Emergency department performance indicators that encompass the patient journey

Evaluation of different machine learning algorithms for predicting the length of stay in the emergency departments: a single-centre study

Background

Recently, crowding in emergency departments (EDs) has become a recognised critical factor impacting global public healthcare, resulting from both the rising supply/demand mismatch in medical services and the paucity of hospital beds available in inpatients units and EDs. The length of stay in the ED (ED-LOS) has been found to be a significant indicator of ED bottlenecks. The time a patient spends in the ED is quantified by measuring the ED-LOS, which can be influenced by inefficient care processes and results in increased mortality and health expenditure. Therefore, it is critical to understand the major factors influencing the ED-LOS through forecasting tools enabling early improvements.

Methods

The purpose of this work is to use a limited set of features impacting ED-LOS, both related to patient characteristics and to ED workflow, to predict it. Different factors were chosen (age, gender, triage level, time of admission, arrival mode) and analysed. Then, machine learning (ML) algorithms were employed to foresee ED-LOS. ML procedures were implemented taking into consideration a dataset of patients obtained from the ED database of the "San Giovanni di Dio e Ruggi d'Aragona" University Hospital (Salerno, Italy) from the period 2014-2019.

Results

For the years considered, 496,172 admissions were evaluated and 143,641 of them (28.9%) revealed a prolonged ED-LOS. Considering the complete data (48.1% female vs. 51.9% male), 51.7% patients with prolonged ED-LOS were male and 47.3% were female. Regarding the age groups, the patients that were most affected by prolonged ED-LOS were over 64 years. The evaluation metrics of Random Forest algorithm proved to be the best; indeed, it achieved the highest accuracy (74.8%), precision (72.8%), and recall (74.8%) in predicting ED-LOS.

Conclusions

Different variables, referring to patients' personal and clinical attributes and to the ED process, have a direct impact on the value of ED-LOS. The suggested prediction model has encouraging results; thus, it may be applied to anticipate and manage ED-LOS, preventing crowding and optimising effectiveness and efficiency of the ED.

Factors associated with longer wait times, admission and reattendances in older patients attending emergency departments: an analysis of linked healthcare data

BACKGROUND AND OBJECTIVE

Care for older patients in the ED is an increasingly important issue with the ageing society. To better assess the quality of care in this patient group, we assessed predictors for three outcomes related to ED care: being seen and discharged within 4 hours of ED arrival; being admitted from ED to hospital and reattending the ED within 30 days. We also used these outcomes to identify better-performing EDs.

METHODS

The CUREd Research Database was used for a retrospective observational study of all 1 039 251 attendances by 368 754 patients aged 75+ years in 18 type 1 EDs in the Yorkshire and the Humber region of England between April 2012 and March 2017. We estimated multilevel logit models, accounting for patients' characteristics and contact with emergency services prior to ED arrival, time variables and the ED itself.

RESULTS

Patients in the oldest category (95+ years vs 75-80 years) were more likely to have a long ED wait (OR=1.13 (95% CI=1.10 to 1.15)), hospital admission (OR=1.26 (95% CI=1.23 to 1.29)) and ED reattendance (OR=1.09 (95% CI=1.06 to 1.12)). Those who had previously attended (3+ vs 0 previous attendances) were more likely to have long wait (OR=1.07 (95% CI=1.06 to 1.08)), hospital admission (OR=1.10 (95% CI=1.09 to 1.12)) and ED attendance (OR=3.13 (95% CI=3.09 to 3.17)). Those who attended out of hours (vs not out of hours) were more likely to have a long ED wait (OR=1.33 (95% CI=1.32 to 1.34)), be admitted to hospital (OR=1.19 (95% CI=1.18 to 1.21)) and have ED reattendance (OR=1.07 (95% CI=1.05 to 1.08)). Those living in less deprived decile (vs most deprived decile) were less likely to have any of these three outcomes: OR=0.93 (95% CI=0.92 to 0.95), 0.92 (95% CI=0.90 to 0.94), 0.86 (95% CI=0.84 to 0.88). These characteristics were not strongly associated with long waits for those who arrived by ambulance. Emergency call handler designation was the strongest predictor of long ED waits and hospital admission: compared with those who did not arrive by ambulance; ORs for these outcomes were 1.18 (95% CI=1.16 to 1.20) and 1.85 (95% CI=1.81 to 1.89) for those designated less urgent; 1.37 (95% CI=1.33 to 1.40) and 2.13 (95% CI=2.07 to 2.18) for urgent attendees; 1.26 (95% CI=1.23 to 1.28) and 2.40 (95% CI=2.36 to 2.45) for emergency attendees; and 1.37 (95% CI=1.28 to 1.45) and 2.42 (95% CI=2.26 to 2.59) for those with life-threatening conditions. We identified two EDs whose patients were less likely to have a long ED, hospital admission or ED reattendance than other EDs in the region.

CONCLUSIONS

Age, previous attendance and attending out of hours were all associated with an increased likelihood of exceeding 4 hours in the ED, hospital admission and reattendance among patients over 75 years. These differences were less pronounced among those arriving by ambulance. Emergency call handler designation could be used to identify those at the highest risk of long ED waits, hospital admission and ED reattendance.

Door to Disposition Key Performance Indicator in Three Saudi Arabian Hospitals' Emergency Departments during COVID-19 Pandemic

Objective: The COVID-19 pandemic impacted health systems and Emergency Departments (ED) services worldwide. This study attempts to assess the impact of COVID-19 on the performance of the emergency department during COVID-19 in three hospitals in Riyadh city, Saudi Arabia. Methods: Ada’a data was used for this retrospective cohort study. The hospitals included in this study were: a 300-bed maternity and children’s hospital; a 643-bed general hospital; and a 1230-bed tertiary hospital. All patients who visited the ED in the time period from September 2019 to December 2021 were included. The outcome variable was the Door to Disposition (DTD) which estimates the percentage of patients seen within 4 h from Door to Disposition. A two-way ANOVA test was used to examine the differences in the outcome variable by hospital and by the phase of COVID-19. Results: Both hospital and the phase of COVID-19 were significantly different in terms of the percentage of patients seen within four hours in the ED (DTD) (p-value < 0.05). On average, the DTD percentages dipped slightly in the early phase of COVID-19 (64.0% vs. 69.8%) and jumped sharply in the later phase (73.6%). Additionally, the average DTD score for the maternity and pediatric hospital (87.6%) was sharply higher than both general and tertiary hospitals (63.2%, and 56.5%, respectively). Conclusion: COVID-19 led to a significant drop in emergency department services performance in the early stage of the pandemic as patients spent more time at the ED. However, for the designated COVID-19 hospital, the ED performance improved as more patients spent less than 4 h at the ED in the early stages of COVID-19. This is a clear indication that careful planning and management of resources for ED services during a pandemic is effective.

Characteristics and retention of emergency department patients who left without being seen (LWBS)

A retrospective observational study was conducted for patients 18 years or older presenting to a Midwestern emergency department (ED) in the United States during February 2019-January 2020 to characterize associated subsequent care utilization in patients who left the ED without being seen. Patients were classified as left without being seen (LWBS) based on documented ED disposition. The healthcare system's records were reviewed for any associated utilizations within 3 weeks following the initial ED encounter. During the study period, 45,456 patients presented to the ED, with 2269 (5.0%) classified as LWBS. The median documented time until patients left the ED was 112 min. Of these patients, 1257 (55.4%) had a subsequent encounter within the health system within 3 weeks and 920 (73.2%) of these visits were determined to be related to the LWBS chief complaint. These visits included 67.5% of patients returning to ED or hospital, 27.5% to primary care or an urgent care clinic, and 5.0% to a specialty or other provider appointment. Of patients returning to ED, 78.1% did so within 72 h. Patients without a subsequent health system associated encounter tended to be younger, female, non-White, and present with possible lower-acuity chief complaints. At least one-half of LWBS patients sought care related to the concerns by a health system provider within 3 weeks of the initial encounter within the same system. The high prevalence of ED returns within a narrow turnaround window highlights a missed opportunity to provide services to these patients during their initial encounter.

Emergency department waiting room nurse role: A key informant perspective

BACKGROUND

Emergency departments have become overcrowded with increased waiting times. Strategies to decrease waiting times include time-based key performance indicators and introduction of a waiting room nurse role. The aim of the waiting room nurse role is to expedite care by assessing and managing patients in the waiting room. There is limited literature examining this role.

METHODS

This paper presents results of semi-structured interviews with five key informants to explore why and how the waiting room nurse role was implemented in Australian emergency departments. Data were thematically analysed.

RESULTS

Five key informants from five emergency departments across two Australian jurisdictions (Victoria and New South Wales) reported that the role was introduced to reduce waiting times and improve quality and safety of care in the ED waiting room. Critical to introducing the role was defining and supporting the scope of practice, experience and preparation of the nurses. Role implementation required champions to overcome identified challenges, including funding. There has been limited evaluation of the role.

CONCLUSIONS

The waiting room nurse role was introduced to decrease waiting times and contributed to risk mitigation. Common to all roles was standing orders, while preparation and experience varied. Further research into the role is required.

Improved quality and efficiency after the introduction of physician-led team triage in an emergency department

BACKGROUND

Overcrowding in the emergency department (ED) may negatively affect patient outcomes, so different triage models have been introduced to improve performance. Physician-led team triage obtains better results than other triage models. We compared efficiency and quality measures before and after reorganization of the triage model in the ED at our county hospital.

MATERIALS AND METHODS

We retrospectively compared two study periods with different triage models: nurse triage in 2008 (baseline) and physician-led team triage in 2012 (follow-up). Physician-led team triage was in use during day-time and early evenings on weekdays. Data were collected from electronic medical charts and the National Mortality Register.

RESULTS

We included 20,073 attendances in 2008 and 23,765 in 2012. The time from registration to physician presentation decreased from 80 to 33 min (P < 0.001), and the length of stay decreased from 219 to 185 min (P < 0.001) from 2008 to 2012, respectively. All of the quality variables differed significantly between the two periods, with better results in 2012. The odds ratio for patients who left before being seen or before treatment was completed was 0.62 (95% confidence interval 0.54-0.72). The corresponding result for unscheduled returns was 0.36 (0.32-0.40), and for the mortality rates within 7 and 30 days 0.72 (0.59-0.88) and 0.84 (0.73-0.97), respectively. The admission rate was 37% at baseline and 32% at follow-up (P < 0.001).

CONCLUSION

Physician-led team triage improved the efficiency and quality in EDs.

Development of a Hospital Standardized Mortality Ratio for Emergency Department Care

STUDY OBJECTIVE

Experts have recommended including measures of mortality in emergency department (ED) performance evaluation frameworks. This study aims to develop a hospital standardized mortality ratio (HSMR) for patients admitted to the hospital with conditions for which ED care may reduce mortality (emergency-sensitive conditions).

METHODS

Data were extracted from Canadian hospital discharge databases from April 1, 2009, to March 31, 2012. The ED HSMR is the ratio of the observed deaths among patients with emergency-sensitive conditions in a hospital during a year to the expected deaths for the same patients during the reference year (2009-2010). The expected deaths were estimated with predictive models fitted for different hospital peer groups (teaching hospitals and large, medium, and small community hospitals).

RESULTS

The data set included 1,770,809 admissions (9.2% deaths). The ED HSMR was calculated for 47% (294/629) of all Canadian hospitals. The majority of exclusions (98%) were for small community hospitals with fewer than 20 expected deaths. Predictive models had good calibration and discrimination, with areas under the curve ranging from 0.80 to 0.81. In comparisons of 2010-2011 and 2011-2012, the classification of hospitals by ED HSMR quartile was stable, with the majority remaining within the same quartile (43.5%) or moving up or down a single quartile (40.2%). Peer-group-level comparisons between ED HSMR measured at different points after admission (2, 7, and 30 days, and hospital discharge) did not demonstrate any significant differences.

CONCLUSION

The ED HSMR appears to be a reliable measure with high discrimination, calibration, and forecasting properties that can be used to guide assessment of ED performance.

Improved outcomes for emergency department patients whose ambulance off-stretcher time is not delayed

OBJECTIVE

To describe and compare characteristics and outcomes of patients who arrive by ambulance to the ED. We aimed to (i) compare patients with a delayed ambulance offload time (AOT) >30 min with those who were not delayed; and (ii) identify predictors of an ED length of stay (LOS) of >4 h for ambulance-arriving patients.

METHODS

A retrospective, multi-site cohort study was undertaken in Australia using 12 months of linked health data (September 2007-2008). Outcomes of AOT delayed and non-delayed presentations were compared. Logistic regression analysis was undertaken to identify predictors of an ED LOS of >4 h.

RESULTS

Of the 40 783 linked, analysable ambulance presentations, AOT delay of >30 min was experienced by 15%, and 63% had an ED LOS of >4 h. Patients with an AOT <30 min had better outcomes for: time to triage; ambulance time at hospital; time to see healthcare professional; proportion seen within recommended triage time frame; and ED LOS for both admitted and non-admitted patients. In-hospital mortality did not differ. Strong predictors of an ED LOS >4 h included: hospital admission, older age, triage category, and offload delay >30 min.

CONCLUSION

Patients arriving to the ED via ambulance and offloaded within 30 min experience better outcomes than those delayed. Given that offload delay is a modifiable predictor of an ED LOS of >4 h, targeted improvements in the ED arrival process for ambulance patients might be useful.

Expanding emergency department capacity: a multisite study

OBJECTIVES

The aims of the present study were to identify predictors of admission and describe outcomes for patients who arrived via ambulance to three Australian public emergency departments (EDs), before and after the opening of 41 additional ED beds within the area.

METHODS

The present study was a retrospective comparative cohort study using deterministically linked health data collected between 3 September 2006 and 2 September 2008. Data included ambulance offload delay, time to see doctor, ED length of stay (LOS), admission requirement, access block, hospital LOS and in-hospital mortality. Logistic regression analysis was undertaken to identify predictors of hospital admission.

RESULTS

Almost one-third of all 286037 ED presentations were via ambulance (n=79196) and 40.3% required admission. After increasing emergency capacity, the only outcome measure to improve was in-hospital mortality. Ambulance offload delay, time to see doctor, ED LOS, admission requirement, access block and hospital LOS did not improve. Strong predictors of admission before and after increased capacity included age >65 years, Australian Triage Scale (ATS) Category 1-3, diagnoses of circulatory or respiratory conditions and ED LOS >4h. With additional capacity, the odds ratios for these predictors increased for age >65 years and ED LOS >4h, and decreased for ATS category and ED diagnoses.

CONCLUSIONS

Expanding ED capacity from 81 to 122 beds within a health service area impacted favourably on mortality outcomes, but not on time-related service outcomes such as ambulance offload time, time to see doctor and ED LOS. To improve all service outcomes, when altering (increasing or decreasing) ED bed numbers, the whole healthcare system needs to be considered.

Emergency department conditions associated with the number of patients who leave a pediatric emergency department before physician assessment

OBJECTIVES

As emergency department (ED) waiting times and volumes increase, substantial numbers of patients leave without being seen (LWBS) by a physician. The objective of this study was to identify ED conditions reflecting patient input, throughput, and output associated with the number of patients who LWBS in a pediatric setting.

METHODS

This study was a retrospective, descriptive study using data from 1 urban, tertiary care pediatric ED. The study population consisted of all patient visits to the ED from April 2005 to March 2007. Multivariate Poisson regression analyses were used to examine the impact of the timing of patient arrival and ED conditions including patient acuity, volume, and waiting times on the number of patients who LWBS.

RESULTS

During the study period, there were 138,361 patient visits corresponding to 2190 consecutive shifts; 11,055 patients (8%) left without being seen by a physician.In the multivariate analysis, the throughput variables, time from triage to physician assessment (rate ratio, 2.11; 95% confidence interval, 2.01-2.21), and time from registration to triage (rate ratio, 1.55; 95% confidence interval, 1.25-1.90) had the largest association with the number of patients who LWBS.

CONCLUSIONS

In the study ED, throughput variables played a more important role than input or output variables on the number of patients who LWBS. This finding, which contrasts with a work done previously in an ED serving primarily adults, highlights the importance of pediatric specific research on the impacts of increasing ED waiting times and volumes.

Evaluation of emergency department performance - a systematic review on recommended performance and quality-in-care measures

BACKGROUND

Evaluation of emergency department (ED) performance remains a difficult task due to the lack of consensus on performance measures that reflects high quality, efficiency, and sustainability.

AIM

To describe, map, and critically evaluate which performance measures that the published literature regard as being most relevant in assessing overall ED performance.

METHODS

Following the PRISMA guidelines, a systematic literature review of review articles reporting accentuated ED performance measures was conducted in the databases of PubMed, Cochrane Library, and Web of Science. Study eligibility criteria includes: 1) the main purpose was to discuss, analyse, or promote performance measures best reflecting ED performance, 2) the article was a review article, and 3) the article reported macro-level performance measures, thus reflecting an overall departmental performance level.

RESULTS

A number of articles addresses this study's objective (n = 14 of 46 unique hits). Time intervals and patient-related measures were dominant in the identified performance measures in review articles from US, UK, Sweden and Canada. Length of stay (LOS), time between patient arrival to initial clinical assessment, and time between patient arrivals to admission were highlighted by the majority of articles. Concurrently, "patients left without being seen" (LWBS), unplanned re-attendance within a maximum of 72 hours, mortality/morbidity, and number of unintended incidents were the most highlighted performance measures that related directly to the patient. Performance measures related to employees were only stated in two of the 14 included articles.

CONCLUSIONS

A total of 55 ED performance measures were identified. ED time intervals were the most recommended performance measures followed by patient centeredness and safety performance measures. ED employee related performance measures were rarely mentioned in the investigated literature. The study's results allow for advancement towards improved performance measurement and standardised assessment across EDs.

Physician-led team triage based on lean principles may be superior for efficiency and quality? A comparison of three emergency departments with different triage models

Background

The management of emergency departments (EDs) principally involves maintaining effective patient flow and care. Different triage models are used today to achieve these two goals. The aim of this study was to compare the performance of different triage models used in three Swedish EDs. Using efficiency and quality indicators, we compared the following triage models: physician-led team triage, nurse first/emergency physician second, and nurse first/junior physician second.

Methods

All data of patients arriving at the three EDs between 08:00- and 21:00 throughout 2008 were collected and merged into a database. The following efficiency indicators were measured: length of stay (LOS) including time to physician, time from physician to discharge, and 4-hour turnover rate. The following quality indicators were measured: rate of patients left before treatment was completed, unscheduled return within 24 and 72 hours, and mortality rate within 7 and 30 days.

Results

Data from 147,579 patients were analysed. The median length of stay was 158 minutes for physician-led team triage, compared with 243 and 197 minutes for nurse/emergency physician and nurse/junior physician triage, respectively (p < 0.001). The rate of patients left before treatment was completed was 3.1% for physician-led team triage, 5.3% for nurse/emergency physician, and 9.6% for nurse/junior physician triage (p < 0.001). Further, the rates of unscheduled return within 24 hours were significantly lower for physician-led team triage, 1.0%, compared with 2.1%, and 2.5% for nurse/emergency physician, and nurse/junior physician, respectively (p < 0.001). The mortality rate within 7 days was 0.8% for physician-led team triage and 1.0% for the two other triage models (p < 0.001).

Conclusions

Physician-led team triage seemed advantageous, both expressed as efficiency and quality indicators, compared with the two other models.

Reviewing deaths in the emergency department: deaths in the department or deaths within 48 h

OBJECTIVES

To investigate an ED death audit process that included deaths occurring within 48 h of admission in addition to deaths in the ED.

METHODS

The study was a review of a prospective audit process undertaken in routine clinical practice that included auditing deaths in the ED and deaths of admitted patients within 48 h of ED presentation. Data were extracted from the audit database and included demography, clinical information and medical recommendations. The hospital incident investigation and monitoring system (IIMS) was searched for major incident reports involving death. The main outcome was the number of medical record audits from each group reported to the clinical governance unit for review, and whether the 48 h audit identified relevant cases to the ED in addition to those identified in the ED audit alone. Secondary outcomes were the number of audits resulting in other actions: ED policy review, education, case discussion or review with the inpatient team.

RESULTS

Over a 2 year period, 303 deaths were reviewed, including 75 deaths in the ED and 228 deaths within 48 h. The ED auditor recommended no further action in 66/75 (88%) ED deaths and 195/228 (86%) 48 h deaths. A major hospital review was recommended in 4/75 (5%) ED deaths and 11/228 (5%) 48 h deaths, with only 3 and 7 of these, respectively, having been detected by the hospital's IIMS. The audit identified 10 of 13 deaths notified to the IIMS and the remaining 3 did not involve error relevant to the ED. Internal review was recommended in one ED death and eight 48 h deaths.

CONCLUSIONS

The present study demonstrates that auditing both ED deaths and 48 h deaths identifies additional issues relevant to the ED compared with auditing ED deaths alone or relying on standard hospital incident reporting.

Efficiency and economic benefits of a payer-based electronic health record in an emergency department

OBJECTIVES

The objective was to evaluate the use of a payer-based electronic health record (P-EHR), which is a clinical summary of a patient's medical and pharmacy claims history, in an emergency department (ED) on length of stay (LOS) and plan payments.

METHODS

A large urban ED partnered with the dominant health plan in the region and implemented P-EHR technology in September 2005 for widespread use for health plan members presenting to the ED. A retrospective observational study design was used to evaluate this previously implemented P-EHR. Health plan and electronic hospital data were used to identify 2,288 ED encounters. Encounters with P-EHR use (n = 779) were identified between September 1, 2005, and February 17, 2006; encounters from the same health plan (n = 1,509) between November 1, 2004, and March 31, 2005, were compared. Outcomes were ED LOS and plan payment for the ED encounter. Analyses evaluated the effect of using the P-EHR in the ED setting on study outcomes using multivariate regressions and the nonparametric bootstrap.

RESULTS

After covariate adjustment, among visits resulting in discharge (ED-only), P-EHR visits were 19 minutes shorter (95% confidence interval [CI] = 5 to 33 minutes) than non-P-EHR visits. Among visits resulting in hospitalization, the P-EHR was associated with an average 77-minute shorter ED LOS (95% CI = 28 to 126 minutes), compared to non-P-EHR visits. The P-EHR was associated with an average of $1,560 (95% CI = $43 to $2,910) lower total plan expenditures for hospitalized visits. No significant difference in total payments was observed among discharged visits.

CONCLUSIONS

In the study ED, the P-EHR was associated with a significant reduction in ED LOS overall and was associated with lower plan payments for visits that resulted in hospitalization.