Machine-Learning-Based Electronic Triage More Accurately Differentiates Patients With Respect to Clinical Outcomes Compared With the Emergency Severity Index.

STUDY OBJECTIVE: Standards for emergency department (ED) triage in the United States rely heavily on subjective assessment and are limited in their ability to risk-stratify patients. This study seeks to evaluate an electronic triage system (e-triage) based on machine learning that predicts likelihood of acute outcomes enabling improved patient differentiation. METHODS: A multisite, retrospective, cross-sectional study of 172,726 ED visits from urban and community EDs was conducted. E-triage is composed of a random forest model applied to triage data (vital signs, chief complaint, and active medical history) that predicts the need for critical care, an emergency procedure, and inpatient hospitalization in parallel and translates risk to triage level designations. Predicted outcomes and secondary outcomes of elevated troponin and lactate levels were evaluated and compared with the Emergency Severity Index (ESI). RESULTS: E-triage predictions had an area under the curve ranging from 0.73 to 0.92 and demonstrated equivalent or improved identification of clinical patient outcomes compared with ESI at both EDs. E-triage provided rationale for risk-based differentiation of the more than 65% of ED visits triaged to ESI level 3. Matching the ESI patient distribution for comparisons, e-triage identified more than 10% (14,326 patients) of ESI level 3 patients requiring up triage who had substantially increased risk of critical care or emergency procedure (1.7% ESI level 3 versus 6.2% up triaged) and hospitalization (18.9% versus 45.4%) across EDs. CONCLUSION: E-triage more accurately classifies ESI level 3 patients and highlights opportunities to use predictive analytics to support triage decisionmaking. Further prospective validation is needed.

Discrete event simulation for healthcare organizations: a tool for decision making.

Healthcare organizations face challenges in efficiently accommodating increased patient demand with limited resources and capacity. The modern reimbursement environment prioritizes the maximization of operational efficiency and the reduction of unnecessary costs (i.e., waste) while maintaining or improving quality. As healthcare organizations adapt, significant pressures are placed on leaders to make difficult operational and budgetary decisions. In lieu of hard data, decision makers often base these decisions on subjective information. Discrete event simulation (DES), a computerized method of imitating the operation of a real-world system (e.g., healthcare delivery facility) over time, can provide decision makers with an evidence-based tool to develop and objectively vet operational solutions prior to implementation. DES in healthcare commonly focuses on (1) improving patient flow, (2) managing bed capacity, (3) scheduling staff, (4) managing patient admission and scheduling procedures, and (5) using ancillary resources (e.g., labs, pharmacies). This article describes applicable scenarios, outlines DES concepts, and describes the steps required for development. An original DES model developed to examine crowding and patient flow for staffing decision making at an urban academic emergency department serves as a practical example.

Designed to fail: how computer simulation can detect fundamental flaws in clinic flow.

Discrete-event simulation can be used as an effective tool for healthcare administrators to "test" various operational decisions. The recent growth in hospital outpatient volumes and a constrained financial environment make discrete-event simulation a cost-effective way to diagnose inefficiency and create and test strategies for improvement. This study shows how discrete-event simulation was used in an adult medicine clinic within a large, tertiary care, academic medical center. Simulation creation steps are discussed, including information gathering, process mapping, data collection, model creation, and results. Results of the simulation indicated that system bottle-necks were present in the medication administration and check-out steps of the clinic process. The simulation predicted that matching resources to excessive demand at appropriate times for these bottleneck steps would reduce patients' mean time in the system (i.e., visit time) from 124.3 (s.d. +/- 65.7) minutes to 87.0 (s.d. +/- 36.4) minutes. Although other factors may affect real-world operations of a clinic, discrete-event simulation allows healthcare administrators and clinic operational decision makers to observe the effects of changing staffing and resource allocations on patient wait and throughput time. Discrete-event simulation is not a cure-all for clinic throughput problems, but can be a strong tool to provide evidentiary guidance for clinic operational redesign.

Readmission Risk Trajectories for Patients With Heart Failure Using a Dynamic Prediction Approach: Retrospective Study.

BACKGROUND: Patients hospitalized with heart failure suffer the highest rates of 30-day readmission among other clinically defined patient populations in the United States. Investigation into the predictability of 30-day readmissions can lead to clinical decision support tools and targeted interventions that can help care providers to improve individual patient care and reduce readmission risk. OBJECTIVE: This study aimed to develop a dynamic readmission risk prediction model that yields daily predictions for patients hospitalized with heart failure toward identifying risk trajectories over time and identifying clinical predictors associated with different patterns in readmission risk trajectories. METHODS: A two-stage predictive modeling approach combining logistic and beta regression was applied to electronic health record data accumulated daily to predict 30-day readmission for 534 hospital encounters of patients with heart failure over 2750 patient days. Unsupervised clustering was performed on predictions to uncover time-dependent trends in readmission risk over the patient's hospital stay. We used data collected between September 1, 2013, and August 31, 2015, from a community hospital in Maryland (United States) for patients with a primary diagnosis of heart failure. Patients who died during the hospital stay or were transferred to other acute care hospitals or hospice care were excluded. RESULTS: Readmission occurred in 107 (107/534, 20.0%) encounters. The out-of-sample area under curve for the 2-stage predictive model was 0.73 (SD 0.08). Dynamic clinical predictors capturing laboratory results and vital signs had the highest predictive value compared with demographic, administrative, medical, and procedural data included. Unsupervised clustering identified four risk trajectory groups: decreasing risk (131/534, 24.5% encounters), high risk (113/534, 21.2%), moderate risk (177/534, 33.1%), and low risk (113/534, 21.2%). The decreasing risk group demonstrated change in average probability of readmission from admission (0.69) to discharge (0.30), whereas the high risk (0.75), moderate risk (0.61), and low risk (0.39) groups maintained consistency over the hospital course. A higher level of hemoglobin, larger decrease in potassium and diastolic blood pressure from admission to discharge, and smaller number of past hospitalizations are associated with decreasing readmission risk (P<.001). CONCLUSIONS: Dynamically predicting readmission and quantifying trends over patients' hospital stay illuminated differing risk trajectory groups. Identifying risk trajectory patterns and distinguishing predictors may shed new light on indicators of readmission and the isolated effects of the index hospitalization.

Emergency department throughput: an intervention.

Shortening emergency department (ED) boarding time and managing hospital bed capacity by expediting the inpatient discharge process have been challenging for hospitals nationwide. The objective of this study is was to explore the effect of an innovative prospective intervention on hospital workflow, specifically on early inpatient discharges and the ED boarding time. The intervention consisted of a structured nursing "admission discharge transfer" (ADT) protocol receiving new admissions from the ED and helping out floor nursing with early discharges. ADT intervention was implemented in a 38-bed hospitalist run inpatient unit at an academic hospital. The study population consisted of 4486 patients (including inpatient and observation admissions) who were hospitalized to the medicine unit from March 2013-March 2014. Of these hospitalizations, 2259 patients received the ADT intervention. Patients' demographics, discharge and ED boarding data were collected for from March 4, 2013 to March 31, 2014 for both intervention and control groups (28 weeks each). Chi-square and unpaired t tests were utilized to compare population characteristics. Poisson regression analysis was conducted to estimate the association between intervention and hospital length of stay adjusted for differences in patient demographics. Mean age of the study population was 58.6 years, 23% were African Americans and 55% were women. A significant reduction in ED boarding time (p < 0.001) and improvement in early (before 2 PM) hospital discharges (p = 0.01) were noticed among patients in the intervention groups. There was a slight but significant reduction in hospital length of stay for observation patients in the intervention group; however, no such difference was noted for inpatient admissions. Our study showed that dedicating nursing resources towards ED-boarded patients and early inpatient discharges can significantly improve hospital workflow and reduce hospital length of stay.

Real-time prediction of inpatient length of stay for discharge prioritization.

OBJECTIVE: Hospitals are challenged to provide timely patient care while maintaining high resource utilization. This has prompted hospital initiatives to increase patient flow and minimize nonvalue added care time. Real-time demand capacity management (RTDC) is one such initiative whereby clinicians convene each morning to predict patients able to leave the same day and prioritize their remaining tasks for early discharge. Our objective is to automate and improve these discharge predictions by applying supervised machine learning methods to readily available health information. MATERIALS AND METHODS: The authors use supervised machine learning methods to predict patients' likelihood of discharge by 2 p.m. and by midnight each day for an inpatient medical unit. Using data collected over 8000 patient stays and 20 000 patient days, the predictive performance of the model is compared to clinicians using sensitivity, specificity, Youden's Index (i.e., sensitivity + specificity - 1), and aggregate accuracy measures. RESULTS: The model compared to clinician predictions demonstrated significantly higher sensitivity (P < .01), lower specificity (P < .01), and a comparable Youden Index (P > .10). Early discharges were less predictable than midnight discharges. The model was more accurate than clinicians in predicting the total number of daily discharges and capable of ranking patients closest to future discharge. CONCLUSIONS: There is potential to use readily available health information to predict daily patient discharges with accuracies comparable to clinician predictions. This approach may be used to automate and support daily RTDC predictions aimed at improving patient flow.

Medication waste reduction in pediatric pharmacy batch processes.

OBJECTIVES: To inform pediatric cart-fill batch scheduling for reductions in pharmaceutical waste using a case study and simulation analysis. METHODS: A pre and post intervention and simulation analysis was conducted during 3 months at a 205-bed children's center. An algorithm was developed to detect wasted medication based on time-stamped computerized provider order entry information. The algorithm was used to quantify pharmaceutical waste and associated costs for both preintervention (1 batch per day) and postintervention (3 batches per day) schedules. Further, simulation was used to systematically test 108 batch schedules outlining general characteristics that have an impact on the likelihood for waste. RESULTS: Switching from a 1-batch-per-day to a 3-batch-per-day schedule resulted in a 31.3% decrease in pharmaceutical waste (28.7% to 19.7%) and annual cost savings of $183,380. Simulation results demonstrate how increasing batch frequency facilitates a more just-in-time process that reduces waste. The most substantial gains are realized by shifting from a schedule of 1 batch per day to at least 2 batches per day. The simulation exhibits how waste reduction is also achievable by avoiding batch preparation during daily time periods where medication administration or medication discontinuations are frequent. Last, the simulation was used to show how reducing batch preparation time per batch provides some, albeit minimal, opportunity to decrease waste. CONCLUSIONS: The case study and simulation analysis demonstrate characteristics of batch scheduling that may support pediatric pharmacy managers in redesign toward minimizing pharmaceutical waste.