“Curmudge, the last time we talked you and some swamp-dwelling possum were telling me that the hills and valleys in hospital routine were man-made. You also said that handling the workload peaks stressed the staff and reduced the patients’ safety and quality of care. Finally, you promised to tell me what can be done to improve this situation. Now’s the time. What’s the word?”
“Lean.”
“Lean? Is that your one-word, one-size-fits-all default answer?”
“Well Jaded Julie, if you must have a second word, it’s heijunka or level loading. But in this case, heijunka is more of a goal than a tool. People in a hospital with high variability need to employ the Lean philosophy and techniques that we have been talking about for the past several months. Kaizen or rapid process improvement events, current- and future-state value stream mapping, go to gemba, A3 reports, the Deming cycle…the whole Lean Geschäft (business).”
“Easy, Curmudge. Please limit your enthusiasm to English words.”
“Getting the right people on the team is most important. If physicians will be impacted, they must become active and engaged team members. Not just any docs; they must be the docs that the others respect. The team should first look at variability and identify that which is man-made. When I say ‘look at,’ that means to go to gemba, gather data, and construct value stream maps. Plan and evaluate trials of changes that might smooth out the variability (plan-do-check-act). Consider any process element where ‘batching’ occurs to be a candidate for smoothing.”
“All of that sounds like Lean with a capital ‘L’, Curmudge. Have any hospitals undertaken this kind of effort?”
“You can read about them in the presentations by Litvak (1) and Paulson (2). They suggest that quality of care can be increased ‘free of charge.’ These are some of the benefits that have been achieved (1):
· Reduced ED diversions
· Greater staff and patient satisfaction
· Reduced mortality and medical errors
· Reduced length of stay
· Increased hospital throughput
· Increased surgical throughput”
“I presume that after a hospital has smoothed out its artificial variability it can better manage its natural variability with queuing theory.”
“That’s the idea, Julie. Characteristics of a system that one must know are the average patient arrival rate and the system’s average service rate. One can then determine staffing and numbers of rooms that will be needed to meet demand. Some of this is shown in the presentations that have been cited (1, 2). Other situations are described in the notes from Litvak’s ‘Queuing Theory’ course offered through IHI. Additional study will be needed to tackle more complex scenarios.”
“It sounds as if you are saying that for tougher situations I’m on my own.”
“Julie, as long as I am here, you’ll never be on your own. And don’t forget my colleagues in the Kaizen Promotion Office.”
Affinity’s Kaizen Curmudgeon
(1)http://www.iom.edu/Object.File/Master/21/207/Litvak_Variability%20in%20Patient%20Flow_updated.pdf
(2)http://www.hospitalcouncil.net/Upload/BEACON_PAULSON_IntrodImprovingInpatientFlow_4-24-07.pdf
Thursday, July 30, 2009
Thursday, July 16, 2009
Managing Variability--The Problem
“’Managing Variability’ doesn’t look too tough, Curmudge. After your mentioning queuing theory and mathematical models in our last discussion, I was afraid you’d dump a pile of differential equations—whatever they are—on me.”
“Don’t worry, Jaded Julie. Math is too complex for a blog. But to temper your exaltation, I’d better warn you that we have a new Japanese word to learn. The word is heijunka, and it means ‘level loading.’”
“I know it must make sense in your ancient alchemist’s mind, Curmudge, but would you please share with me the connection, however tenuous, between heijunka and queuing theory. My brain is spinning, and we’ve only talked for three short paragraphs.”
“It’s really quite logical. Remember Dr. Deming’s writings about common-cause or natural variation as well as special-cause or artificial, man-made variation. We have both kinds in a hospital, such as the random patient arrivals in the ED and the man-made hills and valleys in people’s workloads caused by block scheduling of elective surgeries, morning rounds, and physician-requested early morning blood draws. Although hospital schedules are largely traditional, we must consider them man-made. As Pogo said, ‘we have met the enemy and he is us.’”
“(Pogo? Who’s Pogo? I’d better google him. Oh, Greybeard is probably just recalling some long-ago cartoon character.) So you and this Pogo-person (Pogo-thing?) are saying that traditional hospital routine is a man-made system designed to give hill-and-valley workloads, and ‘surprise!’ that’s what we get.”
“That’s it, Julie. We can’t begin to use queuing theory on the random variation until we have knocked the peaks off the artificial variation. That’s level loading or heijunka, and the study of the hills and valleys is called variability analysis.”
“But, Curmudge, traditional hospital routine is just the way it is. Don’t we have to live with it?”
“What you’re saying sounds a lot like, ‘We’ve always done it that way.’ That’s a very un-Lean statement that flies in the face of ‘continuous improvement.’”
“Sorry, Curmudge. When I get home I’ll wash my mouth out with soap. But what is so bad about variable workloads?”
“Dr. Eugene Litvak, the guru of queuing theory applied to health care, and his disciples feel that the OR is the driver of artificial variability. (1, 2) High-productivity surgeons tend to perform elective surgery on the same day every week. This impacts the whole hospital: admitting, discharging, the lab, housekeeping, and especially the availability of rooms for admissions from the ED. If a hospital’s midnight census is over 90%, it has very little capacity available for new admissions the next morning. (3) This might also cause internal gridlock, when patients can’t be moved from where they are presently (e.g., the ICU) to where they need to be (e.g., med/surg acute care). ‘Optimal care can only be delivered when the right patient is in the right place with the right provider and the right information at the right time.’ (3)”
“This must also affect the scheduling of nurses.”
“It does. If the hospital schedules nurses to accommodate the peaks, it incurs excessive costs. If nurses are scheduled to meet the average census, they are stretched to the limit in handling the peaks.”
“My guess is that nurses would rather have level loads than hills and valleys.”
“It’s more than just ‘rather,’ Julie. According to the literature (2), ‘overworked nurses and stressed resources raise the risk of error, increases staff dissatisfaction, and staff turnover.’ In addition, JCAHO says that 24% of sentinel events are associated with nurse understaffing.”
“That’s alarming, Curmudge. What can be done about it?”
“Next week we’ll talk about countermeasures. I hope you can wait.”
“I can wait if you will answer one question. Over the past year you have quoted a 19th Century novelist (Alexandre Dumas), several 21st Century professors (John Kotter, Jim Collins, and others), and now a 20th Century swamp-dwelling possum (Pogo). A possum? That doesn’t seem very scholarly.”
“Julie, I’d even go to the Okefenokee for the right kind of insight.”
Affinity’s Kaizen Curmudgeon
(1)http://www.iom.edu/Object.File/Master/21/207/Litvak_Variability%20in%20Patient%20Flow_updated.pdf
(2)http://www.hospitalcouncil.net/Upload/BEACON_PAULSON_IntrodImprovingInpatientFlow_4-24-07.pdf
(3) IHI White Paper, Optimizing Patient Flow, Innovation Series 2003. Institute for Healthcare Improvement.
“Don’t worry, Jaded Julie. Math is too complex for a blog. But to temper your exaltation, I’d better warn you that we have a new Japanese word to learn. The word is heijunka, and it means ‘level loading.’”
“I know it must make sense in your ancient alchemist’s mind, Curmudge, but would you please share with me the connection, however tenuous, between heijunka and queuing theory. My brain is spinning, and we’ve only talked for three short paragraphs.”
“It’s really quite logical. Remember Dr. Deming’s writings about common-cause or natural variation as well as special-cause or artificial, man-made variation. We have both kinds in a hospital, such as the random patient arrivals in the ED and the man-made hills and valleys in people’s workloads caused by block scheduling of elective surgeries, morning rounds, and physician-requested early morning blood draws. Although hospital schedules are largely traditional, we must consider them man-made. As Pogo said, ‘we have met the enemy and he is us.’”
“(Pogo? Who’s Pogo? I’d better google him. Oh, Greybeard is probably just recalling some long-ago cartoon character.) So you and this Pogo-person (Pogo-thing?) are saying that traditional hospital routine is a man-made system designed to give hill-and-valley workloads, and ‘surprise!’ that’s what we get.”
“That’s it, Julie. We can’t begin to use queuing theory on the random variation until we have knocked the peaks off the artificial variation. That’s level loading or heijunka, and the study of the hills and valleys is called variability analysis.”
“But, Curmudge, traditional hospital routine is just the way it is. Don’t we have to live with it?”
“What you’re saying sounds a lot like, ‘We’ve always done it that way.’ That’s a very un-Lean statement that flies in the face of ‘continuous improvement.’”
“Sorry, Curmudge. When I get home I’ll wash my mouth out with soap. But what is so bad about variable workloads?”
“Dr. Eugene Litvak, the guru of queuing theory applied to health care, and his disciples feel that the OR is the driver of artificial variability. (1, 2) High-productivity surgeons tend to perform elective surgery on the same day every week. This impacts the whole hospital: admitting, discharging, the lab, housekeeping, and especially the availability of rooms for admissions from the ED. If a hospital’s midnight census is over 90%, it has very little capacity available for new admissions the next morning. (3) This might also cause internal gridlock, when patients can’t be moved from where they are presently (e.g., the ICU) to where they need to be (e.g., med/surg acute care). ‘Optimal care can only be delivered when the right patient is in the right place with the right provider and the right information at the right time.’ (3)”
“This must also affect the scheduling of nurses.”
“It does. If the hospital schedules nurses to accommodate the peaks, it incurs excessive costs. If nurses are scheduled to meet the average census, they are stretched to the limit in handling the peaks.”
“My guess is that nurses would rather have level loads than hills and valleys.”
“It’s more than just ‘rather,’ Julie. According to the literature (2), ‘overworked nurses and stressed resources raise the risk of error, increases staff dissatisfaction, and staff turnover.’ In addition, JCAHO says that 24% of sentinel events are associated with nurse understaffing.”
“That’s alarming, Curmudge. What can be done about it?”
“Next week we’ll talk about countermeasures. I hope you can wait.”
“I can wait if you will answer one question. Over the past year you have quoted a 19th Century novelist (Alexandre Dumas), several 21st Century professors (John Kotter, Jim Collins, and others), and now a 20th Century swamp-dwelling possum (Pogo). A possum? That doesn’t seem very scholarly.”
“Julie, I’d even go to the Okefenokee for the right kind of insight.”
Affinity’s Kaizen Curmudgeon
(1)http://www.iom.edu/Object.File/Master/21/207/Litvak_Variability%20in%20Patient%20Flow_updated.pdf
(2)http://www.hospitalcouncil.net/Upload/BEACON_PAULSON_IntrodImprovingInpatientFlow_4-24-07.pdf
(3) IHI White Paper, Optimizing Patient Flow, Innovation Series 2003. Institute for Healthcare Improvement.
Thursday, July 9, 2009
An Introduction to Queues
“What’s a ‘cue-you,’ Curmudge?”
“It’s not a ‘cue-you,’ Jaded Julie; it’s pronounced just like the letter ‘Q.’ It comes from the French word for tail, and it means a line of people waiting their turn. So when you walk into McDonald’s, you are at the tail end of the tail.”
“Well that’s something everyone should know. If you can justify this enlightening discussion, now would be an excellent time to do it.”
“Of course, Julie. Today we are going to talk about customer flow management, a part of which is queuing theory; and before you ask me ‘why?’ I’m going to tell you why. An important aspect of the Toyota production system is one-piece flow. They build cars in an essentially continuous process with minimal inventory, work in process, and waste. Wouldn’t our patients be delighted if we could treat them in a similar, efficient process?”
“They definitely would like to be treated efficiently but probably not like a partially-built car. While patients are sitting in a waiting room, it probably wouldn’t make them any happier to think of themselves as ‘work in process.’”
“You’re right as usual, Julie. We certainly must treat our patients with respect, dignity, and compassion. But for today’s discussion, let’s think of patients as people who just want to arrive, receive a service, and depart ASAP.”
“Okay, Curmudge. I’m with you, at least for the moment. But if any of our discussion is going to involve queuing theory, I really need to know what that is.”
“Queuing theory uses mathematical models in analyzing how to match demand (usually variable) with capacity (usually fixed). Unfortunately, there’s a major limitation to elementary queuing theory: it can only be applied when customers arrive in a random manner. It doesn’t apply when service is provided on an appointment basis, which excludes most of the hospital. The ED and walk-in clinics would be examples of where patients arrive randomly.”
“If the use of queuing theory is so limited, why should we even study it?”
“Traditionally, when a patient walks into a clinic or hospital they expect they will have to spend time waiting. If queuing theory might help shorten waiting times, it’s certainly a subject with which we should have an acquaintance.”
“I guess it’s okay to get acquainted, but I don’t want to get too close to any mathematical models. Can you give me a non-mathematical introduction?”
“I’d be delighted, Julie. Linear queuing means that the customer waits physically in a line. (1) The simplest version is Single-Queue-Single-Service-Point, like waiting in line to buy your ticket for a popular movie. Several of these side-by-side make a Multiple-Queue-Multiple-Service-Point system like in a grocery store. This can be made more efficient by adding Segmentation, where certain lines handle only customers with 15 items or less. The most advanced form of linear queuing is Single-Queue-Multiple-Service-Points. A common example is the Post Office, where the line is not held up when one of the clerks is delayed by a customer sending a package to Timbuktu.”
“I must admit that I haven’t given much thought to the type of queue that I’m in. I do know that in the grocery store Erma Bombeck’s Law applies, ‘The other line always moves faster.’” (2)
“There are several variations on these simple systems. One is virtual queuing, where, after registering, a customer is sent to a waiting room rather than a physical line. Because the customer does not know his/her exact position in the queue, the staff is able to exercise some segmentation. An example would be an ED in which more urgent patients are seen first. A variation of this, virtual queuing with buffering, is also used in an ED; the next patient to be seen is taken to an exam room (the buffer) to minimize the service provider’s delay.”
“I know there are more, Curmudge. There always are.”
“Another fairly simple one is the multiple stage system in which the customer is served by more than one facility before leaving. This is termed ‘serial servers.’ I’ve seen this in European fast food restaurants where one decides what he wants, lines up to pay for it, and then lines up again to receive the food in exchange for the receipt. A similar system exists in a hospital where an outpatient must go to the lab and then to radiology.”
“Is there no end to this, Curmudge?”
“There is for us today, Julie, and it’s right now. Let’s go down to the cafeteria and stand in a queue.”
Affinity’s Kaizen Curmudgeon
(1)http://www.q-matic.com White Paper: An Introduction to Customer Flow Management
(2)Maister, David H. Note on the Management of Queues, Harvard Business Review (March 17, 1995): 1-14, Reprint #9-680-053.
“It’s not a ‘cue-you,’ Jaded Julie; it’s pronounced just like the letter ‘Q.’ It comes from the French word for tail, and it means a line of people waiting their turn. So when you walk into McDonald’s, you are at the tail end of the tail.”
“Well that’s something everyone should know. If you can justify this enlightening discussion, now would be an excellent time to do it.”
“Of course, Julie. Today we are going to talk about customer flow management, a part of which is queuing theory; and before you ask me ‘why?’ I’m going to tell you why. An important aspect of the Toyota production system is one-piece flow. They build cars in an essentially continuous process with minimal inventory, work in process, and waste. Wouldn’t our patients be delighted if we could treat them in a similar, efficient process?”
“They definitely would like to be treated efficiently but probably not like a partially-built car. While patients are sitting in a waiting room, it probably wouldn’t make them any happier to think of themselves as ‘work in process.’”
“You’re right as usual, Julie. We certainly must treat our patients with respect, dignity, and compassion. But for today’s discussion, let’s think of patients as people who just want to arrive, receive a service, and depart ASAP.”
“Okay, Curmudge. I’m with you, at least for the moment. But if any of our discussion is going to involve queuing theory, I really need to know what that is.”
“Queuing theory uses mathematical models in analyzing how to match demand (usually variable) with capacity (usually fixed). Unfortunately, there’s a major limitation to elementary queuing theory: it can only be applied when customers arrive in a random manner. It doesn’t apply when service is provided on an appointment basis, which excludes most of the hospital. The ED and walk-in clinics would be examples of where patients arrive randomly.”
“If the use of queuing theory is so limited, why should we even study it?”
“Traditionally, when a patient walks into a clinic or hospital they expect they will have to spend time waiting. If queuing theory might help shorten waiting times, it’s certainly a subject with which we should have an acquaintance.”
“I guess it’s okay to get acquainted, but I don’t want to get too close to any mathematical models. Can you give me a non-mathematical introduction?”
“I’d be delighted, Julie. Linear queuing means that the customer waits physically in a line. (1) The simplest version is Single-Queue-Single-Service-Point, like waiting in line to buy your ticket for a popular movie. Several of these side-by-side make a Multiple-Queue-Multiple-Service-Point system like in a grocery store. This can be made more efficient by adding Segmentation, where certain lines handle only customers with 15 items or less. The most advanced form of linear queuing is Single-Queue-Multiple-Service-Points. A common example is the Post Office, where the line is not held up when one of the clerks is delayed by a customer sending a package to Timbuktu.”
“I must admit that I haven’t given much thought to the type of queue that I’m in. I do know that in the grocery store Erma Bombeck’s Law applies, ‘The other line always moves faster.’” (2)
“There are several variations on these simple systems. One is virtual queuing, where, after registering, a customer is sent to a waiting room rather than a physical line. Because the customer does not know his/her exact position in the queue, the staff is able to exercise some segmentation. An example would be an ED in which more urgent patients are seen first. A variation of this, virtual queuing with buffering, is also used in an ED; the next patient to be seen is taken to an exam room (the buffer) to minimize the service provider’s delay.”
“I know there are more, Curmudge. There always are.”
“Another fairly simple one is the multiple stage system in which the customer is served by more than one facility before leaving. This is termed ‘serial servers.’ I’ve seen this in European fast food restaurants where one decides what he wants, lines up to pay for it, and then lines up again to receive the food in exchange for the receipt. A similar system exists in a hospital where an outpatient must go to the lab and then to radiology.”
“Is there no end to this, Curmudge?”
“There is for us today, Julie, and it’s right now. Let’s go down to the cafeteria and stand in a queue.”
Affinity’s Kaizen Curmudgeon
(1)http://www.q-matic.com White Paper: An Introduction to Customer Flow Management
(2)Maister, David H. Note on the Management of Queues, Harvard Business Review (March 17, 1995): 1-14, Reprint #9-680-053.
Subscribe to:
Posts (Atom)