Portal de Periódicos da CAPES

Keynote Address—The Innovator's Prescription: An Examination of the Future of Health Care Through the Lenses of Disruptive Innovation

2009; American Medical Association; Volume: 133; Issue: 4 Linguagem: Inglês

10.5858/133.4.513

1543-2165

Jason Hwang,

Biotechnology and Related Fields

I want to give credit where it is due and mention that during the past 3 years I have had the privilege of working with Professor Clayton Christensen at the Harvard Business School (Boston, Massachusetts), and his ideas and groundbreaking research are really the foundation of many of the concepts that I am going to present to you.The whole idea of disruptive innovation comes from the world of business, and it is rooted in the observation that if you were to go through and scan through the annals of business history, most of the companies that were at one time considered to be unassailably successful, within a decade or two, found themselves running in the middle of the pack, or worse, at the bottom of the heap. And this observation really did not make sense at first because at one time when these companies were high flyers, all of the trade magazines and leaders of the industry commended the management of these companies, talking about how brilliant they were in leading these companies to the top of their business, and yet a decade or two later when these same managers were in charge, they were criticized for running their companies into the ground. It did not make sense that it happened with such frequency and regularity that all of a sudden managers could suddenly become stupid and turn their companies to dust. The observation was made in Professor Christensen's initial research that it was actually the principles of good management that were being taught at Harvard and other business schools around the country and the world that were causing these successful businesses to fail. The initial message that came out of the first book, The Innovator's Dilemma, was that if you somehow became wildly successful by doing everything right, you were doomed.1 (Unfortunately, they did not tell me this until I paid 2 years' worth of tuition to Harvard Business School.)Nevertheless, the model of disruptive innovation is actually quite simple, so here I have graphed performance of a product or service over time (Figure). In any given market, you have a trajectory that is reflected here by the red dotted line that represents the rate of improvement in the performance of that product or service that customers can possibly consume or use. And, of course, in every market there are tiers of customers, and that distribution includes the very high end at the top, where you have the most demanding customers who are never satisfied with the best products that companies can deliver to them, and, at the low end, you have people with simpler problems who can be overserved by products and services that are actually very easy to produce.The second trajectory in every market is reflected here by the blue solid line that represents the rate of performance improvement that is delivered by innovative companies that continue to introduce better and better products every year. Sometimes these leaps in innovative progress can be small and incremental; other times they can be breakthrough innovations, for example, from the imaging industry, where we went from x-rays to computed tomography to magnetic resonance imaging to positron emission tomography scanning. Each of those innovations represented billions of dollars worth of technologic progress. What is important to note about this model is that the pace of technologic improvement that comes from these innovative companies almost always outstrips the ability of customers to use that performance, and that is going to have dramatic ramifications as we will soon see.Whether the performance improvement is just a little incremental improvement or if it is a big leap forward in technology, it does not really matter because they all have the same effect of sustaining the existing trajectory of performance improvement, and it does not help in predicting who will ultimately win and be the leaders of the industry. In situations in which you are talking about sustaining innovation and that type of performance improvement, almost always the incumbents will win.In our analysis of other industries, what we noticed was every once in awhile you would have something called a disruptive innovation that came into play. We call it disruptive not because it was groundbreaking or evolutionary but because, rather than sustaining the existing trajectory of performance improvement, it actually was a product that was not as good as what was being already sold by the leaders of the industry. But because it was simpler and more affordable, it could take root in a less demanding tier of the market and, with that as a foothold, start to improve itself and move up-market and start to steal market share away from the sustaining innovator. Whenever there was a case of disruptive innovation being introduced into a marketplace, almost always it was the entrants that came in and killed off the leader.Why is it that there is such an extreme difference in the odds of success when you are talking about sustaining versus disruptive innovation? I am going to use an example from the steel industry to illustrate why—because it has nothing to do with health care, but then when you hear it, you'll realize that it actually has everything to do with health care.For those of you who have not had the chance to make a lot of steel in your lives, there are actually 2 ways of making steel. First of all, historically most of the world's steel has been made in these gigantic, integrated mills, and it costs about $8 billion to build one of these. But there is a second way that came out in the 1960s called the mini-mill, and in a mini-mill you basically melt down scrap in an electric furnace. And because you can get that molten steel in such a small space, it allows you to scale down all the other steps of the steel-making process, and you could probably fit 2 mini-mills in this very room—and hence they call it a mini-mill.The important thing to note about mini-mills is that it actually allows you to produce steel of any quality for 20% less than it would cost you to make that very same product in an integrated mill. If you were the chief executive officer of a steel company, even in a great year your profits were probably about 2% of your overall sales, and here you have this new technology in the form of mini-mills that would allow you to reduce your full cost of production by 20%. Don't you think you would adopt that technology? And yet not a single integrated mill has ever successfully built and operated a mini-mill. So why is it that something that seems to make so much sense be so difficult for a leader of a company or leader of an industry to implement?Let me describe the picture of the steel industry as we begin this story. Like in every other market, you have different tiers of customers. So at the low end you have the very simple products that are easy to make, and in the steel industry that is concrete reinforcing bar, also known as rebar. At the very high end, you have sheet steel, which is used to make automobiles and appliances—very hard to make of sufficient quality and therefore very profitable. At the beginning of the story, the integrated mills are making this entire product line and serving each of the tiers of the market.At the beginning of the 1960s, the mini-mill technology finally became viable, but because these mini-mills were essentially melting scrap in these electric furnaces, the quality of steel that they produced was just not very good. The only companies that would buy from these initial mini-mills were the ones that were looking for rebar, primarily because there are almost no specs for rebar and once you buried it in concrete nobody could ever verify whether they met specification. And so it was actually the perfect market for pretty crummy products, at least initially.As the mini-mills began to attack the rebar market, the reaction of the integrated mills was man were they glad to get out of that market, because it was just a dog-eat-dog commodity, and the margins that they made in that market were only 7%, and it only represented 4% of the overall tonnage of their production, and why would you ever invest to defend a lower profit-making part of your product line when you had the opportunity to strengthen your share of angle iron for which you made profit margins of 12%. The integrated mills basically shut down their lines that were making rebar and reconfigured them to start making more angle iron. A funny thing happened on their financial statements, because as these integrated mills were lopping off the low-margin products of their portfolio, their profit margins went up, and Wall Street applauded them for it. And so it felt great for them to get out of the rebar business, and meanwhile, because the mini-mills had a 20% cost advantage, it felt great for them to get in, because they were making loads of money. And things were kind of at peace until about 1979 when the mini-mills succeeded in driving out the very last high-cost integrated mill out of the rebar business. When that happened, all of a sudden, the prices of rebar collapsed by about 20%. And why did that happen? It is a funny thing about strategy that a low-cost strategy only works when there is a high-cost competitor in the marketplace. And once they drove out all the high-cost integrated mills, nothing was left except low-cost mini-mill duking it out against low-cost mini-mill in a commodity business. That competition drove prices down to the point where nobody could make any money.So what are these poor mini-mills going to do? They had no choice but to move up-market, where they saw the 12% margins being made in angle iron, and they basically figured out that they had no choice but to stretch their capabilities from making rebar to start making angle iron. When the mini-mills began attacking the angle iron market, the integrated mills' reaction was man were they glad to get out of that market, because it was just a dog-eat-dog commodity and why would they ever invest to defend a market in which they only made 12% gross margins, when they could shut down their lines and reconfigure them in structural steel for which they can make even higher margins. Once again, the financial statements that they reported to Wall Street made it look like a great decision, because their profit margins were up, because they were focusing in investments for which the profits were most attractive, and meanwhile the mini-mills were happy because of their 20% cost advantage and their profitability had returned and so they were happy. And once again things were kind of at peace until 1984, and that is when the mini-mills succeeded in driving out the very last high-cost integrated mill out of angle iron, and then all of a sudden, once again, the prices collapsed by 20%. So what are the poor mini-mills going to do? Once again, the reward for success was they couldn't make any money.If you go back to the issues of Iron Age in the 1980s, all of the experts in the industry could see these barbarians at the gate, but all of the experts were convinced that there was no way to ever roll structural steel (these big beams) within the confines of a small mini-mill, and you just could not get the properties necessary to meet spec through the smaller mini-mill process, because those properties could only be imparted by the 7-step rolling process of the integrated mills. But what these technical experts failed to realize was just how desperate these mini-mills were, because once again they had no choice; they had to figure out a way to move up-market to survive. The types of innovations that mini-mills were able to apply to start moving into just the very low end of structural steel, stretching their capabilities from making angle iron into just little 4-inch I-beams that were used to support mobile homes and things like that, were just extraordinarily clever. As the mini-mills began to attack this low end of the structural steel market, the reaction of the integrated mills was man were they glad to get out of that market.It did not take very much longer before the mini-mills were looking at sheet steel. Today, 60% of the market in steel is controlled by the mini-mills, and all but one of the integrated mills has gone bankrupt. You can see the extreme difference in the results of what happens to the leaders of an industry versus the disruptive entrants when one of these disruptive innovations is in play.The moral of the story is basically, whenever you are dealing with a sustaining innovation, if the question is who is going to lead the way in developing better and better products that can be sold for higher profits to the best customers in an industry, I will always bet on the leaders of an industry. On the other hand (and this is important for health care), if the issue is who is going to come up with more simple, affordable solutions, I will always bet on a new entrant to come in and kill off the leaders of the industry.How is it that the odds of success are so different? If you think about it from the perspective of say you are David fighting off Goliath: If you try to fight Goliath on his own turf and, in the language of sustaining innovation, if you try to come up with a better product that can be sold for better profits to the giant's best customers, the giant is going to kill you, because the giant has all the resources at its disposal and is motivated to win. If you are the disruptive entrant, and again you are David, and you decide to pick a fight at the low end of the market, you've created a situation actually in which the giant is motivated by its own self-interest to flee rather than fight. And that explains why there is such a difference in the odds of success.As we began to apply these ideas to health care, one of the questions we were often asked is, given that health care continues to introduce advanced technologies and sophisticated equipment each and every year, how is it possible that health care has not been disrupted to any degree already? How is it possible that health care only seems to get more expensive, less convenient, and less accessible and is going in the opposite direction of what we might predict based on our observation of business history? To better answer that question and to explain that phenomenon, we decided to go back and break down disruptive innovation into 3 discrete steps, and we have called these the enablers of disruption. And really for successful disruptive innovation to occur, all 3 of these steps really need to happen.The first step is something we call a technologic enabler, which is the technology that makes a product or service simpler and more affordable. But what is missing in health care very often is that second step, which is the business model innovation in which you embed that technology into a business model that can take that technology in a cost-effective way into the marketplace. And this is where health care often struggles.Third, you need a commercial system to coalesce around a disruptive business model, to support it as it goes to market, because very rarely does a disruption actually happen at just a single point in the whole value chain of an industry, and really what we usually see is an entire value chain disrupting the one from the past.First, on the technologic enabler, this is an area in which pathology has tremendous influence. I am going to use a case from outside health care to kind of explain where I am going.In the early days of literally every industry, the technical problems of that industry can only be solved through a process of experimentation and trial-and-error problem-solving. An easy way to visualize this is if you think back 70 years, when you needed a new molecule from which to build an organic fiber, the only way you could do it was through this type of problem-solving, and there were only maybe 50 people in the entire world who could make one of these new molecules, and DuPont (Wilmington, Delaware) employed almost all of them. The way they would create one of these new molecules was they would mix a few atoms in a beaker and then the scientists would stir it together, heat it up, draw out a fiber, look at it under a microscope, take it down the hallway, knock on a colleague's door, and ask him, “What do you think we've got here?” The colleague would say, “I'm not sure, but let's heat it up for another 10 minutes and see what happens.” The miracle molecules we know today as polyester, nylon, Kevlar, and acetate all came from the labs of DuPont through exactly this type of trial-and-error process. But in this type of work you have to rely on such costly expertise and just a very few individuals who have the experience to be able to carry this out, and it required just the deepest intuition in organic chemistry, and that is why nobody could play that game like DuPont. But as DuPont scientists practiced this art over and over again, patterns began to emerge. You started to move into this pattern-recognition realm in which there was no cookbook yet, but you could start to teach other scientists how to do some of this work. So you could teach a new scientist, always start here, never there. If this happens, then do this, but never that. You still had to be reasonably smart, but you no longer had to be the best in the world to play this game, and a lot more companies were able to enter the industry in addition to DuPont.Eventually quantum theory came to be better understood and applied. What quantum theory allowed you to do was to essentially predict in advance that if you had this given structure of a molecule then these will be its properties. And that understanding of cause and effect basically allowed you to write software that would allow a user to type in the properties that were desired, and the software program would reverse-engineer the structure of the molecule that came closest to delivering those properties. Professor Christensen likes to joke that all you need today is a BS in chemical engineering from your average state college and a great piece of software and you can design better molecules than could the world's best scientists just a generation ago.If you were to stand up and turn around 360° you could lay your eyes or your fingers on perhaps 20 or 30 different fibers that have just been an unmitigated blessing to mankind in their appearance, their functionality, their durability, and their cost, but that blessing did not come about by somehow replicating the costly experience and expertise of DuPont scientists, but rather it came from commoditizing their expertise through technologic progress. You can see where I am going. What we need are similar technologies in health care that actually commoditize physician expertise so that, little by little, it is actually not needed any more.In health care, once we started to apply these ideas (and forgive me for this oversimplification because we often talk to nonclinicians), we described the human body as having a very limited vocabulary from which it can draw on to express when something is wrong. This vocabulary basically consists of its gross physical symptoms, and there just are not enough symptoms to go around for all the diseases that there are, as you all know looking at specimens under the microscope, so all the diseases basically had to get together and agree to share symptoms.In that kind of world, you may treat 1 patient with a given therapy for a symptom and he or she will respond, but if you apply that very same treatment to somebody with the very same symptom, he or she may not. In that type of uncertain world, the practice of medicine really requires a deep intuition of medicine and physician expertise, much like the work at DuPont required deep expertise and intuition in the world of organic chemistry. We call this type of medicine “intuitive medicine,” precisely because it relies on the intuition of physicians to practice. But as physicians practice their art over and over again, patterns began to emerge, and we move into an area of medicine that we call “empirical medicine.” Some of you may know it as best practices, other people call it evidence-based medicine, but basically in that type of world when we see something that works more often than other possible treatments, then we choose to adopt that technology.There is another step forward in which we move into precision medicine, and I know we are going to have a session later about this as well, but basically the new technologies that are coming out, specifically things such as molecular diagnostics and new imaging techniques, are allowing us to precisely diagnose diseases and therefore come up with predictably effective therapies for the treatment of these precisely defined diseases. An example that you are all aware of is cancers; in the past we defined them based on symptoms and ended up with geographical definitions of disease. We have lung cancer, brain cancer, and breast cancer, and now with these new tools we are realizing that within each of these given geographies, there is actually a multitude of cancers, different types, different histologies, different pathologies, each with potentially unique treatments that could have a dramatically improved success rate.We are discovering with these tools that at the level of our genes and at the molecular level that our human bodies are actually very precise and very articulate in the way that they express themselves when something is wrong. It is technologic enablers, like the ones that you are going to hear about later in this conference, that are enabling health care to move down this cascade. When you get to the point where things are completely rules-based, which is when physician expertise is truly commoditized, these technologies essentially allow nonphysicians, people with less skill and less training, to do more and more sophisticated things because the work no longer requires the intuition of physicians.One of the big problems in health care is that we do not have the right business models when we come up with these great technologic enablers. In fact, as an example, when I was walking through the exhibits last night, very often I saw digital images being produced, and I am willing to bet that most of the vendors that try to sell digital imaging to pathologists are talking and selling the point that it helps you do your job better. Although there's nothing wrong with that, that is a sustaining application of that technology. What would be disruptive is if somehow you could follow the model, for example, that is used in radiology, and take advantage of the portability of digital imaging and send it to centers for reading. That is one potential business model application that might be very different from the way that we traditionally think about how to apply digital imaging technology in pathology.Before I get more into that, let me just talk in general terms about what composes a business model. Every successful business model starts with a value proposition. Basically, that is the product or service that helps customers do something that they are trying to get done. To deliver that value proposition, a business will put into place a set of resources, and that is basically composed of people, equipment, cash, and so forth to perform that value proposition. After doing this several times, over and over again, a set of processes will coalesce, and processes basically just define the ways of working together to consistently deliver that value proposition.Very quickly, a profit formula will emerge. Basically, it says given the resources and processes that we have in place to deliver this value proposition, what sort of margins do we need to set to support those costs? How fast do we need to turn over our assets and things like that? That profit formula very soon starts to dictate which value propositions a company can and cannot take on. This goes back to this initial disruptive innovation model that I told you about and why it is so difficult for a sustaining innovator, a successful incumbent leader, to do things that are disruptive—because these arrows become bidirectional, and the process that I just described begins to work in reverse. And you can see that something that is sustaining, in which you are able to create a better and better product for higher profit margins to your best customers, can easily fit into this business model, but anything that falls out of it, something that perhaps is not as good, that does not target your best customers, that makes lower margins, just cannot make its way through a company and get approved within an established, successful business model.In our study of innovation, we noticed that there are 3 types of business models. The first one we call a solution shop, and representative of solution shops are consulting firms such as McKinsey (New York, New York) and high-end law firms, and basically these are businesses in which we pay professionals a lot of money to diagnose what is wrong with our company or situation and basically recommend solutions for us. Research and development organizations do this, high-end marketing firms will do this, and, interestingly, the diagnostic activities of hospitals are solution shop activities, for which essentially we are paid to figure out what is wrong with a patient and to recommend a treatment based on that diagnosis.What is interesting about solution shop work is that it almost always has to be paid on a fee-for-service basis. Every now and then you will have a company such as Bain Consulting (Boston, Massachusetts) come in and say … we will take a part of our compensation based on the effectiveness and value of our outcomes. But it never works, because the outcomes are just dependent on so many factors out of their control, and they are so unpredictable that they always end up reverting to fee for service.There is a second type of business model that we have seen, and that is called a value-adding process business. Representative of this model are things such as manufacturing, in which you take something that is incomplete into your facility, you do stuff to it, and then you ship it out in a more complete form: auto manufacturing, in which you bring in steel, you do stuff to it, and you ship cars out; food services, in which you bring in raw ingredients, you do stuff to it, and you ship prepared meals out. In health care, one prime example of this type of activity is the procedures that we do after definitive diagnoses have been made. You can probably already think of ambulatory surgical centers, very focused specialty centers such as Shouldice Hospital (Toronto, Ontario) in Canada that does nothing but repair hernias, and Aravind Eye Hospital (Madurai, India) in India. These are all examples of this value-adding process business type of model.The interesting thing about value-adding process businesses is that you can pay these guys by fee for outcome. In fact, most of these businesses can offer guarantees of their quality, or else they will give you your money back or repair it under warranty—and when has a hospital ever declared a warranty on their repair? The interesting thing is that for solution shops the reason that hospitals cannot do that is the work of solution shops is very highly dependent on the people, the professionals that are delivering care at the hospitals. To compete in that kind of world, the only way to deliver more value to the customer is to hire the best talent available. In value-adding process businesses, the value that you deliver to customers is really inherent in the processes of the business and not the people, and because your outcomes are not as people-dependent, you can make it easier for a business to guarantee your outcomes.The third type of business model is what we call facilitated user networks, and these are underrepresented in health care. But we have seen them elsewhere, and examples would include telecommunications. These are businesses in which essentially users of the network trade with one another: in telecommunications, I send data to you, you send data to me; in insurance, I put premiums into a pool and I take claims out of it. These are examples of user networks, and in these sorts of businesses, the company that usually makes the money is the one that facilitates the transactions. So a familiar example in today's world is eBay (San Jose, California).In health care, there are some examples in which there may not be a lot of profit being made, but Alcoholics Anonymous (New York, New York) is an example of a network in which users of that network share information or protocols with one another. Weight Watchers (New York, New York) is a long-standing example of a user network, and one of the newer ones I have up here is a company called dLife (Westport, Connecticut), which is basically a network for diabetic patients and their families to share information with each other.It is our belief that these user networks are really going to be an ideal business model for treating chronic diseases. Today the cost of treating chronic diseases is so high because we are trying to fit them into the business models of hospitals and physician practices that simply were not set up to do that well, whereas user networks are ideally suited for that type of exchange and support network necessary for lifelong diseases.Knowing these 3 types of business models and what is unique to each of them helped us realize that today's hospitals are really mixtures of all 3 of these types of business models, but primarily of the first 2. And as a result of their attempt to try to be multiple business models, they end up being very costly and very inefficient at delivering a growing list of conditions. Given the type of technologic enablers that are coming down the road, what we would envision for a new health care system is for hosp