The split-flow model has emerged as one of the most efficient and cost-effective strategies for improving emergency department patient flow, yielding thousands or even millions of dollars in annual financial improvement when deployed correctly.
At a Glance
Split-flow models used in conjunction with demand-to-capacity staffing and Lean strategies can resuscitate emergency department (ED) metrics by:
- Expediting front-end triage and intake processes
- Segregating patient flow into two categories, with different care processes for each
- Establishing a process of resource pooling in which physicians and nurses conduct joint assessments
- Opening ED beds through the strategic use of results-pending waiting areas
In competitive healthcare markets, patients often will use the nearest hospital if it is within a 10-minute drive-time radius of home. If the nearest hospital is not so close, however, distance becomes a less significant factor in driving patient choice. To capture these patients, hospitals must distinguish themselves by providing superior customer service, a center of excellence for specialty care, or both.
Because superior customer service demands expedited care, emergency departments (EDs) should focus on having patients be seen by a physician as soon as possible. But experience has shown that faster "door-to-doc" times alone do not improve patient satisfaction, as evidenced by experience with the physician-in-triage model. In this model, a triage-based physician (that is, a physician who does not discharge patients) performs a focused exam to expedite the ordering of diagnostic studies on patients arriving in the ED. Patients then return to the waiting room until called back to the main ED, where a second physician completes a more comprehensive exam.
The physician-in-triage model reduces the door-to-doc time and gives purpose to the patient's wait for a bed, as the patient is concurrently waiting for test results. However, this model often does not decrease overall length of stay (LOS); in some instances, LOS can actually increase due to rework caused by variation in ordering practices between the two physicians involved and the need for additional diagnostic studies uncovered by the second provider. Also, many patients complain when subjected to two exams during the same visit. Furthermore, when arrivals exceed the capacity of the triage physician, patients begin queuing in the waiting room before seeing the triage physician. Thus, by addressing inflow without concurrently addressing the entire throughput process, this model creates a "hurry up and wait" perception of the hospital's ED service.
To drive patient satisfaction and create perceptions of a "no wait" ED, hospitals should go beyond inflow design (the focus of the physician-in-triage model) to redesign the patient throughput process. Hospitals should seek a model that is revenue neutral, at a minimum. To use nursing and physician time most efficiently, redesigns should be Lean-based-eliminating unnecessary processing, motion, searching, and waiting-and should use demand-to-capacity staffing. The goal is to achieve a balance between provider staffing and capacity, at any given moment, to meet patient demand, avoid queuing, and prevent the overstaffing that results in idle providers-all objectives that may be accomplished using a split-flow model.
Overview of the Split-Flow Model
In the split-flow model, arriving patients are greeted by a "pivot" or "quick-look" nurse who uses an emergency severity index (ESI) score and segmentation protocols to separate patients into two groups: sicker (emergent care) and less sick (urgent care). Patients who are deemed sicker bypass triage and go immediately to the main ED for care. Patients who are clearly less sick, by ESI class or protocol, also bypass triage and go immediately to the split-flow intake area where they are examined promptly, as shown in the exhibit below.
Only patients in the "unsure" category-that is, those who were not assigned to the sicker or less-sick groups by the pivot nurse-are triaged. The resulting reduction in triage workload in combination with the implementation of "mini triage"-in which the triage nurse obtains only the patient's chief complaint, vital signs, ESI classification, allergies, and pain score-markedly reduces triage workload, enabling some larger EDs to change triage staffing patterns.
Patients directed to the main ED generally undergo traditional processing. In contrast, patients directed to the split-flow intake area are processed simultaneously by a three-person team consisting of a nurse, a physician or midlevel provider, and a patient care technician, a method known as server pooling in queuing theory vernacular. These joint clinical assessments help prevent inefficient patient over-processing.
After the intake exam, one of four things can happen:
- The patient is immediately discharged.
- The patient is sent to a waiting area to await diagnostic study results, with beds provided only for patients who require them.
- The patient is directed to a treatment room for a procedure or therapy requiring a bed.
- The patient is deemed too sick for the split-flow area and is transferred to the main ED for care.
Those who wish to deploy a split-flow design should understand demand-to-capacity staffing, which is instrumental in calculating many factors associated with the split-flow model, including the optimal number of intake beds, results- waiting chairs, treatment beds, nurses, and physicians and midlevel providers. Using demand-to-capacity staffing also drives decisions about when to deploy scribes, what type of split-flow model to deploy, and when to transition from one split-flow model to another. It also can help reveal why a split-flow model might be failing.
Demand-to-capacity analysis begins by identifying each queuing interface patients will likely experience as they receive care. A queuing interface occurs whenever a patient must wait (queue) for a given server-such as a nurse, a physician, or a bed-to provide a specific service. In essence, patients flow from one server to the next, moving only as fast as the slowest, or least capacitated, server. The goal is for each server to have enough capacity to prevent patient queuing without ever becoming idle. An overview of the methodology is provided in the sidebar below.
Because split-flow areas are often bolted onto existing ED footprints, due consideration should be given to applying Lean principles to optimize provider productivity. In split-flow projects, it is helpful to deploy tools that aid in the overall strategic goal of expediting care at minimal cost.
Results of the Split-Flow Design
Split-flow models, used in conjunction with demand-to-capacity staffing and Lean strategies, can be effective in resuscitating ED metrics, including the percentage of patients who leave without treatment as well as LOS and patient volume. All of these improvements can have a positive impact on a hospital's financial results.
For an average-sized community hospital with 36,000 ED visits per year, for example, results such as the following may be anticipated:
- A reduction in LOS from 300 minutes to 180 minutes for all patients seen in the ED, as care processes become more efficient and waiting times fall
- A drop in ambulance diversions by two patients per day, as LOS improves and more beds become available for patients transported by emergency medical services
- A lowering of the left-without-treatment (LWOT) rate from 5 percent to 2 percent, as enhanced efficiency in the ED decreases wait times
- A 1 percent boost in year-over-year patient volume, as word of the hospital's efficiency spreads throughout the community
The hospital's revenue gain from the use of a split-flow design for ED throughput may be calculated by applying a conservative average-revenue-per-visit value of $400 to this example. Specifically, the LWOT reduction contributes $432,000 per year, even before considering the financial impact of those captured patient visits that result in an inpatient admission. Assuming that half of the additional ambulance patients are admitted to the hospital, the net revenue gain is more than $2.7 million, using a conservative rate of $7,500 per admission. The impact of a 1 percent growth in ED volume yields an additional 360 patients per year, resulting in an additional $144,000 to the hospital's bottom line.
Another benefit of the split-flow model comes from "virtual expansion." That is, as patient flow is improved, it frees up capacity to accommodate more patients. For example, a 60-minute reduction in LOS allows for an additional 30 patients per day to be treated without expanding space in the ED. Thus, if 15 additional patients are treated per day, based on a conservative estimate of charges of $400 per visit, an additional $2.1 million in revenue will be generated each year.
Euclid Hospital in Euclid, Ohio, part of the Cleveland Clinic Health System, provides a real-world example of a split-flow success story, having achieved significant improvements after implementing a split-flow design. Prior to implementing the new flow model, the 33,000-visits-per- year ED had a LWOT rate of 4 percent. Four months after adopting a split-flow design, that rate fell to 2.3 percent. If approximately 560 additional patient visits were captured, at $400 average revenue per visit, that improvement would result in an additional $224,000 in revenue per year.
Euclid also reduced its door-to-doc time by 45 percent for all ED patients and 57 percent for split-flow patients. The ED LOS dropped by more than an hour overall and by more than 2.5 hours for split-flow patients. As a result of those improvements, the hospital was able to accommodate more patients, and its patient satisfaction improved from a mean score of 75.3 to 87.8.
A Cost-Effective Model
Split-flow designs are a dramatic departure from the standard ED workflow model. When properly sized, staffed, and operated, they can be a cost-effective solution for improving ED metrics and patient satisfaction, helping to transform average EDs into centers of excellence while improving financial results.
Mark Harris, MD, FACEP, is president, TeamHealth Northwest, Seattle (firstname.lastname@example.org)
Jeff Wood, RN, is vice president, hospital-based services, TeamHealth, Dallas (email@example.com)
Demand-to-Capacity Staffing Methodology
By understanding the number of arrivals by hour, the minutes of workload generated by each arrival, the way in which workload is distributed over the LOS, and the amount of extra capacity required to allow for variations in interarrival and service times, it is possible to predict the number of servers (including physicians, nurses, intake beds, treatment beds, and results-waiting chairs) required for any given split-flow design, ensuring optimal patient flow with minimal overstaffing. The process for calculating server demand comprises five steps:
- Determine the number of patients arriving to a server for each hour of operation.
- Using the server's average productivity (in patients per hour), calculate the average minutes of workload for the server generated by each arriving patient.
- Distribute this workload rationally over the patient's length of stay. Algorithms and software can simplify this process.
- Calculate the minutes of patient workload for each server by hour of operation, thereby ensuring enough server capacity to meet demand throughout the day.
- As an additional precaution, apply queuing theory to add just enough additional server capacity to counter day-to-day variations in patient arrivals and provider service times. Given their complexity, these calculations are best performed using queuing-based software.
Publication Date: Monday, December 03, 2012