WHITE PAPER Key Strategies for Improving Hospital Flow BY: Kirk B. Jensen, MD, MBA, FACEP Content The ED is a part of a hospital-wide system of patient flow Every system produces the results it is designed to produce Key strategic concepts to improve patient flow You cannot optimize a system by optimizing just one part of the system
CONTENTS
KEY STRATEGIES FOR IMPROVING HOSPITAL FLOW Introduction... 2 Every System Is Designed to Produce the Results It Produces... 3 Demand-Capacity Management... 3 Real-Time Monitoring of Patient Flow... 4 Forecasting... 4 Queuing Theory... 5 The Theory of Constraints... 6 Managing Variation... 7 The Appreciation of a System... 8 In Summary... 8 About the Author... 9 Contact... 10
INTRODUCTION We are all engaged in a hospital-wide a system of patient flow or patient care. We are each part of the whole. The emergency department is connected to the ICU. The ICU is connected to the OR. The discharge and discharge processes are connected to our admission capabilities and capacity. It s like the Dry Bones song you learned as a child, The foot bone s connected to the leg bone, the leg bone s connected to the knee bone, the knee bone s connected to the thigh bone and so forth. Overall flow, or the system, can only be improved by applying several key strategic concepts to these disparate but equal parts. 2
KEY STRATEGIES FOR IMPROVING HOSPITAL FLOW Every System Is Designed to Produce the Results It Produces Back at that time when I was training, I was an idealistic emergency physician managing an inner city ER in downtown Los Angeles serving the poor. I stumbled across an article written by Don Berwick that said every system is perfectly designed to produce the results it produces. It changed my life. (Berwick MD, Donald M, Continuous Improvement as an Ideal in Health Care, NEJM, January 5th, 1989, p50). I started looking at the ER as a set of processes, as a system, with inputs, throughputs, and outputs. I had to change those processes if I wanted different results or outputs, because they were often perfectly designed to produce some of the very results that neither I, my co-workers, nor my patients liked. As Thomas Edison has pointed out, Discontent is the first necessity of progress. So the question became: Are waits and delays inevitable? We think the answer is no. There are several key strategic concepts which can be applied to improve patient flow: Demand-capacity management; Real-time monitoring of patient flow; Forecasting service and service demand; A basic understanding of queuing theory; The theory of constraints; Managing variation; and Appreciation of a system. Demand-Capacity Management What should capacity look like to guarantee quality care? Here s a profile of demand and capacity. NUMBER OF PATIENTS TIME The horizontal axis is time and the vertical axis is the actual number of emergency department patient by hour. In the ED this often seems totally unpredictable, however, if you look closely, you will notice there is lot of white space above and below the graph and a line drawn through the middle of the curve begins to approximate predicted demand by hour of the day or day of the week. How do we match demand and capacity? You need to predict demand based upon your historical data. You need to match your service capacity to your patient demand. You need daily patient flow predictions by hour of the day and plans to service those predictions. You also need to implement a real-time dashboard for key operational cycle times and then measure, monitor, and service those cycle times. You need to do this while respecting the desires, concerns, and goals of your people, of your team. Demand-capacity management is critical. We cannot plan on or even hope for our capacity routinely meeting our demand without planning, foresight, and action. 3
Real-Time Monitoring of Patient Flow McDonald s does it. They have cameras they nicknamed Hyperactive Bob that monitor incoming customer traffic. They use software to predict what the soccer mom in the van with three kids is going to order. They can get five to ten minutes upstream on predicting what their demand is going to be. McDonald s reduced their waiting time and cut waste in half using this process. Now imagine a busy freeway. It s a dark Friday night and you re driving 75 miles an hour on this busy freeway. Now, turn off the dashboard and headlights. How long do you think you could stand it? How long could you maintain your sense of safety and mastery? This is what we do every day in our hospitals; we show up, our real-time dashboard is not available, our headlights are turned off, and then we brag about how well we can manage the chaos. There is a better way. You need an inpatient flow dashboard. It can be grease board, pen, or electronic, but you need some way of monitoring flow in real time. All by setting up a system where units could grade how busy they were, put a temporary halt to new admissions while they managed the patients they had, monitor hospital-wide patient flow, and disseminate this information in real-time to all of the key departments and personnel. Forecasting Do you look at the weather forecast before you leave home? Forecasting plays a significant role in your life on a daily basis and ought to play a significant role in your life within the hospital. You can predict patient flow. You can predict unscheduled arrivals with about 80 to 85% accuracy. You don t know their names but you do or should know who s coming, what s coming, why they are coming, and what set of resources they ll need. How many Friday nights does it take before you decide your next Friday night is going to be different? How many flu seasons does it take before you decide your next flu season s going to be different? Patient flow is predictable. Here s what happened at Luther Middlefort Hospital once they introduced a hospital-wide intranet dashboard and a system for capping patient flow to units that were overwhelmed: Increased patient throughput, resulting in increased revenue of about $200,000 per month; Increased the percentage of patients who were put into bed within one hour from 23% to 40%; Reduced ED diversions from 12% to 1 to 2%; and Decreased the overall number of open nursing positions from about 10% to 1%. 0 :0 0 1 :0 0 2 :0 0 3 :0 0 4 :0 0 5 :0 0 6 :0 0 7 :0 0 8 :0 0 9 :0 0 1 0 :0 0 1 1 :0 0 1 2 :0 0 1 3 :0 0 1 4 :0 0 1 5 :0 0 1 6 :0 0 1 7 :0 0 1 8 :0 0 1 9 :0 0 2 0 :0 0 2 1 :0 0 2 2 :0 0 2 3 :0 0 1 5 4 1 4 9 1 2 0 8 1 8 3 7 9 9 9 1 5 3 1 6 6 2 6 9 2 5 3 2 7 7 2 3 5 2 6 0 2 7 4 2 6 8 2 9 4 3 0 7 3 3 2 3 5 2 3 4 5 2 9 9 2 7 8 2 1 1 F Y2 0 0 4 Q -1 F Y2 0 0 5 Q -1 1 6 0 1 1 9 1 0 7 8 3 7 1 7 6 8 5 1 0 6 1 5 6 2 0 8 2 2 6 2 3 0 2 6 0 2 4 3 2 6 0 2 6 0 3 0 4 2 8 6 3 0 2 3 3 3 2 8 7 2 7 0 2 6 0 1 9 8 This emergency department patient flow graph charts arrivals, patient volume, and arrival by hour of the day. 4
KEY STRATEGIES FOR IMPROVING HOSPITAL FLOW This is your emergency department. This could actually be any emergency department in the United States. The actual number of patients per hour may vary but the curve is the same. Here s a queuing model in action in a MICU: The key questions are: How many patients are coming? When are they coming? What are they going to need? Is our service capacity going to match patient demand? Queuing Theory Queuing theory is the art and science of matching fixed resources to unscheduled demand. You are engaged in a queuing model any time you are in a system with unscheduled or uncontrolled arrivals. Whether you re in a line waiting to register, at Starbucks awaiting your coffee, at McDonald s to order a burger, or serving patients in the ER, that is a queuing system. Facts about queuing systems: Systems that are serving unscheduled or uncontrolled arrivals behave in a characteristic fashion. When patient inflow and service times are random, the response to increasing utilization is nonlinear. As utilization rises above 80 to 85%, waits and rejections increase exponentially. The great thing is that at high levels of utilization, small changes can lead to big improvements. The horizontal axis is the utilization percentage. At 0%, the MICU is totally empty. At 100%, every bed is full. The vertical axis is the rejection rate. If the MICU is totally empty, the odds of you being rejected for a requested bed are zero. If the MICU is totally full, the odds of rejection are close to 100%. Notice the shape of the curve. As you get to 60% utilization, it starts to take off. You will note this is not a linear curve. It rises steeply and the sweet spot is probably about 80 to 85%. Operating at 100% capacity or occupancy, in a system with unscheduled arrivals and/or variation in service times, is absolutely the wrong way to go. There is simply no way you can run a queuing system at 100% utilization. You want to be at about 80 to 85%. In unit after unit where they have been able to drop down to 80 to 85, or even 90% utilization, hospitals are better able to handle the inflows and the variation. As a result, throughput goes up, profits go up, and healthcare worker satisfaction goes up. 5
The Theory of Constraints The Theory of Constraints was articulated by Eli Goldratt. His novel is called The Goal. It s a quick read. It s about a guy who is trying to save his job and his marriage. He is trying to save his relationship with his son, and he does it through the understanding and application of the theory of constraints. After reading this, you will never ever walk into an ER or a hospital without thinking about the theory of constraints. satisfaction went up, but were disappointed to learn that throughput cycle time, their most important objective, did not change. Does that mean that an admission nurse is a bad idea? Absolutely not. What it means is that in this particular hospital, the functions of the admission nurse were not a critical bottleneck to the flow or process of getting the admitted patient to an inpatient unit. What the hospital should have done was map out their process flow, look at the cycle times, and then experiment with an admission nurse to see if it reduced the process times. Not only does the theory of constraints allow you to understand how a system works, but it also helps you identify which bottlenecks are irritants and which bottlenecks are essential to the critical path. The key principles: Patient care is a network of queues and service transitions. An hour lost at a bottleneck is an hour lost for the whole system. Time saved at a non-bottleneck is a mirage. Efforts spent improving a non-critical bottleneck will not improve the overall performance of your process or system. Here s one example: A hospital was trying to reduce the cycle time for admitting patients from the ER to upstairs. They thought hiring an admission nurse would be the solution. They hired four people and deployed a full-time admission nurse. They pleasantly discovered that patient satisfaction, nursing satisfaction, and emergency physician 6 Efforts spent improving a non-critical bottleneck will not improve the overall performance of your system. To make matters even more complex, bottlenecks can jump around. For a certain period of time, the bottleneck may be a bed. For another period of time, it may be lab. For another period of time, it may be the discharge process. Another example. One hospital found that getting patients out of the PACU was a problem. Patients were delayed in the PACU, potentially backing up the OR. Through further investigation, they discovered that all orthopedic patients leave on an anticoagulation regimen. The true bottleneck was actually getting medication delivered on the day of discharge.
AVERAGE NUMBER OF PEOPLE ON HOLD 20 18 16 14 12 10 8 6 4 2 0 NOTE: - Average Call lasts 2 minutes. - Calls are answered by one person full time LOW MEDIUM HIGH Calls / hr = 29 Calls / hr = 28 Calls / hr = 27 Calls / hr = 25 VARIATION OF CALL LENGTH Now you are confronted with several possibilities. One you can wait to try to speed things up the day of discharge. Two you can try to speed things up the day before discharge. Or three you can get the pharmacy to deliver the meds on post-op day one. They had a just-in-time delivery system that wasn t just in time. Patients need this medicine when they re discharged. This was prolonging discharge, sometimes by a whole day. This was a key or critical constraint. The solution was to move delivery up to day one. After analysis, rapid-cycle testing and andingprototype the improved outcome was 43 out of 43 patients received medication on post-op day one with a simple process change. Managing Variation A telephone helpline is a queuing system. Here is an interesting example of the challenges we face: Let s say, that on average, a call to this helpline lasts two minutes. The calls are answered by one full-time person, or a single-server queue. Can the system handle 30 calls an hour without putting people on hold? If we have 25 calls per hour with low variability, almost nobody ends up on hold. With high variability and 25 calls per hour, we have two to three people on hold. But if there are 29 calls in an hour (high volume) and high variability, we end up with 18 people on hold. It is the variation that wreaks havoc with your systems, your service, and your people. That is one of several reasons why the sweet spot for patient flow is not 100% utilization. Because of variability, we need a slack to handle that variation. There are multiple sources of variation in health care. Are all our sick patients the same? Is every congestive heart failure patient the same as every other congestive heart failure patient? Is every AV malformation the same as every other AV malformation? You know the answer is a resounding no. We have clinical variability. We have flow variability. We have variability in our processes. We have variability in how people work. Each of these sources of variation is additive. They do not cancel out each other; they add up. 7
Look at the variation in health care patients, processes, and service delivery and think about what that does to your systems. Some of the key leverage points are: focus on operational efficiency, prioritize available resources, flexed responses, flexible scheduling, predict demand, smooth demand, and improve the accuracy of your predictions. The Appreciation of a System This is articulated in Deming s system of profound knowledge. There are four components to Deming s system of profound knowledge: 1. Appreciation of a system; 2. Knowledge of variation; 3. Theory of knowledge; and 4. Knowledge of psychology. A system must have an aim. Without aim, there is no system. The aim of the system must be clear to everyone in the system. You have to decide what the aim of your system is. Otherwise, you will have good people working under significantly different priorities. For some people, it s patient safety. For others, it s patient satisfaction. For some, it s patient flow. It could also be workforce satisfaction or controlling costs. You must look at patient flow and operations through a global lens to optimize the entire system. You cannot optimize a system by optimizing each part of the system. You can t have every part of the system busy all of the time and have a system that s optimized and performing perfectly. Some components of the system need sufficient slack to take up some of the work and variation of the other parts of the system. You cannot have every part of the system functioning at 100% utilization or capacity if you want flow through the system to be maximized. The larger the system, the harder it is to optimize but the greater the total benefits. While it may be difficult to improve emergency department patient flow, it may be much harder to improve hospital-wide patient flow. It is a challenge to improve flow through the OR, but it may be even more difficult to improve flow throughout the entire system. In Summary The key strategic concepts that you need to have a working understanding and appreciation for are: demand-capacity management, real-time monitoring of patient flow, forecasting, queuing theory, the theory of constraints, managing variation, and appreciation of a system. Genius is 1 percent inspiration and 99% perspiration. Thomas Edison The work may be hard, but the results are gratifying. The number one reason to get this right is that it is good for your patients and it is good for your people who take care of your patients. 8
KEY STRATEGIES FOR IMPROVING HOSPITAL FLOW About the Author Kirk B. Jensen, MD, MBA, FACEP, is Chief Medical Officer for BestPractices, Inc., a leading emergency physician staffing and practice management group. He has spent over 20 years in Emergency Medicine management and clinical care and is directly responsible for the coaching, mentoring, and career development of BestPractices Medical Directors. Kirk B. Jensen, MD, MBA, FACEP, Dr. Jensen is a faculty member for the Institute for Healthcare Improvement (IHI) and has held numerous leadership positions with IHI focusing on quality improvement, patient satisfaction, and patient flow both within the ED and throughout the hospital. Dr. Jensen also serves as a Medical Director for Studer Group. He was honored by the American College of Emergency Physicians as the 2010-2011 Outstanding Speaker of the Year. Copyright 2011 BestPractices Inc. All rights reserved. This publication may not be reproduced, stored in a retrieval system, or transmitted in any form or by any means electronic, mechanical, photocopying, recording, or otherwise without prior permission of the copyright owner. This White Paper is an informational document. Readers should note that this document does not represent an endorsement by any entity. All page headers and custom graphics are service marks, trademarks, and/or trade dress of BestPractices, Inc. All other trademarks, product names, and company names or logos cited herein are the property of their respective owners. Any comments relating to the material contained in this document may be sent to the BestPractices Marketing Department: Email: info@best-practices.com Mail: BestPractices, Inc. Marketing Director 10306 Eaton Place, Suite 180 Fairfax, VA 22030 9
Contact 10306 Eaton Place Suite 180 Fairfax, VA 22030 (800) 910-3796 info@best-practices.com www.best-practices.com