Utilizing Computer Modeling To Identify Bottlenecks To Patient Flow And Streamline Care In A Psychiatric Emergency Room

This computer simulation modeled obstacles to patient flow in a psychiatric emergency service (PES) by varying the number of providers, number of independent (MD, APRN) vs. supervised providers (LCSW, medical residents), shift provider added, time to hospitalization and available beds to identify bottlenecks in this system. A computer simulation using JaamSim software predicted changes to patient flow by comparing evaluation and documentation time for licensed independent providers (LIPs) vs. those requiring supervision (SPs), time of day provider added, number of ED beds and wait for hospitalization. Cost analysis was performed for the factor that most improved the patient flow metrics. The largest improvements in patient flow occurred with addition of a LIP from 4pm-12am with reductions in time to bed, time to provider and length of stay by 82%, 68% and 31%, respectively. This compared to reductions by 50%, 40% and 16% with addition of an SP during the same shift. Addition of beds and decreasing the time to hospitalization achieved only modest improvements. However, once an independent provider was added to the evening shift, time to hospitalization was shown to be the most important factor in further lowering the patient length of stay. Cost analysis of adding either an attending M.D., advanced nurse practitioner (APRN), social worker (LCSW) or medical resident showed that the addition of an APRN to the evening shift provided the most cost-efficient improvement in patient flow as it costs 42%, 31% and 26% less than adding an M.D., LCSW or psychiatry resident, respectively. In this particular system, the simulation identified available providers as the primary bottleneck and time to hospitalization as the secondary bottleneck to delivery of care. Addition of personnel during the evening shift showed the greatest improvement in flow metrics. Modeling the decreased efficiency by M.D.s with supervising responsibilities allowed us to calculate the relative cost needed to reduce patient length of stay by 1 hour based on the type of provider added. Overall, the simulation’s ability to simulate changes to one parameter at a time allows bespoke analysis in a variety of clinical setting