had higher acuity than the other groups. Disposition differed for HSUs and controls. Further study is required to identify ways to safely reduce ED utilization by HSUs.

Introduction: There is ongoing concern about the burden placed on healthcare systems by lab tests. Although these concerns are widespread, it is difficult to quantify the extent of the problem. One approach involves use of a metric known as the Mean Abnormal Response Rate (MARR), which is the proportion of tests ordered that return an abnormal result; a higher MARR value indicates higher yield. The primary objective of this study was to calculate MARRs for tests ordered between April 2014 and March 2019 at the four adult emergency departments (EDs) covering a metropolitan population of 1.3 million. Secondary objectives included identifying tests with highest and lowest MARRs; comparison of MARRs for nurse- and physician-initiated orders; correlation of the number of tests per order requisition to MARR; and correlation of physician experience to MARR. Methods: In total, 40 laboratory tests met inclusion criteria for this study. Administrative data on these tests as ordered at the four EDs were obtained and analyzed. Multi-component test results, such as from CBC, were consolidated such that an abnormal result for any component was coded as an abnormal result for the entire test. Repeat tests ordered within a single patient visit were excluded. Physician experience was quantified for 209 ED physicians as number of years since licensure. Analyses were descriptive where appropriate for whole-population data. Risk of bias was attenuated by the focus on administrative data. Results: The population dataset comprised 33,757,004 test results on 415,665 unique patients. Of these results, 30.3% were the outcomes of nurse-initiated orders. The 5-year MARRs for the four hospitals were 38.3%, 40.0%, 40.7% and 40.9%. The highest per-test MARRs were for BNP (80.5%) and CBC (62.6%), while the lowest were for glucose (7.9%) and sodium (11.6%). MARRs were higher for nurse-initiated orders than for physician-initiated orders (44.7% vs. 38.1%), likely due to the greater order frequency of high-yield CBC in nurse-initiated orders (38.6% vs. 18.1%). The number of tests per order requisition was inversely associated with MARR (r = -0.90, p < 0.001). Finally, the number of years since licensure was modestly but significantly associated with MARR (r = 0.28, p < 0.001). Conclusion: This is the first and largest study to apply the MARR in an ED setting. As a metric, MARR effectively identifies differences in test ordering practices on per-test and per-hospital bases, which could be useful for data-informed practice optimization.

Introduction: The Choosing Wisely Canada (CWC) initiative is dedicated towards optimizing patient care and reduce unnecessary resource use. Different specialty organizations create recommendations lists towards these outcomes. The goal of this study was to examine the applicability of non-Emergency Medicine (EM) recommendations towards EM practice. Methods: The entire master recommendations listings spreadsheet was downloaded from the CWC website (March 2019; n = 333). The EM-specific items from the CAEP checklist were deliberately excluded (n = 10). Items were rated by Niagara community EM physicians (n = 7) using the previously validated Best Evidence in Emergency Medicine (BEEM) rating scale (7 point Likert scale) to determine potential impact on EM practice. Items rated “6 or 7/7” were determine as “high relevance.” Redundant items were consolidated. Results: From the retrieved CWC master list, a total of 102 “highly relevant” recommendations were identified (41 items scored 6/7 [12%], 61 scored 7/7 [18%]; total 31%). Redundant items consolidated included antimicrobial avoidance (n = 18), opioid avoidance for pain (n = 11), reduction of unnecessary imaging (n = 11), and avoidance of routine low back imaging (n = 7). Conclusion: There are a large number of non-EM specialty recommendations highly relevant to EM practice in the CWC database (31%). Quality improvement initiatives looking to operational CWC recommendations in Canadian Emergency Departments should be aware of these as a part of optimizing patient care.

Introduction: Overcrowding in the Emergency Department (ED) results in delays in care, and increased patient morbidity and mortality. Innovative departmental approaches have the potential to make patient flow through the ED more efficient and reduce overcrowding by improving patient throughput. The Calgary zone ED recently piloted a new physician role, the Emergency Physician Lead (EPL), a senior physician working closely with the charge nurse and consulting services to provide physician leadership, and to troubleshoot flow issues and safety breeches such as EMS offload delays and long emergency inpatient (EIP) stays. The objective of this study was to evaluate the efficacy of the EPL by determining its effect on key metrics of patient flow, and by identifying which specific EPL interventions were most effective at improving patient throughput. Methods: A retrospective cohort design was used to compare Foothills Medical Centre (FMC) ED patients seen by the EPL from March-June 2019 (n = 1343 patients) with a control group from the same period in 2018 (n = 5530). An EMR search was used to collect patient data and generate descriptive statistics, which were compared between groups by Mann-Whitney U-test. Patient handover notes left by the EPL were also collected and analyzed by two independent assessors to develop a list of actions taken by the EPL. Each patient was then coded based on the actions in the handover note, and means for each coded group were compared to control to find correlations between action and changes in key flow metrics. Results: Patients whose care involved the EPL had a 40% shorter average ED length of stay (ELOS) compared to control (515 vs 865 min, p < 0.001). The EPL was especially effective for patients with ELOS above the 90th percentile, with a 58% relative reduction. EPL patients also had lower average times from first contact with the department to first order being placed (79 vs 143 min, p < 0.001), and spent less time as EIPs after being admitted (390 vs 515 mins, p < 0.001). EPL actions aimed at early ordering of investigations or early management showed the largest relative reductions in ELOS, followed by actions related to resolving issues with consulting services (56% and 48% respectively, p < 0.001). Conclusion: The EPL role appears to be associated with improvements in several key metrics of patient flow. Specific EPL actions were correlated with marked decreases in length of stay. The EPL may be an effective strategy to improve patient throughput and combat ED overcrowding.

Introduction: Emergency Department (ED) crowding is the primary threat to emergency care quality. Input and outflow factors are important factors, but EDs must optimize throughput efficiency by improving internal processes from triage to disposition, and triage is the first throughput phase. Triage throughput interventions exclude strategies that direct patients away from the ED (these modify input rather than throughput). Previous research has described physicians in triage, team triage, telemedical triage, and nurse practitioner (NP) or physician assistant (PA) led triage, but their impact has never been systematically evaluated. Methods: We conducted systematic database searches in Medline, Embase, CINAHL, and the Cochrane Central Register of Controlled Trials without the use of filters or language restrictions of all triage interventions that effected ED throughput (PROSPERO:CRD42019125651). Two independent reviewers screened studies. Study quality was assessed using the Cochrane Risk of Bias tool (version 2) for randomized controlled trials, and the National Heart, Lung, and Blood Institute quality assessment tool for other designs. Results: 18 studies met inclusion criteria (Cohen's k = 0.69). Study results were not pooled due to high statistical heterogeneity as assessed by chi-squared and I-squared statistics. Studies were grouped into physician led, NP or PA led, and team triage interventions. Six physician in triage interventions reported LOS changes between -82 and + 18 minutes. Five NP/PA led triage interventions resulted in LOS changes of -106 to + 19 minutes. Five team triage interventions reported LOS reductions of 4 to 34 minutes. One telemedicine triage study reported a non-significant 8 minute increase in LOS. Six physician at triage interventions yielded significant LWBS rate improvement (relative risk {RR}= 0.29-0.82). Team triage interventions generated LWBS rate changes ranging from meaningful improvement (RR = 0.58) to substantial deterioration (RR = 1.68). Five studies have low risk of bias, 11 studies have some risk of bias, and 2 studies have high risk of bias (Cohen's kappa = 0.58). Conclusion: Fourteen of 18 triage interventions reduced EDLOS and/or LWBS rate. Physician, NP and PA led triage were the most effective triage interventions. To aid widespread adoption, future research should focus on interrupted time series or RCT designs, and more comprehensive descriptions of the contextual factors affecting implementation of these interventions.

Introduction: Emergency Department (ED) crowding is an intensifying crisis. While input, throughput, and output factors all contribute to crowding, throughput factors are the most dependent on ED staff and process. Diagnostic testing is a fundamental ED process that has not been systematically evaluated. We present a systematic review of interventions designed to reduce ED length of stay (LOS) by optimizing laboratory or imaging turnaround time, or by introducing point-of-care testing (POCT). Methods: We conducted systematic database searches in Medline, Embase, CINAHL, and the Cochrane Central Register of Controlled Trials without filters or language restrictions, of all interventions on diagnostic technology that affected ED throughput (PROSPERO:CRD42019125651). Studies were screened by two independent reviewers. Study quality was assessed using the Cochrane ROB-2 tools for randomized controlled trials (RCTs), and the National Heart, Lung, and Blood Institute tool for all other study designs. Results: 18 studies met inclusion criteria (Cohen's kappa = 0.69). Study results were not pooled due to high statistical heterogeneity as assessed by chi-squared and I-squared statistics. 12 POCT intervention studies reported LOS changes ranging from -114 to + 8 minutes (-26.8% to + 3.8%), although three were non-significant findings. Four studies that initiated POCT or lab-ordering at triage reported LOS reductions ranging from 22 to 174 minutes, but only one of these, at 29 minutes (16%), was statistically significant. One study of improved laboratory troponin processing reported a LOS reduction of 43 minutes (12.3%). Another, which allowed triage nurses to order ankle x-rays using the Ottawa ankle rules, reported a non-significant LOS reduction of 28 minutes for patients with ankle injuries. LOS improvements reflected the population of patients who underwent the testing modality, rather than overall ED LOS. Seven studies had low risk of bias, 11 studies had some risk of bias, and no studies had high risk of bias (Cohen's kappa = 0.58). Conclusion: Eleven of 18 diagnostic testing studies reported LOS reductions. POCT was the most common intervention type, and usually reduced EDLOS within relevant patient subsets, while triage-initiated testing generally did not. To aid widespread adoption, future research should focus on interrupted time series or RCT designs, and more comprehensive descriptions of the contextual factors affecting implementation of these interventions.

Introduction: CTAS is a validated five-level triage score utilized in EDs across Canada and internationally. Moderate interrater reliability between prehospital paramedic and triage nurse application of CTAS during clinical practice has been found. This study is the first assessment of the variation in distribution of CTAS scores with increasing departmental pressure as measured by the NEDOCs scale comparing triage allocations made by triage nurses with those made by triage paramedics. Methods: We conducted a retrospective, observational cohort study of EDIS data of all patients triaged in the Halifax Infirmary Emergency Department from January 1, 2017-May 30, 2017 and January 1, 2018 - May 30, 2018. CTAS score assignment by nursing and paramedic triage staff were compared with increasing levels of ED overcrowding, as determined by the department NEDOCS score. Results: Nurses were more likely to assign higher acuity scores in all situations of department crowding; there was a 3% increased probability that a nurse, as compared to a paramedic, would triage as emergent when the ED was not overcrowded (Pearson chi-square(1) = 4.21, p < 0.05, Cramer's v = 0.028, n = 5314), and a 10% increased probability that a nurse, as compared to a paramedic, would triage a patient as emergent when EDs were overcrowded (Pearson chi-square(1) = 623.83, p < 0.001, Cramer's v = 0.11, n = 56 018). Conclusion: Increasing levels of ED overcrowding influence triage nurse CTAS score assignment towards higher acuity to a greater degree than scores assigned by triage paramedics.

Introduction: Emergency department (ED) crowding is a major problem across Canada. We studied the ability of artificial intelligence methods to improve patient flow through the ED by predicting patient disposition using information available at triage and shortly after patients’ arrival in the ED. Methods: This retrospective study included all visits to an urban, academic, adult ED between May 2012 and June 2019. For each visit, 489 variables were extracted including triage data that had been collected for use in the Canadian Triage Assessment Scale (CTAS) and information regarding laboratory tests, radiological tests, consultations and admissions. A training set consisting of all visits from April 2012 up to December 2018 was used to train 5 classes of machine learning models to predict admission to the hospital from the ED. The models were trained to predict admission at the time of the patient's arrival in the ED and every 30 minutes after arrival until 6 hours into their ED stay. The performance of models was compared using the area under the ROC curve (AUC) on a test set consisting of all visits from January 2019 to June 2019. Results: The study included 536,332 visits and the admission rate was 15.0%. Gradient boosting models generally outperformed other machine learning models. A gradient boosting model using all available data at 2 hours after patient arrival in the ED yielded a test set AUC 0.92 [95% CI 0.91-0.93], while a model using only data available at triage yielded an AUC 0.90 [95% CI 0.89-0.91]. The quality of predictions generally improved as predictions were made later in the patient's ED stay leading to an AUC 0.95 [95% CI 0.93-0.96] at 6 hours after arrival. A gradient boosting model with 20 variables available at 2 hours after patient arrival in the ED yielded an AUC 0.91 [95% CI 0.89-0.93]. A gradient boosting model that makes predictions at 2 hours after arrival in ED using only variables that are available at all EDs in the province of Quebec yielded an AUC 0.91 [95% 0.89-0.92]. Conclusion: Machine learning can predict admission to a hospital from the ED using variables that area collected as part of routine ED care. Machine learning tools may potentially be used to help ED physicians to make faster and more appropriate disposition decisions, to decrease unnecessary testing and alleviate ED crowding.

Background: Every 30-minute delay to ST-Elevation Myocardial Infarction (STEMI) reperfusion increases one-year mortality by 7.5%. A local audit found that the third of patient electrocardiograms (ECGs) not initially meeting classic STEMI criteria had an ECG-to-Activation (ETA) time of over 90 minutes, more than five times that of classic STEMIs. However, three quarters of “STEMI negative” ECGs met STEMI-equivalent patterns or rules for subtle occlusion, uncovering an opportunity for improvement. Aim Statement: We aimed to reduce ETA time, from initial emergency department (ED) ECG to activation of the cath lab, for patients whose ECGs did not meet classic STEMI criteria, by 30 minutes within one year (i.e. by Dec 2019). Measures & Design: We reviewed all ED Code STEMIs over a 35-month pre-intervention period. Root Cause analyses, including Ishikawa diagram and Pareto chart, led to our Plan-Do-Study-Act cycles: 1) a survey to engage our team; 2) a Grand Rounds presentation as an educational strategy; and 3) weekly web-based feedback to all ED physicians on STEMI-equivalents and subtle occlusions, using recent local cases. Our outcome measures were ETA times, stratified by ECGs not initially meeting STEMI criteria (primary) and those that did (secondary). Our process measures were the number of website visits and page views. Our balancing measure was the proportion of Code STEMIs without culprit lesion. We used Statistical Process Control (SPC) charts with usual special cause variation rules. Evaluation/Results: ETA time for the 37.5% of 56 ECGs that did not meet classic STEMI criteria decreased from 97.5 to 53.7 minutes (min), a 43.8-min absolute decrease (p = 0.037), while those meeting STEMI criteria remained the same (16.5 to 18.2min, p = 0.75). SPC charts did not show special cause variation. There were 2,634 page views (65.9/week) and 1,092 visits (27.3/week), in a group of 80 physicians—i.e. a third of the group each week. There was no change in Code STEMIs without culprit lesions (28.0 % to 23.3%, p = 0.41). Discussion/Impact: We reduced ETA time by 43.8min for the one third of patients with culprit lesions not initially meeting classic STEMI criteria, a magnitude associated with mortality impact. To do so, we used a multi-modal educational strategy including a novel web-based feedback approach to all ED physicians. Local feedback and education on this challenging-to-diagnose subgroup, guided by ETA time as a quality metric, could be replicated in other centres.

Background: Pediatric pain is often under-treated in emergency departments (EDs), causing short and long-term harm. In Alberta EDs, children's pain outcomes were unknown. A recent quality improvement collaborative (QIC) led by our team improved children's pain care in 4 urban EDs. We then spread to all EDs in Alberta using the Institute for Healthcare Improvement Framework for Going to Full Scale. Aim Statement: To increase the proportion of children <12 years who receive topical anesthetic before needle procedures from 11% to 50%; and for children <17 years with fractures: to 1) increase the proportion receiving analgesia from 31% to 50%; 2) increase the proportion with pain score documentation from 24% to 50%, and 3) reduce time to analgesia from 60 to 30 minutes, within 1 year. Measures & Design: All 97 EDs in Alberta that treat children were invited. Each was asked to form a project team, attend webinars, develop key driver diagrams and perform PDSA tests of change. Sites were given a monthly list of randomly selected charts for audit and entered data in REDCap for upload to a provincial run chart dashboard. Baseline performance measurement informed aims. Measures included proportion of children <12 years undergoing a lab test who received topical anesthetic, and for children <17 years with fracture, the proportion with a pain score, proportion receiving analgesia and median minutes to analgesia. Length of stay and use of opioids were balancing measures. Control charts were used to detect special cause. Interrupted time series (ITS) was performed to assess significance and trends. Evaluation/Results: 36 sites (37%) participated, including rural and urban sites from all regions. 8417 visits were audited. 23/36 sites completed audits before and after tests of change and were analyzed. Special cause occurred for all aims. The proportion receiving topical anesthetic increased from 11% to 30% (ITS p < 0. 001). For children with fractures, the proportion with pain scores increased from 24% to 34% (ITS p = 0.21, underlying trend present), proportion receiving analgesic medication increased from 31% to 39% (ITS p = 0.41, underlying trend present) and minutes to analgesia decreased from 60 to 28 (ITS p < 0. 01). There was no increase in length of stay or use of opioid medications. Discussion/Impact: A pragmatic approach encouraging locally led change was well-received and key to success. The QIC method shows promise for improving outcomes in diverse EDs across large geographic areas. Next steps include further spread and sustainability measurement.