Simulation to Predict Effect of Citywide Events on Emergency Department Operations.

Medical emergency preparedness has been an issue of medical relevance since the advent of hospital care. Studies have simulated emergency department (ED) overcrowding but not yet characterized effects of large-scale, planned events that drastically alter a city's demography, such as in Philadelphia, Pennsylvania during the 2015 World Meeting of Families. A discrete event simulation of the ED at the Children's Hospital of Philadelphia was designed and validated using past data. The model was used to predict the patient length of stay (LOS) and number of admitted patients if the arrival stream to the ED were to change by 50% from typical arrivals in either direction. We compared the model's estimations with data produced during the papal visit that had 39.65% fewer patient arrivals. For validation, the simulated mean LOS was 226.1 ± 173.3 minutes (mean ± SD) for all patients and 352.1 ± 170.3 minutes for admitted patients. Real-world mean LOSs for the fiscal year 2014 were 230.6 ± 134.8 for all patients and 345.0 ± 147.7 for admitted patients. For the estimation of the World Meeting of Families, the simulation accurately estimated the LOS of both patients overall and admitted patients within 10%. These results show that it is possible to use simulations to project the patient flow effects in EDs in case of large-scale events. Providing efficient care is essential to emergency operations, and projections of demand are crucial for targeting appropriate changes during large-scale events. Analysis of validated computer simulations allows for evidence-based decision making in a complex clinical environment.

Reducing postponements of elective pediatric cardiac procedures: analysis and implementation of a discrete event simulation model.

BACKGROUND: This study describes the use of discrete event simulation (DES) to model and analyze a large academic pediatric and test cardiac center. The objective was to identify a strategy, and to predict and test the effectiveness of that strategy, to minimize the number of elective cardiac procedures that are postponed because of a lack of available cardiac intensive care unit (CICU) capacity. METHODS: A DES of the cardiac center at The Children's Hospital of Philadelphia was developed and was validated by use of 1 year of deidentified administrative patient data. The model was then used to analyze strategies for reducing postponements of cases requiring CICU care through improved scheduling of multipurpose space. Each of five alternative scenarios was simulated for ten independent 1-year runs. RESULTS: Reductions in simulated elective procedure postponements were found when a multipurpose procedure room (the hybrid room) was used for operations on Wednesday and Thursday, compared with Friday (as was the real-world use). The reduction Wednesday was statistically significant, with postponements dropping from 27.8 to 23.3 annually (95% confidence interval 18.8-27.8). Thus, we anticipate a relative reduction in postponements of 16.2%. Since the implementation, there have been two postponements from July 1 to November 21, 2014, compared with ten for the same time period in 2013. CONCLUSIONS: Simulation allows us to test planned changes in complex environments, including pediatric cardiac care. Reduction in postponements of cardiac procedures requiring CICU care is predicted through reshuffling schedules of existing multipurpose capacity, and these reductions appear to be achievable in the real world after implementation.

Capacity planning for maternal-fetal medicine using discrete event simulation.

BACKGROUND: Maternal-fetal medicine is a rapidly growing field requiring collaboration from many subspecialties. We provide an evidence-based estimate of capacity needs for our clinic, as well as demonstrate how simulation can aid in capacity planning in similar environments. METHODS: A Discrete Event Simulation of the Center for Fetal Diagnosis and Treatment and Special Delivery Unit at The Children's Hospital of Philadelphia was designed and validated. This model was then used to determine the time until demand overwhelms inpatient bed availability under increasing capacity. FINDINGS: No significant deviation was found between historical inpatient censuses and simulated censuses for the validation phase (p = 0.889). Prospectively increasing capacity was found to delay time to balk (the inability of the center to provide bed space for a patient in need of admission). With current capacity, the model predicts mean time to balk of 276 days. Adding three beds delays mean time to first balk to 762 days; an additional six beds to 1,335 days. CONCLUSION: Providing sufficient access is a patient safety issue, and good planning is crucial for targeting infrastructure investments appropriately. Computer-simulated analysis can provide an evidence base for both medical and administrative decision making in a complex clinical environment.

Incorporating discrete event simulation into quality improvement efforts in health care systems.

Quality improvement (QI) efforts are an indispensable aspect of health care delivery, particularly in an environment of increasing financial and regulatory pressures. The ability to test predictions of proposed changes to flow, policy, staffing, and other process-level changes using discrete event simulation (DES) has shown significant promise and is well reported in the literature. This article describes how to incorporate DES into QI departments and programs in order to support QI efforts, develop high-fidelity simulation models, conduct experiments, make recommendations, and support adoption of results. The authors describe how DES-enabled QI teams can partner with clinical services and administration to plan, conduct, and sustain QI investigations.

Addressing the variation of post-surgical inpatient census with computer simulation.

OBJECTIVE: This study describes the development of a Discrete Event Simulation (DES) of a large pediatric perioperative department, and its use to compare the effectiveness of increasing the number of post-surgical inpatient beds vs. implementing a new discharge strategy on the proportion of patients admitted to the surgical unit to recover. MATERIALS AND METHODS: A DES of the system was developed and simulated data were compared with 1 year of inpatient data to establish baseline validity. Ten years of simulated data generated by the baseline simulation (control) was compared to 10 years of simulated data generated by the simulation for the experimental scenarios. Outcome and validation measures include percentage of patients recovering in post-surgical beds vs. "off floor" in medical beds, and daily census of inpatient volumes. RESULTS: The proportion of patients admitted to the surgical inpatient unit rose from 79.0% (95% CI, 77.9-80.1%) to 89.4% (95% CI, 88.7-90.0%) in the discharge strategy scenario, and to 94.2% (95% CI, 93.5-95.0%) in the additional bed scenario. The daily mean number of patients admitted to medical beds fell from 9.3 ± 5.9 (mean ± SD) to 4.9 ± 4.5 in the discharge scenario, and to 2.4 ± 3.2 in the additional bed scenario. DISCUSSION: Every hospital is tasked with placing the right patient in the right bed at the right time. Appropriately validated DES models can provide important insight into system dynamics. No significant variation was found between the baseline simulation and real-world data. This allows us to draw conclusions about the ramifications of changes to system capacity or discharge policy, thus meeting desired system performance measures.

An Agent-Based Modeling Template for a Cohort of Veterans with Diabetic Retinopathy.

BACKGROUND: Agent-based models are valuable for examining systems where large numbers of discrete individuals interact with each other, or with some environment. Diabetic Veterans seeking eye care at a Veterans Administration hospital represent one such cohort. OBJECTIVE: The objective of this study was to develop an agent-based template to be used as a model for a patient with diabetic retinopathy (DR). This template may be replicated arbitrarily many times in order to generate a large cohort which is representative of a real-world population, upon which in-silico experimentation may be conducted. METHODS: Agent-based template development was performed in java-based computer simulation suite AnyLogic Professional 6.6. The model was informed by medical data abstracted from 535 patient records representing a retrospective cohort of current patients of the VA St. Louis Healthcare System Eye clinic. Logistic regression was performed to determine the predictors associated with advancing stages of DR. Predicted probabilities obtained from logistic regression were used to generate the stage of DR in the simulated cohort. RESULTS: The simulated cohort of DR patients exhibited no significant deviation from the test population of real-world patients in proportion of stage of DR, duration of diabetes mellitus (DM), or the other abstracted predictors. Simulated patients after 10 years were significantly more likely to exhibit proliferative DR (P<0.001). CONCLUSIONS: Agent-based modeling is an emerging platform, capable of simulating large cohorts of individuals based on manageable data abstraction efforts. The modeling method described may be useful in simulating many different conditions where course of disease is described in categorical stages.

Decreased length of stay after addition of healthcare provider in emergency department triage: a comparison between computer-simulated and real-world interventions.

OBJECTIVE: (1) To determine the effects of adding a provider in triage on average length of stay (LOS) and proportion of patients with >6 h LOS. (2) To assess the accuracy of computer simulation in predicting the magnitude of such effects on these metrics. METHODS: A group-level quasi-experimental trial comparing the St. Louis Veterans Affairs Medical Center emergency department (1) before intervention, (2) after institution of provider in triage, and discrete event simulation (DES) models of similar (3) 'before' and (4) 'after' conditions. The outcome measures were daily mean LOS and percentage of patients with LOS >6 h. RESULTS: The DES-modelled intervention predicted a decrease in the %6-hour LOS from 19.0% to 13.1%, and a drop in the daily mean LOS from 249 to 200 min (p<0.0001). Following (actual) intervention, the number of patients with LOS >6 h decreased from 19.9% to 14.3% (p<0.0001), with the daily mean LOS decreasing from 247 to 210 min (p<0.0001). CONCLUSION: Physician and mid-level provider coverage at triage significantly reduced emergency department LOS in this setting. DES accurately predicted the magnitude of this effect. These results suggest further work in the generalisability of triage providers and in the utility of DES for predicting quantitative effects of process changes.