Fewer emergency department alarms is associated with reduced use of medications for acute agitation.

BACKGROUND AND OBJECTIVES: Patient monitoring systems provide critical information but often produce loud, frequent alarms that worsen patient agitation and stress. This may increase the use of physical and chemical restraints with implications for patient morbidity and autonomy. This study analyzes how augmenting alarm thresholds affects the proportion of alarm-free time and the frequency of medications administered to treat acute agitation. METHODS: Our emergency department's patient monitoring system was modified on June 28, 2022 to increase the tachycardia alarm threshold from 130 to 150 and to remove alarm sounds for several arrhythmias, including bigeminy and premature ventricular beats. A pre-post study was performed lasting 55 days before and 55 days after this intervention. The primary outcome was change in number of daily patient alarms. The secondary outcomes were alarm-free time per day and median number of antipsychotic and benzodiazepine medications administered per day. The safety outcome was the median number of patients transferred daily to the resuscitation area. We used quantile regression to compare outcomes between the pre- and post-intervention period and linear regression to correlate alarm-free time with the number of sedating medications administered. RESULTS: Between the pre- and post-intervention period, the median number of alarms per day decreased from 1332 to 845 (-37%). This was primarily driven by reduced low-priority arrhythmia alarms from 262 to 21 (-92%), while the median daily census was unchanged (33 vs 32). Median hours per day free from alarms increased from 1.0 to 2.4 (difference 1.4, 95% CI 0.8-2.1). The median number of sedating medications administered per day decreased from 14 to 10 (difference - 4, 95% CI -1 to -7) while the number of escalations in level of care to our resuscitation care area did not change significantly. Multivariable linear regression showed a 60-min increase of alarm-free time per day was associated with 0.8 (95% CI 0.1-1.4) fewer administrations of sedating medication while an additional patient on the behavioral health census was associated with 0.5 (95% CI 0.0-1.1) more administrations of sedating medication. CONCLUSION: A reasonable change in alarm parameter settings may increase the time patients and healthcare workers spend in the emergency department without alarm noise, which in this study was associated with fewer doses of sedating medications administered.

Multisite development and validation of machine learning models to predict severe outcomes and guide decision-making for emergency department patients with influenza.

Objective: Millions of Americans are infected by influenza annually. A minority seek care in the emergency department (ED) and, of those, only a limited number experience severe disease or death. ED clinicians must distinguish those at risk for deterioration from those who can be safely discharged. Methods: We developed random forest machine learning (ML) models to estimate needs for critical care within 24 h and inpatient care within 72 h in ED patients with influenza. Predictor data were limited to those recorded prior to ED disposition decision: demographics, ED complaint, medical problems, vital signs, supplemental oxygen use, and laboratory results. Our study population was comprised of adults diagnosed with influenza at one of five EDs in our university health system between January 1, 2017 and May 18, 2022; visits were divided into two cohorts to facilitate model development and validation. Prediction performance was assessed by the area under the receiver operating characteristic curve (AUC) and the Brier score. Results: Among 8032 patients with laboratory-confirmed influenza, incidence of critical care needs was 6.3% and incidence of inpatient care needs was 19.6%. The most common reasons for ED visit were symptoms of respiratory tract infection, fever, and shortness of breath. Model AUCs were 0.89 (95% CI 0.86-0.93) for prediction of critical care and 0.90 (95% CI 0.88-0.93) for inpatient care needs; Brier scores were 0.026 and 0.042, respectively. Importantpredictors included shortness of breath, increasing respiratory rate, and a high number of comorbid diseases. Conclusions: ML methods can be used to accurately predict clinical deterioration in ED patients with influenza and have potential to support ED disposition decision-making.

Effect of Automated Real-Time Feedback on Early-Sepsis Care: A Pragmatic Clinical Trial.

OBJECTIVES: To determine if a real-time monitoring system with automated clinician alerts improves 3-hour sepsis bundle adherence. DESIGN: Prospective, pragmatic clinical trial. Allocation alternated every 7 days. SETTING: Quaternary hospital from December 1, 2020 to November 30, 2021. PATIENTS: Adult emergency department or inpatients meeting objective sepsis criteria triggered an electronic medical record (EMR)-embedded best practice advisory. Enrollment occurred when clinicians acknowledged the advisory indicating they felt sepsis was likely. INTERVENTION: Real-time automated EMR monitoring identified suspected sepsis patients with incomplete bundle measures within 1-hour of completion deadlines and generated reminder pages. Clinicians responsible for intervention group patients received reminder pages; no pages were sent for controls. The primary analysis cohort was the subset of enrolled patients at risk of bundle nonadherent care that had reminder pages generated. MEASUREMENTS AND MAIN RESULTS: The primary outcome was orders for all 3-hour bundle elements within guideline time limits. Secondary outcomes included guideline-adherent delivery of all 3-hour bundle elements, 28-day mortality, antibiotic discontinuation within 48-hours, and pathogen recovery from any culture within 7 days of time-zero. Among 3,269 enrolled patients, 1,377 had reminder pages generated and were included in the primary analysis. There were 670 (48.7%) at-risk patients randomized to paging alerts and 707 (51.3%) to control. Bundle-adherent orders were placed for 198 intervention patients (29.6%) versus 149 (21.1%) controls (difference: 8.5%; 95% CI, 3.9-13.1%; p = 0.0003). Bundle-adherent care was delivered for 152 (22.7%) intervention versus 121 (17.1%) control patients (difference: 5.6%; 95% CI, 1.4-9.8%; p = 0.0095). Mortality was similar between groups (8.4% vs 8.3%), as were early antibiotic discontinuation (35.1% vs 33.4%) and pan-culture negativity (69.0% vs 68.2%). CONCLUSIONS: Real-time monitoring and paging alerts significantly increased orders for and delivery of guideline-adherent care for suspected sepsis patients at risk of 3-hour bundle nonadherence. The trial was underpowered to determine whether adherence affected mortality. Despite enrolling patients with clinically suspected sepsis, early antibiotic discontinuation and pan-culture negativity were common, highlighting challenges in identifying appropriate patients for sepsis bundle application.

Evaluation and implementation of a Just-In-Time bed-assignment strategy to reduce wait times for surgical inpatients.

Early bed assignments of elective surgical patients can be a useful planning tool for hospital staff; they provide certainty in patient placement and allow nursing staff to prepare for patients' arrivals to the unit. However, given the variability in the surgical schedule, they can also result in timing mismatches-beds remain empty while their assigned patients are still in surgery, while other ready-to-move patients are waiting for their beds to become available. In this study, we used data from four surgical units in a large academic medical center to build a discrete-event simulation with which we show how a Just-In-Time (JIT) bed assignment, in which ready-to-move patients are assigned to ready-beds, would decrease bed idle time and increase access to general care beds for all surgical patients. Additionally, our simulation demonstrates the potential synergistic effects of combining the JIT assignment policy with a strategy that co-locates short-stay surgical patients out of inpatient beds, increasing the bed supply. The simulation results motivated hospital leadership to implement both strategies across these four surgical inpatient units in early 2017. In the several months post-implementation, the average patient wait time decreased 25.0% overall, driven by decreases of 32.9% for ED-to-floor transfers (from 3.66 to 2.45 hours on average) and 37.4% for PACU-to-floor transfers (from 2.36 to 1.48 hours), the two major sources of admissions to the surgical floors, without adding additional capacity.

Forecasting hospital-level COVID-19 admissions using real-time mobility data.

BACKGROUND: For each of the COVID-19 pandemic waves, hospitals have had to plan for deploying surge capacity and resources to manage large but transient increases in COVID-19 admissions. While a lot of effort has gone into predicting regional trends in COVID-19 cases and hospitalizations, there are far fewer successful tools for creating accurate hospital-level forecasts. METHODS: Large-scale, anonymized mobile phone data has been shown to correlate with regional case counts during the first two waves of the pandemic (spring 2020, and fall/winter 2021). Building off this success, we developed a multi-step, recursive forecasting model to predict individual hospital admissions; this model incorporates the following data: (i) hospital-level COVID-19 admissions, (ii) statewide test positivity data, and (iii) aggregate measures of large-scale human mobility, contact patterns, and commuting volume. RESULTS: Incorporating large-scale, aggregate mobility data as exogenous variables in prediction models allows us to make hospital-specific COVID-19 admission forecasts 21 days ahead. We show this through highly accurate predictions of hospital admissions for five hospitals in Massachusetts during the first year of the COVID-19 pandemic. CONCLUSIONS: The high predictive capability of the model was achieved by combining anonymized, aggregated mobile device data about users' contact patterns, commuting volume, and mobility range with COVID hospitalizations and test-positivity data. Mobility-informed forecasting models can increase the lead-time of accurate predictions for individual hospitals, giving managers valuable time to strategize how best to allocate resources to manage forthcoming surges.

During the COVID-19 pandemic, hospitals have needed to make challenging decisions around staffing and preparedness based on estimates of the number of admissions multiple weeks ahead. Forecasting techniques using methods from machine learning have been successfully applied to predict hospital admissions statewide, but the ability to accurately predict individual hospital admissions has proved elusive. Here, we incorporate details of the movement of people obtained from mobile phone data into a model that makes accurate predictions of the number of people who will be hospitalized 21 days ahead. This model will be useful for administrators and healthcare workers to plan staffing and discharge of patients to ensure adequate capacity to deal with forthcoming hospital admissions.

Impact of a hospital policy to redistribute admission flow across clinical services for capacity relief during COVID-19 surges.

BACKGROUND: Increased hospital admissions due to COVID-19 place a disproportionate strain on inpatient general medicine service (GMS) capacity compared to other services. OBJECTIVE: To study the impact on capacity and safety of a hospital-wide policy to redistribute admissions from GMS to non-GMS based on admitting diagnosis during surge periods. DESIGN, SETTING, AND PARTICIPANTS: Retrospective case-controlled study at a large teaching hospital. The intervention included adult patients admitted to general care wards during two surge periods (January-February 2021 and 2022) whose admission diagnosis was impacted by the policy. The control cohort included admissions during a matched number of days preceding the intervention. MAIN OUTCOMES AND MEASURES: Capacity measures included average daily admissions and hospital census occupied on GMS. Safety measures included length of stay (LOS) and adverse outcomes (death, rapid response, floor-to-intensive care unit transfer, and 30-day readmission). RESULTS: In the control cohort, there were 365 encounters with 299 (81.9%) GMS admissions and 66 (18.1%) non-GMS versus the intervention with 384 encounters, including 94 (24.5%) GMS admissions and 290 (75.5%) non-GMS (p < .001). The average GMS census decreased from 17.9 and 21.5 during control periods to 5.5 and 8.5 during intervention periods. An interrupted time series analysis confirmed a decrease in GMS daily admissions (p < .001) and average daily hospital census (p = .014; p < .001). There were no significant differences in LOS (5.9 vs. 5.9 days, p = .059) or adverse outcomes (53, 14.5% vs. 63, 16.4%; p = .482). CONCLUSION: Admission redistribution based on diagnosis is a safe lever to reduce capacity strain on GMS during COVID-19 surges.

The Power of Modeling in Emergency Preparedness for COVID-19: A Moonshot Moment for Hospitals.

Before coronavirus disease 2019 (COVID-19), few hospitals had fully tested emergency surge plans. Uncertainty in the timing and degree of surge complicates planning efforts, putting hospitals at risk of being overwhelmed. Many lack access to hospital-specific, data-driven projections of future patient demand to guide operational planning. Our hospital experienced one of the largest surges in New England. We developed statistical models to project hospitalizations during the first wave of the pandemic. We describe how we used these models to meet key planning objectives. To build the models successfully, we emphasize the criticality of having a team that combines data scientists with frontline operational and clinical leadership. While modeling was a cornerstone of our response, models currently available to most hospitals are built outside of their institution and are difficult to translate to their environment for operational planning. Creating data-driven, hospital-specific, and operationally relevant surge targets and activation triggers should be a major objective of all health systems.

Impact of Emergency Department Crowding on Discharged Patient Experience.

INTRODUCTION: While emergency department (ED) crowding has deleterious effects on patient care outcomes and operational efficiency, impacts on the experience for patients discharged from the ED are unknown. We aimed to study how patient-reported experience is affected by ED crowding to characterize which factors most impact discharged patient experience. METHODS: This institutional review board-exempt, retrospective, cohort study included all discharged adult ED patients July 1, 2020-June 30, 2021 with at least some response data to the the National Research Corporation Health survey, sent to most patients discharged from our large, academic medical center ED. Our query yielded 9,401 unique encounters for 9,221 patients. Based on responses to the summary question of whether the patient was likely to recommend our ED, patients were categorized as "detractors" (scores 0-6) or "non-detractors" (scores 7-10). We assessed the relationship between census and patient experience by 1) computing percentage of detractors within each care area and assessing for differences in census and boarder burden between detractors and non-detractors, and 2) multivariable logistic regression assessing the relationship between likelihood of being a detractor in terms of the ED census and the patient's last ED care area. A second logistic regression controlled for additional patient- and encounter-specific covariates. RESULTS: Survey response rate was 24.8%. Overall, 13.9% of responders were detractors. There was a significant difference in the average overall ED census for detractors (average 3.70 more patients physically present at the time of arrival, 95% CI 2.33-5.07). In unadjusted multivariable analyses, three lower acuity ED care areas showed statistically significant differences of detractor likelihood with changes in patient census. The overall area under the curve (AUC) for the unadjusted model was 0.594 (CI 0.577-0.610). The adjusted model had higher AUC (0.673, CI 0.657-.690]; P<0.001), with the same three care areas having significant differences in detractor likelihood based on patient census changes. Length of stay (OR 1.71, CI 1.50-1.95), leaving against medical advice/without being seen (OR 5.15, CI 3.84-6.89), and the number of ED care areas a patient visited (OR 1.16, CI 1.01-1.33) was associated with an increase in detractor likelihood. CONCLUSION: Patients arriving to a crowded ED and ultimately discharged are more likely to have negative patient experience. Future studies should characterize which variables most impact patient experience of discharged ED patients.

Design and Performance of a COVID-19 Hospital Recovery Model.

To determine the accuracy of a predictive model for inpatient occupancy that was implemented at a large New England hospital to aid hospital recovery planning from the COVID-19 surge. Background: During recovery from COVID surges, hospitals must plan for multiple patient populations vying for inpatient capacity, so that they maintain access for emergency department (ED) patients while enabling time-sensitive scheduled procedures to go forward. To guide pandemic recovery planning, we implemented a model to predict hospital occupancy for COVID and non-COVID patients. Methods: At a quaternary care hospital in New England, we included hospitalizations from March 10 to July 12, 2020 and subdivided them into COVID, non-COVID nonscheduled (NCNS), and non-COVID scheduled operating room (OR) hospitalizations. For the recovery period from May 25 to July 12, the model made daily hospital occupancy predictions for each population. The primary outcome was the daily mean absolute percentage error (MAPE) and mean absolute error (MAE) when comparing the predicted versus actual occupancy. Results: There were 444 COVID, 5637 NCNS, and 1218 non-COVID scheduled OR hospitalizations during the recovery period. For all populations, the MAPE and MAE for total occupancy were 2.8% or 22.3 hospitalizations per day; for general care, 2.6% or 17.8 hospitalizations per day; and for intensive care unit, 9.7% or 11.0 hospitalizations per day. Conclusions: The model was accurate in predicting hospital occupancy during the recovery period. Such models may aid hospital recovery planning so that enough capacity is maintained to care for ED hospitalizations while ensuring scheduled procedures can efficiently return.

Development and Validation of a Machine Learning Model to Aid Discharge Processes for Inpatient Surgical Care.

Importance: Inpatient overcrowding is associated with delays in care, including the deferral of surgical care until beds are available to accommodate postoperative patients. Timely patient discharge is critical to address inpatient overcrowding and requires coordination among surgeons, nurses, case managers, and others. This is difficult to achieve without early identification and systemwide transparency of discharge candidates and their respective barriers to discharge. Objective: To validate the performance of a clinically interpretable feedforward neural network model that could improve the discharge process by predicting which patients would be discharged within 24 hours and their clinical and nonclinical barriers. Design, Setting, and Participants: This prognostic study included adult patients discharged from inpatient surgical care from May 1, 2016, to August 31, 2017, at a quaternary care teaching hospital. Model performance was assessed with standard cross-validation techniques. The model's performance was compared with a baseline model using historical procedure median length of stay to predict discharges. In prospective cohort analysis, the feedforward neural network model was used to make predictions on general surgical care floors with 63 beds. If patients were not discharged when predicted, the causes of delay were recorded. Main Outcomes and Measures: The primary outcome was the out-of-sample area under the receiver operating characteristic curve of the model. Secondary outcomes included the causes of discharge delay and the number of avoidable bed-days. Results: The model was trained on 15 201 patients (median [interquartile range] age, 60 [46-70] years; 7623 [50.1%] men) discharged from inpatient surgical care. The estimated out-of-sample area under the receiver operating characteristic curve of the model was 0.840 (SD, 0.008; 95% CI, 0.839-0.844). Compared with the baseline model, the neural network model had higher sensitivity (52.5% vs 56.6%) and specificity (51.7% vs 82.6%). The neural network model identified 65 barriers to discharge. In the prospective study of 605 patients, causes of delays included clinical barriers (41 patients [30.1%]), variation in clinical practice (30 patients [22.1%]), and nonclinical reasons (65 patients [47.8%]). Summing patients who were not discharged owing to variation in clinical practice and nonclinical reasons, 128 bed-days, or 1.2 beds per day, were classified as avoidable. Conclusions and Relevance: This cohort study found that a neural network model could predict daily inpatient surgical care discharges and their barriers. The model identified systemic causes of discharge delays. Such models should be studied for their ability to increase the timeliness of discharges.

Implementation of the Vocera Communication System in a Quaternary Perioperative Environment.

In the hospital, fast and efficient communication among clinicians and other employees is paramount to ensure optimal patient care, workflow efficiency, patient safety and patient comfort. The implementation of the wireless Vocera® Badge, a hands-free wearable device distributed to perioperative team members, has increased communication efficiency across the perioperative environment at Massachusetts General Hospital (MGH). This quality improvement project, based upon identical pre- and post-implementation surveys, used qualitative and quantitative analysis to determine if and how the Vocera system affected the timeliness of information flow, ease of communication, and operating room noise levels throughout the perioperative environment. Overall, the system increased the speed of information flow and eased communication between coworkers yet was perceived to have raised the overall noise level in and around the operating rooms (ORs). The perceived increase in noise was outweighed by the closed-loop communication between clinicians. Further education of the system's features in regard to speech recognition and privacy along with expected conversation protocol are necessary to ensure hassle-free communication for all staff.