A Guide to Implementing the Theory of Constraints
Caring About Healthcare
“Here is Edward Bear, coming downstairs now, bump, bump, bump, on the back of his head. It is, as far as he knows, the only way of coming downstairs, but sometimes he feels that there really is another way, if only he could stop bumping for a moment and think of it. And then he feels that perhaps there isn't.” ‑‑ A. A. Milne.
How many Edward Bears are there here I wonder? How many people, like Edward Bear, are dissatisfied with the current situation, wonder if there can’t be something better, and then doubt that there is? Probably quite a few I suspect. Well, the following pages were written with the Edward Bears of healthcare in mind. However, for these pages to be of any benefit, we will have to stop bumping the back of our head for more than just a moment – there is quite a bit to think about; things that we might not have thought about before.
If you have just “landed” on this page as the consequence of some frustrated internet search on healthcare and improvement and Lean, or some other such facet, then don’t push the “back button” yet. Hold on for a moment and I will provide some rationale.
You may feel uneasy seeing headings above such as “production” and “supply chain” and “projects,” as this indicates some operational bias to this website. But this is exactly as it should be; most of the problems in healthcare are operational. Most of the problems in any modern enterprise of more than one department are operational. Why is this? Well, operations are full of people, and with that there are a few challenges that we need to learn to overcome. We are going to do that by invoking a management philosophy known as Theory of Constraints. If you suffer from improvement fad fatigue, then you have my sympathy (which is of no use to you), and you also have my understanding. We are going to look at why several of the current and past fads have failed to yield the necessary results and also look at the hugely valuable and more fundamental aspects that do yield results. Outside of the Toyota Production System, Theory of Constraints is the only consistent and fundamental approach to process improvement.
I am going to make an assumption that many people in healthcare are very busy – after all everyone else is, so I doubt that healthcare is much different, and maybe it is somewhat worse. Therefore, this page will not only serve as an introduction but also as a directory to other parts of the website where there is material that you may need to follow up. As we learn to bump our heads less often, we will find that we have more time.
People who have visited this website before will know that I have left healthcare until after production, supply chain, and projects – the three major logistical arms that Theory of Constraints applies to. However, my interest in healthcare predates all of these. It was my interest in service operations that sent me in search of better approaches to the very real chaos that prevails throughout professional consultancies. After all, if seemingly clever professionals such as engineers and scientists are in this state – and a rather permanent state at that – what then of other forms of business? Were we missing something? Could we learn? And I rather picked that of all the service operations, surely modern healthcare services must be the most complex of the lot, surely anything “new” must stand or fall in this environment – hence part of the attraction. I’ve since learnt that nothing is too complex, but most things are simply made way too complicated, complicated beyond belief. The world is not a complicated place, but we certainly force it to behave in a very complicated manner.
Much of the problem in modern healthcare – like so much else in the modern world – has to do with the dynamics of our organizations, not the detail, after all we are all expert in the day-to-day detail of our profession or area or what have you. In fact, from the start we have used two terms here from systems thinking; “dynamic complexity,” and “detail complexity.” Somewhere along the way we have asserted that; “dynamic complexity isn’t.” And indeed dynamic complexity isn’t complex at all. It is our psychology and our failure to understand what has gone on in the past 100 years that is the cause of our current situation and the source of the apparent complexity.
So, take a step back for a moment and think. Think how little removed we are from the discovery of modern antibiotics, how little removed we are from anesthetic via a chloroform mask, indeed how little removed we are from formerly complicated surgery such as appendectomy that used to require an incision that seemed to extend from the pelvis to the arm.
All of this, of course, is detail complexity. There is so much we can do today that was unimaginable 10 or 20 or 30 years ago. Relative expectations right around the world are also so much greater as a consequence. But this is true of any aspect of human endeavor. People are unlikely to forgo their Ipods for 78’s (a 78 was a flat plastic disk with music recorded as analogue grooves that could be played back using a needle – strange but true), so too with healthcare. What is missing is an effective approach to dynamic complexity.
Think back again for a minute. Think back again to hospitalization in the time of our parents or our grandparents. Was it different? Of course it was. It was vastly different. It was much simpler. It doesn’t matter what industry you run this test on, it will be the same. It was simpler in that there were less dependencies with which to deal with, and less choice in any case when such a dependency was brought to bear. I guess that we all long for a former simpler experience that we seem to have lost, but at the same time we don’t want to forgo the improved detail that we have gained since.
However this isn’t a win-lose situation. It is win-win, if only we are willing to learn how. And in order to be willing to learn how, we must be willing to suspend some of the things that we have learnt in the past and still hold to be true. There is a fundamental paradox in that what we have learnt through personal experience is exactly what we don’t need if we are to successfully operate a modern healthcare system.
There’s that damn word “system” again, it will undoubtedly keep popping up. The peculiarities of the dynamic complexity systems that we are going to look at is that they are composed of tasks or steps or stages that are strongly linear, serial, and dependent. Many stages have a specialist staff that only do one or two particular things (and very well at that). The trouble is that all of “this” has only happened in the last 100 years or so, barely a couple of generations. We don’t teach people how to manage this situation because we fail to recognize that it is different. We “know” how to manage each task or step or stage – or so we think, if they existed in isolation – and we extrapolate that out to the belief that competent management of the whole system is simply the sum of the competent management of each and every task along the way.
In these webpages we have consistently called this the Reductionist/local optima approach.
But there is another approach that we must use, the Systemic/global optimum approach.
You see, healthcare from a dynamical point of view is like a chain, and a chain as we know is only as strong as its weakest link. We must seek out the dynamical weakest link and strengthen it and protect it, and once strengthened and protected, we must search for the next, and so forth, in a process of on-going and continual improvement. It is not just the detail of healthcare solutions that must improve, it is the dynamics of the delivery that must improve as well – and we are starting from behind.
We’ve been here before, many times before, since modern hospitalization first occurred; scratch the surface and many things that we now take for granted have their origins in Scientific Management from the early 1900’s. At the moment Lean is the lead mechanism for healthcare process improvement. That healthcare professionals are receptive to Lean methodologies indicates only too well the underlying desire to do better with the limited resources that we currently do have. After all, everybody wants to do their best – right?
But Lean is an amalgam of a number of previous attempts at system improvement. These older strands weave in and out of the current approach, however, one or two important strands that ought to be in the mix have been dropped out, and one or two that ought not to be in the mix have been added in. Let’s address these briefly.
Lean is essentially an explicit Western academic interpretation of a tacit Eastern industrial reality that we know today as the Toyota Production System. The Toyota Production System isn’t a recent development, the philosophical roots go back to the late 1880’s within the Toyoda family, the intent to manufacture cars extends back to 1911, and the actuality of beginning to build cars dates from the early 1930’s (1).
The two pillars of the system are;
§ Autonomation (automation with a human touch).
These come from the silk spinning & weaving origins of the company and the Toyoda family. Integral within this is a focus on the absolute minimization of waste. This major thread, the absolute minimization of waste comes via the contemporaneous work of Frederick Taylor and Frank and Lillian Gilbreth in North America in the late 1890’s and early 1900’s (2). Lillian Gilbreth visited Japan on a number of occasions and Taylor’s work was translated and vigorously advocated by Japanese nationals at the time. Not only has the elimination of waste had a pedigree extending back to North American industrialization in the 1900’s, it has also reappeared under the auspices of; Kaizen, Total Quality Management, and World Class Manufacturing – each of these being an attempt to describe key Japanese industrial expertise. As I said, we have been here many times before.
A strand that has been left out of the Lean mix is the impact that W. Edwards Deming’s methods had on Toyota from 1960 onwards, and many other major Japanese industries since 1950. Deming’s work is based upon the foundations established by Walter Shewhart at Bell Labs in the mid-1920’s (3). These days the work of Deming is pretty much bastardized under the banner of Six Sigma and we are all the poorer for it. And this is why people have to talk about Lean and Six Sigma as a pair within the same breath and within the same sentence.
These three approaches; the Toyoda’s, Taylor, and Deming/Shewhart were concomitant with the development of modern industrialization and the needs that arose out of that. By the mid-1920’s we had the essentials of what we require except for one thing – an understanding that we have only had since the mid-1980’s. We will come back to that soon enough.
Another strand that is left out of the Lean mix is the logistical backbone of Just-In-Time, either the use of Kanban logistics or Tact time. The reason for this is two-fold. Firstly Kanban and Tact time really are best suited for repetitive manufacturing. There are nevertheless places in healthcare where they could be used – drug and consumable resupply for instance – but generally they are not. This is because of the second reason, the West’s predilection for data (detail complexity) and computer information systems, especially those known as material resource planning (MRP II) or enterprise resource planning (ERP). It seems to be poorly understood is that Kanban logistics has the essential function of stopping the waste of over-production.
More critically, Kanban is the focusing mechanism or driver for process improvement within the Toyota Production System. A common enough analogy is a boat floating over hidden rocks. The rocks (our problems) are hidden in a sea of work-in-process. In healthcare we call these queues, and they contain people called patients. The Kanban system functions by reducing the sea of work-in-process step-by-step until it uncovers a new problem. The new problem is then addressed using various techniques and improvements are made until there is no longer a problem. The cycle then continues. Without a focusing system for the whole process we are left with only local initiatives, and that unfortunately is how Lean in healthcare functions at the present.
What then of the threads that have been added to Lean that ought not to have been? Well, the predominant one is value stream mapping. The Toyota Production System is recognized as a new sociotechnical system (4); it is characterized by people who work within the system all of their working lives, it is characterized by people who move around within the system, 6 months here, 2 years there, learning how the process works. In the West where, to paraphrase Deming, we don’t have such constancy of purpose, people don’t actual know the entire stream of their process, and if we bring in consultants, which isn’t such an unusual occurrence, the matter is made even worse.
We can’t blame our people, after all our whole management system is built around local efficiency and departmental optimization, and if that wasn’t enough, our career progression is determined by individual achievement, not group achievement. That, however, isn’t a good reason to adopt value stream mapping. With focus, you can bring about real improvement in the time that most people are still trying to work out how to map their value stream.
In the section called “& More ...” we discussed in some detail; Deming, Taylor, and Toyota – in fact a page for each. The point of this was to illustrate how they have each been systematically mis-understood. They have been systematically mis-understood because we operate under a paradox or a “cloud” whereby our personal experience as individuals stops us from learning from our everyday experience in industry. We called this the fundamental cloud. You can find two files in the PowerPoint section that address this in general terms.
Lean and the predecessors such as Kaizen, Total Quality Management, and World Class Manufacturing will not help us because they are bereft of context. And the context must be systemic. The original work of Deming, Taylor, and Toyota had the context embedded within it. Descriptions of the derivatives lack this aspect. They are, as Taylor described it in 1911, mechanisms without the essence. You can see the problem isn’t a new one.
Well, it is a both a curious and fortunate thing that healthcare professionals are actually exposed to this context, the essence, within their professional lives. Healthcare professionals do understand systems – albeit a plethora of sub-systems – in the form of the human body. And each of these systems has some rate limiting factor, a weakest link. So too with our healthcare processes. Indeed the healthcare process is best viewed as a patient.
Lets take a look.
Healthcare professionals know about rate limiting steps. Rate limiting steps are of course a dynamic entity, however, we are going to take a step back for a moment and use a simple static analogy – a chain.
Everyone knows that;
The weakest link in our static chain is analogous to the rate limiting step in a dynamic system. A static chain should not be our preferred analogy, but let’s continue to work with it for a short while. It has its advantages – for instance, you can go and cut a plastic chain and use it as a real-life analogy.
What about this chain then?
Where is the weakest link? It clearly has to have one. If we were to pull the chain it would eventually break at one of the links, but which one? This is often very much closer to peoples’ real experience. The analogy makes sense but the location of the weakest link is no longer so clear – or it is clear in so much as everyone’s own department is “probably” the weakest link. But that won’t do. There is only one. This brings us to the next point.
Unfortunately in Western management we break the system down into pieces and seek overall efficiency and optimization from every link – that’s why it feels as though “our” link is always the weakest. If we don’t believe that, then we just need to go and look at the KPI’s that people are expected to perform to, or go and look at the management accounting figures – the subdivision and allocation of costs across the system as though each link is independent. Either of these approaches ignores the serial dependency, in fact interdependency, between the various links.
We must replace this with a different paradigm;
The key to this quest is the identification of the weakest link, the rate limiting step in our process. Once we know where this is, it then offers us huge leveragability.
But let’s now change our analogy; change it away from a physical chain to one of a “group of groups” of people. Let’s have a look.
Firstly, here is a group of people.
They might be a department or a team, we don’t need to know exactly what, they are after all just an analogy.
And here we have a group of groups of people.
Well that’s nice. And in our local efficiency/departmentalize view of the world we might accept this diagram at face value. But the chain analogy begs the question, what are the links? There are links, let’s have a look.
The linkage is provided by patients! Scary thought – right? After all, if there were no patients then healthcare would be an extremely efficient enterprise. The patients transmit things. I don’t mean diseases, I mean things like variability. Basically we have a process – a serial dependent process.
Let’s have a look at the basic components of the process.
Each group, whether they are defined by location or specialty or some other factor, have some inherent variability. The less mechanized a particular step is, the more variability it will have. This is why manufacturing tries to mechanize, and once mechanized, tries to automate.
Let’s look at this another way. Jaques and Cason define work as; “the exercise of judgment and discretion in making the decisions necessary to solve and overcome the problems that arise in the course of carrying out tasks (5).” This is a non-trivial definition. If we no longer need to exercise judgment and discretion we can mechanize or computerize things and maybe automate them as well. This is where Ohno’s “autonomation” comes in – and remember Ohno’s initial context was automated silk spinning and weaving, way before any form of electronic control – machines replace the judgment and discretion that was once the exclusive domain of people. And why would we want to maintain such exclusivity if we can off-load it for more humanistically rewarding endeavors?
So ask yourself; is the exercise of judgment and discretion a feature of healthcare? That’s a rhetorical question, isn’t it? Of course the exercise of judgment and discretion is a feature of healthcare. That is why our health specialists train for so long, and in fact never stop learning. Therefore, we should expect some variability. And don’t forget the patients – they aren’t exactly uniform either. The potential mix of any one single patient’s “complicating factors” means variability is rife within this process.
If variability is rife, then it is only natural that we should attempt to reduce it, and hence the attraction of the various methodologies that we have mentioned. If we can locally reduce variability we ought to expect a better local outcome. Let’s show this.
In fact, it seems that we have spent an inordinate amount of effort in reducing medical, surgical, and technical variation, but little effort to date on process variation – the dynamic “complexity” – the bit that deals with the dependency between the different groups. There is a simple test for this, has the total output gone up as the local outcomes have improved? What is the answer? Shouldn’t we expect it to do so?
Well, the results across the world says no, the total output does not improve. We have buried money into this system in the last decade and the productivity has been static, people have to be more ill to be seen, and they wait longer and longer to be seen. Why, why, why, is this?
Let’s have a look.
There are two issues, or two parts of the same issue. The first part of the issue is that we have a system. A system is irreducible, it has no parts. That of course doesn’t stop us from trying to make it have parts. But that is a problem with our lack of experience with modern industrial systems. They didn’t exist a couple of generations ago. We think that we can “wing it” with our old knowledge, but we can’t. In fact, we are in denial that we can’t. The fact that almost everyone else is equally hobbled only makes it look more as though we can make it. And those organizations that we term exceptional, we tend to explain away with all manner of excuses, to do otherwise would be to accept responsibility for what we do not know.
It’s sad that since at least 1911 we have known that; “In the past man has been first; in the future the system must be first (6).” But we still don’t understand what this means. We blame “the system,” even though we are part of it, because we don’t understand it; and we don’t understand it in two mutually exclusive directions;
§ Vertical hierarchy – the requisite number of layers and necessary competence.
§ Horizontal process – the existence of serial dependency and variability.
And it is within the horizontal process that is the second part of the issue with systems lies.
Let’s have a look.
The second part of the issue is that all systems have a weakest link, a rate limiting step. Sure, most industry strives for “balanced” lines, but this is just a manifestation of our own illusion, or should that be delusion, of independence. There are no balanced lines. There is no way that we could balance health (or any other service) even if such a concept existed, we simply don’t have control over the rate at which new work enters the system. We might hope that on average we have enough capacity, but reality is that we must have considerably more.
The two earlier logistical systems, Ford’s process chain, and Toyota’s Kanban system implicitly acknowledge the existence of a rate limiting step. Ford uses mechanical linkages and Toyota uses cards to communicate, across the system, the speed of the rate limiting step – these systems can’t operate faster than the rate limiting step and all other parts can not over-produce relative to the rate limiting step. Theory of Constraints continues in this fashion using time as a logistical “rope.” It is also the first to explicitly recognize the role of the constraint. This is the one thing that has been missing since the 1920’s; explicit recognition of the constraint – the rate limiting step.
Deming was so very close to this understanding when he described the obligation of a component to the whole (7). And yes I know I said there were no parts, but our language lacks the words to describe this.
Let’s draw the understanding as described by Deming.
Here, each step in the process is obligated or subordinates to the overall whole. But this is not quite sufficient. It might do in a network where there is considerably more independence between sub-parts, but it won’t do where there is strong serial dependency.
Rather we must have this;
In order to make the best of the system as a whole, we must exploit the constraint or rate limiting step and we must subordinate everything else. The Kanban approach in the Toyota Production System – the thing that Lean most often leaves out – is the synchronization mechanism that ensures the stronger areas with more capacity are truly subordinated to the rate limiting step.
It may seem counterintuitive, but the discipline for good subordination is far more important than the discipline for exploitation. We know how to exploit, we try to do it everywhere all of the time (just trying to doing our best), we need to focus instead on just the one place where exploitation needs to be done in order to improve the system as a whole, and subordinate everything else.
We need a systematic and systemic focusing process in healthcare to find and identify the places to exploit and the places to subordinate if we are to replace local improvements in outcomes with system-wide improvements in output. Let’s have look at such a process.
Taylor, Deming, and Toyota all have a systemic context embedded within their methodologies, but none of these previous approaches explicitly recognizes the role of rate limiting step in the process. Theory of Constraints does. Moreover, it recognizes time rather than materials as the fundamental measure. By doing so, a series of similar logistical solutions have been implemented that bring about rapid, significant, and sustained improvement to industrialized processes; manufacturing of all types, supply chain, and projects. The same principles can be applied directly to healthcare with the same results, and have been for more than a decade (8).
Knowing what you now know, would you choose to improve a step that is not rate limiting? You might if there is a significant deficiency in the current outcome at that step – a medical or technical or quality issue. But would that improve the output of the whole chain? Unfortunately not. This is why, cost of personnel excluded, so much money is expended upon healthcare for so little improvement in output.
What we have lacked is focus, and the logistical solutions to go with it. However, both do exist. Let’s look at focus first.
Goldratt proposed a focusing process as follows (9);
(1) Identify the system’s constraints.
(2) Decide how to Exploit the system’s constraints.
(3) Subordinate everything else to the above decisions.
(4) Elevate the system’s constraints.
(5) If in the previous steps a constraint has been broken, Go back to step 1, but do not allow inertia to cause a system constraint.
The background to this is discussed in detail on the page called Process of Change, the central importance of time is discussed in detail on a page called Evaluating Change, and nature of subordination is discussed on a page called Paradigms in the Strategy section. If there is a “mantra” in Theory of Constraints, then the above focusing process is it. Failure to use it, will mean failure to succeed.
However, there is one caveat, you don’t have to go through the sequence; identify, exploit, subordinate, elevate, every time. Some constraints can be broken at exploitation and you short-circuit the loop and start again. The various “loops” are discussed in a subpage off the Evaluating Change page called 5 Step Method and also on the Strategy page where you can be so bold as to proactively select a constraint rather than reactively identify one. But these are all things that have to be learnt – and this takes time, there is skill involved.
The one thing that we want to avoid, however, is that almost all non-systemic improvement methodologies start at the start by identifying a constraint and then jump directly to the 4th step and try to elevate the constraint by adding more of it; that is increasing production rather than increasing productivity. This always involves increased expenditure of some sort. We will never ever learn if continue to do that.
We are much “tighter” than that. We expect to ensure that we are getting the fullest potential out of our existing constraint first, and that all of the other steps have fallen into line with this and are protecting the constraint so that we don’t waste any of its valuable capacity. That is a hugely different mind-set at first for many people. Believe me.
The exploitation and subordination tactics come from the work of Taylor, Ohno/Shigeo, Deming, and a rather special understanding of the buffering of safety from Goldratt. I urge people to acquaint themselves with original work on Toyota, and the original work of Deming. A health service is also a rather special sociotechnical system but it doesn’t seem to know it.
Having a focusing mechanism with which to direct our toolsets is necessary, but this of itself is not sufficient. We need to articulate the broader context. This is what I mean, we need to;
(1) Define the system.
(2) Define the goal of the system.
(3) Define the necessary conditions.
(4) Define the fundamental measurements.
(5) Define the role of the constraints.
(6) Define the role of the non-constraints.
We were introduced to these concepts on the page on Measurements. A more healthcare specific approach was presented on the page for Replenishment and Healthcare in the Supply Chain section. There must be a strategic intent if are to know how to apply our new found tactics.
We summarized this on the page for Paradigms as follows;
This shows the relationship between the focusing process, our plan of attack, and the broader context of the environment, our rules of engagement. These are well discussed throughout the website.
Essentially we must replace the former reductionist/local optima approach of our pre-industrial psychology with the systemic/global optimum approach of our industrial reality. We must move past a fixation with detail complexity and recognize that dynamic complexity is an equally valid component. Slowly we will learn that most dynamic complexity is really quite simple – if only we would stop to listen and to learn.
We know enough now to continue with some specifics of healthcare. Let’s summarize and get on with it.
In order to improve healthcare we must approach it in a systemic manner. We must understand the overall context and then begin to look for the constraints that stop us from improving our output. Currently we are approaching the problem without the rigor of a focusing process, we are applying improvement methodologies with the very best of intent, but without focus. Indeed some of the improvement methodologies are derivatives of more systemic approaches, but we need to return to the fundamentals of the systemic approaches of Taylor, Toyota, and Deming if we are to avoid another crash and burn.
It is possible to implement rapid, significant, and sustainable improvement in this environment – healthcare. It is being done so right now. The knowledge and skills that have been developed in other environments are portable to this one. Sure there are some singular challenges in healthcare, but nothing that can’t be overcome. In the end it is about people, we create the problems ourselves, and we can fix them too.
Remember that Deming said; “The system is such that almost nobody can do his best. You have to know what to do, then do your best. Sure we need everybody’s best – everybody working together with a common aim. And knowing something about how to achieve it. Not just with what seem to be brilliant ideas, but with a system of improvement (10).”
It is just amazing that more than 20 years after those words were last aired in the West and more than 60 years after the Japanese listened, we are still feeling around in the dark. We don’t usually have a system of improvement. Who are we kidding? Surely, only ourselves.
In the next 3 pages there are three specific logistical approaches to healthcare. They cover the beginning, the middle and the end points of public service hospitalization, that is; emergency department, non-acute surgical intervention, and medical/surgical nursing and discharge. They will require that most people not do their best in most areas most of the time, and to do their best in probably just one. Both the outcome and the output will improve as a consequence. If we use our experience developed with rate limiting processes then we in fact understand this already.
Taiichi Ohno said it all in 1978 (11); “A business organization is like the human body.”
Anyway, there is lot to learn, let’s get on with it.
(1) Ohno, T., (1978) The Toyota production system: beyond large-scale production. English Translation 1988, Productivity Press, 143 pp.
(2) Shingo, S., and Robinson, A., (editor) (1990) Modern Approaches to Manufacturing Improvement: The Shingo System. Productivity Press pp 21-45.
(3) Neave, H. R., (1990) The Deming Dimension. SPC Press, pp 22-23.
(4) Hurst, D. K., (1995) Crisis and renewal: meeting the challenge of organizational change. Harvard Business School Press, pp 120-123.
(5) Jaques, E., and Cason, K., (1994) Human capability: a study of individual potential and its application. Cason Hall & Co., pg 10.
(6) Taylor, F. W., (1911) The principles of scientific management. Dover Publications reprint (1998), pg iv.
(7) Deming, W. E., (1994) The new economics: for industry, government, education. Second edition, MIT Press, pp 95-97.
(8) Phipps, B., (1999) Hitting the bottleneck. Health Management Magazine, February, pp 16-17.
(9) Goldratt, E. M., (1990) What is this thing called Theory of Constraints and how should it be implemented? North River Press, pp 3-21.
(10) Walton, M., (1986) The Deming Management Method. Perigee, pg 32.
(11) Ohno, T., (1978) The Toyota production system: beyond large-scale production. English Translation 1988, Productivity Press, pg 45.
This Webpage Copyright © 2008-2009 by Dr K. J. Youngman