Emergency Medicine: An Operations ... - Wiley Online Library

PRESENTATION

Emergency Medicine: An Operations Management View Olan A. Soremekun, MD, MBA, Christian Terwiesch, PhD, and Jesse M. Pines, MD, MBA, MSCE

Abstract Operations management (OM) is the science of understanding and improving business processes. For the emergency department (ED), OM principles can be used to reduce and alleviate the effects of crowding. A fundamental principle of OM is the waiting time formula, which has clear implications in the ED given that waiting time is fundamental to patient-centered emergency care. The waiting time formula consists of the activity time (how long it takes to complete a process), the utilization rate (the proportion of time a particular resource such a staff is working), and two measures of variation: the variation in patient interarrival times and the variation in patient processing times. Understanding the waiting time formula is important because it presents the fundamental parameters that can be managed to reduce waiting times and length of stay. An additional useful OM principle that is applicable to the ED is the efficient frontier. The efficient frontier compares the performance of EDs with respect to two dimensions: responsiveness (i.e., 1 ⁄ wait time) and utilization rates. Some EDs may be ‘‘on the frontier,’’ maximizing their responsiveness at their given utilization rates. However, most EDs likely have opportunities to move toward the frontier. Increasing capacity is a movement along the frontier and to truly move toward the frontier (i.e., improving responsiveness at a fixed capacity), we articulate three possible options: eliminating waste, reducing variability, or increasing flexibility. When conceptualizing ED crowding interventions, these are the major strategies to consider. ACADEMIC EMERGENCY MEDICINE 2011; 18:1262–1268 ª 2011 by the Society for Academic Emergency Medicine

O

perations management (OM) is the science of business operations. OM utilizes key concepts and theories to understand and improve business processes to increase profitability and customer satisfaction. The science of OM has been used for decades to improve the service provided in industries outside of health care, such as the restaurant, hotels, and airline industries. More recently, key OM principles are being applied in health care environments, such as operating rooms and outpatient clinics, to improve care processes.1–4 However, despite its relevance and potential utility to improve patient flow in the emergency

department (ED), OM principles are not well known in the ED administrative or research communities. The purpose of this paper is to present a basic vocabulary of OM and present two specific OM principles that may be useful to the ED administrator and researchers in conceptualizing interventions to reduce ED crowding. The first principle is the waiting time formula, which is a mathematical calculation of expected wait times that allows understanding of the relationship between key variables that drive how long patients wait for care—a key concept in the ED. The second is the efficient frontier, which presents a way to compare EDs on

From the Department of Emergency Medicine (OAS), and the Department of Operations and Information Management, The Wharton School (CT), University of Pennsylvania, Philadelphia, PA; and the Departments of Emergency Medicine and Health Policy, George Washington University (JMP), Washington, DC. This manuscript was a component of the 2011 Academic Emergency Medicine Consensus Conference entitled ‘‘Interventions to Assure Quality in the Crowded Emergency Department (ED)’’ held in Boston, MA. Funding for this conference was made possible (in part) by 1R13HS020139-01 from the Agency for Healthcare Research and Quality (AHRQ). The views expressed in written conference materials or publications and by speakers and moderators do not necessarily reflect the official policies of the Department of Health and Human Services, nor does mention of trade names, commercial practices, or organizations imply endorsement by the U.S. Government. This issue of Academic Emergency Medicine is funded by the Robert Wood Johnson Foundation. The authors have no relevant financial information or potential conflicts of interest to disclose. Supervising Editor: James Miner, MD. Address for correspondence and reprints: Olan A. Soremekun, MD, MBA; e-mail: [email protected].

1262

ISSN 1069-6563 PII ISSN 1069-6563583

ª 2011 by the Society for Academic Emergency Medicine doi: 10.1111/j.1553-2712.2011.01226.x

ACADEMIC EMERGENCY MEDICINE • December 2011, Vol. 18, No. 12

responsiveness (the inverse of waiting time) and utilization rates. Finally, we discuss the three specific levers that ED operations managers can use to improve ED responsiveness without increasing capacity: eliminating waste, reducing variability, or increasing flexibility.

THE BASIC VOCABULARY OF OPERATIONS MANAGEMENT Below are some key terms used in OM: 1. Activity times and capacity: The activity time represents the amount of time a resource spends on the task. Therefore, the capacity of that resource equals one divided by the activity time. For example, assuming an emergency physician (EP) has an activity time of 0.33 hours per patient evaluation, the laboratory has an activity time of 0.2 hours per test, and radiology has an activity time of 1 hour per study. The physician, laboratory, and radiology capacity thus is three, five, and one patient per hour, respectively. 2. Bottleneck and process capacity: In a multistep process, the bottleneck is the process step with the lowest capacity. The bottleneck capacity thus determines the overall capacity of the entire process, as the process can only go as fast as the slowest step. Using a simplified example from above and assuming all patients require physician evaluation (capacity three patients per hour), basic laboratory tests (capacity five tests per hour), and radiologic studies (capacity one study per hour), in this ED, the bottleneck and overall process capacity is one patient per hour. In a complicated service environment such as the ED, identifying bottlenecks is challenging as these bottlenecks shift based on changes in capacity of each process step and ⁄ or patient mix. 3. Flow time, flow rate, and utilization: The flow time is the duration of time for one flow unit to complete a process. The flow rate is the total number of units that flow through the process and is a function of flow time and customer demand. Utilization rate is thus the flow rate divided by the process capacity. From the above example, the flow time for a patient who requires physician evaluation, laboratory tests, and radiology studies, arriving with no wait for any of these components of the care process, equals 1.53 hours (0.33 hours for physician evaluation, 0.2 hours for laboratory tests, and 1 hour for radiologic study). Assuming sequential processing, the total number of patients that can be cared for in a day equals 16 patients. If only 10 patients per day visit the ED (flow rate), the utilization rate equals 10 divided by 16 or 62.5%. 4. Variation (V) and coefficient of variation (CV): Variation in the service process generally consists of two types: interarrival variation (Va) and service time variation (Vp). While the standard deviation of interarrival and service times is a measure of absolute variability, the CV measures variability in relative terms and is defined as standard deviation of interarrival time or service time divided by their respective means.

•

www.aemj.org

1263

OPERATIONS MANAGER’S PERSPECTIVE OF AN ED The ‘‘Ideal’’ Versus the ‘‘Real’’ ED From an operations manager’s perspective, an ‘‘ideal’’ ED is one with fixed patient arrival time intervals and fixed service times that are equal to or less than the arrival time intervals. Figure 1A is an illustration that demonstrates this concept. With patients arriving every 5 minutes and a fixed service time of 4 minutes per patient, the system will be 80% utilized with no wait times. The ‘‘real’’ ED bears little resemblance to an operations manager’s ideal with high variability of both patient arrival intervals and service times. Figure 1B illustrates the variability in patient arrival and service times that occurs in a typical ED. Assuming a simplified single-server model, this variability in patient arrival time and service intervals leads to some patients having to wait to receive service. Variability in arrival rates and service times is a problem and leads to waiting even when an ED has spare capacity in its overall system. In the setting of this variability, one of the goals of the operations manager is to determine the level of capacity needed to ensure that wait times are less than a certain level the majority of the time. For example, the ED operations manager, using the expected waiting time formula described below, may determine to set capacity at a level such that 95% of patients arriving have an expected wait time less than 30 minutes.

Figure 1. (A) The ideal ED: operations manager’s perspective. (B) The real ED: operations manager’s perspective.

1264

Soremekun et al.

PRINCIPLES IN OPERATIONS MANAGEMENT Waiting Time Formula The waiting time is a function of activity time (or capacity), demand, and variability. Mathematically, the expected waiting time can be expressed as the following:

Wait time ¼ Activity time ½Utilization=ð1 UtilizationÞ ½ðCV2a þ CV2p Þ=2 (see the vocabulary section for definitions of CV and other terms). According to this formula, the expected waiting time is determined by three key variables: utilization rate, interarrival variation, and processing time variation. The utilization rate, which is a function of activity time, capacity, and customer demand, disproportionately affects the wait times with exponential increases in wait times as utilization rates approach 100% (Figure 2). This means that as a higher proportion of ED capacity (treatment bays, staff time, laboratory testing, radiology, etc.) is used, the expected wait time increases. Increasing variability also increases expected waiting times and shifts the curve upwards. That is, at any given utilization rate, an increase in variability leads to longer waiting times (Figure 2). CVa and CVp refer to two types of variation, both of which can be managed to reduce the waiting time (see the ‘‘Moving Toward Efficient Frontier’’ section for more on managing variability). In a simplified example of the waiting time formula, a hospital has 100 inpatients beds, 10 to 20 admissions from the ED per day, and 10 to 20 elective admissions per day. The daily demand for beds thus ranges from 20 to 40 beds per day. If there is an average 50% utilization rate of hospital beds, there will almost always be beds available for the ED and elective admissions. However, at an average 80% utilization rate of hospital beds, there will be days when 40 beds are needed, and on

Figure 2. Relationship between wait times, utilization, and variability.

•

OPERATIONS MANAGEMENT VIEW OF EM

these days there will be significant wait times for inpatient beds. This demonstrates the key tradeoffs between utilization rates and waiting times. Running the hospital at higher utilization rates is more profitable, maximizing revenue and reducing cost per patient (fixed costs shared over more patients); however, this strategy is associated with longer waiting times. Efficient Frontier Another concept that is of particular interest to EDs is the efficient frontier. The efficient frontier allows the comparison of multiple EDs on their responsiveness (responsiveness = 1 ⁄ wait time) and utilization rates. In addition, it can be used to understand how interventions such as changing capacity may affect responsiveness. An efficient frontier is empirically determined, with the EDs on the frontier maximizing their responsiveness at given utilization rates. This is illustrated in Figure 3A. EDs A, B, C, and D are on the efficient frontier and maximize their responsiveness at their respective utilization rates and variability in demand. Given

Figure 3. (A) ED efficient frontier. (B) ED efficient frontier: moving along versus toward the frontier.


that the majority of EDs may be on the lower left of the frontier, managers have the opportunity to implement changes to move toward this frontier. If an ED is off the frontier, an opportunity exists to increase utilization as well as responsiveness, even though according to the mathematical wait time formula, which does not account for operational inefficiencies, these two variables are inversely related. MOVING ALONG THE EFFICIENT FRONTIER Moving along the efficient frontier involves increasing capacity without improving efficiency. Adding capacity will reduce utilization rates and increase responsiveness, but should be differentiated from a move towards the efficient frontier. As illustrated in Figure 3B, ED X by adding more capacity moves to ED X+cap and, while by applying concepts described in the next section, moves toward the frontier to ED X+frontier. Building more capacity in the ED requires hiring more people and building more treatment spaces. To maximally increase ED capacity and increase responsiveness, capacity increases should focus at process bottlenecks. For EDs with significant crowding, a key process bottleneck is the boarding of inpatients.5 Therefore, increasing treatment spaces without reducing the boarding of inpatients in the ED should be expected to be minimally effective. This has been demonstrated in simulation models where reduction in boarding had more of an effect on crowding than increasing treatment spaces.6 Empirical evidence also exists demonstrating the limited effect of an increase in treatment spaces on ED crowding.7 This is why interventions such as surgical schedule smoothing and full-capacity protocols may be very effective in improving patient flow, because they reduce boarding and address a key process bottleneck. Placing an additional physician in triage is another common ED intervention that serves to reduce time to test results and ⁄ or increases the functional ED capacity. This intervention essentially turns the waiting room to another diagnostic and treatment area. The physician in triage, by safely arranging the disposition of certain patients from the waiting room, reduces the demand for ED treatment beds and essentially increases capacity. However, this intervention has had mixed results with institutions where the physician in triage focuses on the treatment and disposition of straightforward patients, demonstrating more of an effect on patient flow than those institutions that focus on initiating workups in the waiting room to reduce lead time.8–10 The institutions that use physician triage to increase capacity move along the efficient frontier and will increase responsiveness.10 The institutions that focus on reduction in testing lead time have found limited benefit, as this design fails to increase capacity, and reduction in testing lead time does not address the key process bottlenecks.8,9

•

www.aemj.org

1265

responsiveness. In OM science, there are three ways to move toward the frontier: 1) eliminating waste, 2) reducing variability, and 3) increasing flexibility. Eliminating Waste Identifying and eliminating parts of the process that add no value to the customer will lead to movement toward the efficient frontier. The analysis can be done on the overall care process (e.g., analyzing the system as a patient moves through the care process) or at the level of each resource (e.g., analyzing how key resources such as physicians and nurses utilize their time). This approach of identifying and eliminating waste to increasing capacity and efficiency is referred to as Lean management and has been used in some EDs to improve patient flow.1 A simplified example is illustrated in Figure 4. The EP during a shift spends a significant portion of time on non–value-added activities such as searching for equipment or idling. In addition, she spends a significant portion of time on value-added activities that could be provided by a much cheaper resource. The use of medical scribes is an example of an intervention that addresses this mismatch and helps improve physician productivity.11 Redesigning processes to eliminate these two types of activities will in effect increase the capacity of the EP. However, as described above, if physician capacity is not a critical process bottleneck, increasing capacity may be associated with increasing costs with no significant improvement in patient flow. A second option to eliminate waste is to reduce unnecessary demand. This would mean directing patients to alternative settings outside the ED for care or bypassing the ED and directly admitting patients to the hospital. However, the effectiveness of creating alternatives to ED-based care and how it might reduce demand is unclear. For example, after implementation of universal health insurance in Massachusetts, it was expected that ED visits would decrease; however, the opposite happened, as both the number and the acuity

MOVING TOWARD THE EFFICIENT FRONTIER Moving toward the efficient frontier requires increasing efficiency that leads to higher utilization rates and

Figure 4. Lean management: identifying waste in physician resource.

1266

of visits increased.12 Some interventions to reduce ED demand may be more effective, like increasing patient access to primary care providers and urgent laboratory and radiologic testing. While this may reduce the demand for ED services, this requires significant investments in human resources and infrastructure. Reducing Variability Reducing variation is another option to move toward the efficient frontier. There are two components of variability, service time variability and demand variability, that can be managed. The first option to reduce service time variability is to reduce provider variability. In every ED, there is significant variation in individual provider practice. Recent studies performed across pediatric and adult EDs have demonstrated that provider utilization of certain resources can vary up to eightfold in the workup of some common ED complaints.13,14 Assuming that the quality of care provided is equivalent, understanding and reducing variations in care (decreasing test utilization) will increase efficiency and movement towards the efficient frontier. In addition to provider variation, several other factors, such as presenting complaints, patient acuity, and availability of inpatient beds, contribute to service time variability. The creation of dedicated fast tracks, while increasing demand variability (pooling of patients generally reduce demand variation), significantly reduces service time variability. The reduction in service time variability by creation of a dedicated fast track, in certain EDs, outweighs the reduction in demand variation and has led to significant movement toward the frontier.15,16 The variation in processing time secondary to availability of inpatient beds can also be reduced through interventions such as surgical schedule smoothing and full capacity protocols. Both interventions reduce the variability in processing time of admitted ED patients and decrease the ED utilization rates. Finally, reduction in demand variability can lead to reducing overall variation and movement toward the efficient frontier. There are limited options to reduce demand variability. A common intervention to reduce demand variability is the use of ambulance diversion. This intervention has a very limited effect on reducing variability for a couple of reasons. First, ambulance traffic accounts for a fraction of ED patient arrivals and thus is not an effective way to reduce variation in demand.17 Second, due to the lack of spare capacity in the overall system, an ED going on diversion frequently leads to a domino-like effect, causing crowding and the need for diversion at nearby EDs.18,19 Increasing Flexibility Despite demand variability, some demand is predictable, with up to one-third of ED demand variation explained by the day of week alone.20 There is also a somewhat predictable demand for noncritical inpatient beds per day.20,21 Using historic data to develop predictive models can allow for better matching of capacity with demand. A simple question that a manager may ask would be whether staffing is the same on Sundays and Mondays. If the answer is yes, given that most EDs have significantly higher demand on Mondays, there

Soremekun et al.

•


may be an opportunity for using predictive modeling to appropriately match capacity with demand. Predictive models have also been used to predict patient arrival and develop staggered staffing patterns that better match demand patterns.22 However, even the best predictive model will have some forecasting error. Therefore, in addition to building predictive models, building flexibility will also be required to move toward the efficient frontier. To supplement predictive demand models, increasing flexibility allows for responsiveness to unpredicted demand. Reactive capacity and flow flexibility are two approaches to increase flexibility. Reactive capacity is the ability of the system to rapidly increase capacity in response to a higher than predicted level of demand. Reactive capacity is especially important in the fashion industry where trends and customer demands change rapidly. A specific fashion trend may last as little as 3 months, after which retailers have to significantly mark down whatever is not sold. By producing in proximity to their core markets, as opposed to in Asia like most of their competitors, the retailer Zara has slightly higher costs but more reactive capacity, allowing it to get products on retail shelves sooner and thus selling their product at regular price longer than their less reactive competitors. In the ED, reactive capacity that is on demand can be helpful when demand exceeds current capacity. Full-capacity protocols and canceling elective surgical cases may be effective ways to develop reactive inpatient bed capacity to reduce boarding.23 Also, an on-call system of EPs and nurses may be helpful in building flexibility or possibly opening additional space during high-volume days. A warning, though: if your reactive capacity is always needed, then it is really not reactive and is a sign that you will need more capacity. For example, if an ED has a predictive capacity model that aims to match capacity to demand and additional reactive capacity, but must activate its reactive capacity every Monday, the predictive capacity model should be changed for Monday to increase the baseline capacity for Monday. Designing processes to allow for flow flexibility is another option to increase flexibility. Dynamic routing and chaining are two additional concepts that may be used to build flow flexibility. Both concepts were developed in the automotive industry and basically are process designs that allow multiple plants to manufacture multiple different cars. Therefore, when demand for a specific car increases, there are several options to fulfill the demand; essentially there is always a ‘‘backup’’ to handle unexpected demand. In the ED and hospital, the use of dynamic routing and chaining designs has been limited, but should be explored. Chaining can be employed in bed management where new ED patients can be cared for in ED bays, fast track, observation units, urgent care centers, or primary care physician offices depending on demand. It can also be employed for hospital bed management where inpatient beds can be used by several different services depending on demand. Hospitals whose bed management strategy involves grouping certain types of patients on certain floors can benefit from chaining. While the strategy of grouping patients


together create floors with certain expertise, the lack of a backup location where equally high-quality care can be provided reduces flexibility, prolongs waits for beds, and can affect the quality of care when patients have to go to other floors. The number of backup floors could be determined by balancing the variation in demand with the cost of educating and creating backups. CONCLUSIONS Emergency departments are challenging to manage because of high variability in demand and service times. While maximization of utilization rates and associated high profits are goals for managers, variability in demand and service times can lead to long wait times at high utilization rates. Therefore, managers must find the right balance of utilization, profitability, and wait times. The manager must also seek to move the operations to the efficient frontier using one or all the three options available: eliminating waste, reducing variability, or building flexibility. Using Lean management techniques, wastes in the process can be identified and eliminated. Also, while additional capacity can be built by hiring staff and adding treatment bays, this generally results in movement along the efficient frontier and not toward the frontier. Increasing capacity without improving efficiencies will have limited effect on flow. Capacity increases should also focus on the current process bottlenecks (e.g., the boarding of inpatients) to achieve a maximal increase in overall process capacity. Building predictive models that allow for better matching of capacity with demand and identification of reactive capacity are other important interventions to improve flexibility. Designing dynamic routing systems and chaining may create flexibility and give patients other options for urgent medical evaluation. While the complicated nature of emergency care makes operations management challenging, effective application of these concepts in designing interventions can lead to reduction to wait times and crowding and improvement in overall patient flow. References 1. Holden RJ. Lean thinking in emergency departments: a critical review. Ann Emerg Med. 2011; 57:265–78. 2. Ryckman FC, Yelton PA, Anneken AM, Kiessling PE, Schoettker PJ, Kotagal UR. Redesigning intensive care unit flow using variability management to improve access and safety. Joint Comm J. 2009; 35:535–43. 3. Guerriero F, Guido R. Operational research in the management of the operating theatre: a survey. Health Care Manag Sci. 2011; 14:89–114. 4. Chand S, Moskowitz H, Norris JB, Shade S, Willis DR. Improving patient flow at an outpatient clinic: study of sources of variability and improvement factors. Health Care Manag Sci. 2009; 12:325–40. 5. Hoot NR, Aronsky D. Systematic review of emergency department crowding: causes, effects, and solutions. Ann Emerg Med. 2008; 52:126–36.

•

www.aemj.org

1267

6. Khare RK, Powell ES, Reinhardt G, Lucenti M. Adding more beds to the emergency department or reducing admitted patient boarding times: which has a more significant influence on emergency department congestion? Ann Emerg Med. 2009; 53:575–85. 7. Zhou HJ, France C, Daniel J, et al. The effect of emergency department expansion on emergency department overcrowding. Acad Emerg Med. 2007; 14:338–43. 8. Han JH, France DJ, Levin SR, Jones ID, Storrow AB, Aronsky D. The effect of physician triage on emergency department length of stay. J Emerg Med. 2010; 39:227–33. 9. Holroyd BR, Bullard MJ, Latoszek K, et al. Impact of a triage liaison physician on emergency department overcrowding and throughput: a randomized controlled trial. Acad Emerg Med. 2007; 14:702–8. 10. Soremekun OA, Biddinger PD, White BA, et al. Operational and financial impact of physician screening in the ED. Am J Emerg Med. 2011; in press. 11. Arya R, Salovich DM, Ohman-Strickland P, Merlin MA. Impact of scribes on performance indicators in the emergency department. Acad Emerg Med. 2010; 17:490–4. 12. Smulowitz PB, Lipton R, Wharam JF, et al. Emergency department utilization after the implementation of Massachusetts health reform. Ann Emerg Med. 2011; 58:225–34. 13. Jain S, Elon LK, Johnson BA, Frank G, Deguzman M. Physician practice variation in the pediatric emergency department and its impact on resource use and quality of care. Pediatr Emerg Care. 2010; 26:902–8. 14. Kim AS, Sidney S, Klingman JG, Johnston SC. Practice variation in neuroimaging to evaluate dizziness in the ED. Am J Emerg Med. 2011; in press. 15. Considine J, Kropman M, Kelly E, Winter C. Effect of emergency department fast track on emergency department length of stay: a case-control study. Emerg Med J. 2008; 25:815–9. 16. Sanchez M, Smally AJ, Grant RJ, Jacobs LM. Effects of a fast-track area on emergency department performance. J Emerg Med. 2006; 31:117–20. 17. Niska R, Bhuiya F, Xu J. National Hospital Ambulatory Medical Care Survey: 2007 emergency department summary. Nat Health Stat Rep. 2010; (26):1–31. 18. Sun BC, Mohanty SA, Weiss R, et al. Effects of hospital closures and hospital characteristics on emergency department ambulance diversion, Los Angeles County, 1998 to 2004. Ann Emerg Med. 2006; 47:309–16. 19. Vilke GM, Brown L, Skogland P, Simmons C, Guss DA. Approach to decreasing emergency department ambulance diversion hours. J Emerg Med. 2004; 26:189–92. 20. Wargon M, Guidet B, Hoang TD, Hejblum G. A systematic review of models for forecasting the number of emergency department visits. Emerg Med J. 2009; 26:395–9.

1268

21. Sun Y, Heng BH, Seow YT, Seow E. Forecasting daily attendances at an emergency department to aid resource planning. BMC Emerg Med. 2009; 9:e1. 22. Green LV, Soares J, Giglio JF, Green RA. Using queueing theory to increase the effectiveness of emergency department provider staffing. Acad Emerg Med. 2006; 13:61–8.

Soremekun et al.

•


23. Viccellio AF, Santora C, Singer AJ, Thode HC Jr, Henry MC. The association between transfer of emergency department boarders to inpatient hallways and mortality: a 4-year experience. Ann Emerg Med. 2009; 54:487–91.