>> This is Adam Wilcox, and today I'm talking about issues of clinical decision support as an introduction to the next section that we'll talk in much more detail about that. In thinking about clinical decision support, it's important to put it in the context of the overall goal of what we want to do in healthcare, and that's to make a difference, or make an impact in, in healthcare. And part of the, the considerations when you try to make a difference is that you want to maximize your efforts and make the biggest difference possible. So you need to first identify what's the important differences that can be made and also where the variation is, what is variable. Because when things that when there's a variation between different results, there's an opportunity there to by implementing information systems to consolidate, or at least standardize the, the care or the processes the delivered and so you remove that variation. And so understanding what's variable is often a really good indicator about where the opportunities are. ^M00:01:06 [ Background noise ] ^M00:01:14 >> So here we have an image of an analysis that was done in an institution, this institution had a health plan as well, it was an integrated delivery network, and they were looking at the variation in costs for certain services that were delivered, so we have the rows represent the different specialties, and each bubble represents a different type of service. Now, over to the right, that's where you'd find the areas where they'd been spending more for their institution, and over to the left it would be where the total costs were a little lower. So if one is to choose, where do you want to make the biggest difference? There's two things to consider. Number one is the total cost. But number two is the size of the bubble because the larger the bubble that represented the larger variation that was going on in that area, so if you have pregnancy with complications of any delivery type, what you see on the far right in the center, that was the most expensive area, but it was also the area where they had the largest variation. So, by standardizing the processes around that, you could narrow that variation down on one side of that area and you could make very large improvements. For example, by narrowing the variation down closer to the 75 million dollars, that's a savings of, you know, about 5 million -- think about medicine, we have to think about where the variation might be and where the issues or the opportunities may be. Well, there's really kind of five areas that we've seen recently about where the variation can be and I want to go through all of these instances. In Part 2, emphasize the last two about efficiency and quality and the relationship of healthcare reform, as areas of very important variation. In 1999, the Institute of Medicine came out with a report on medication -- or not just medication errors, but errors in medicine, called to error is human, and they followed that up in 2007 with a bigger report that focused specifically on medication errors. In '99, part of that report indicated that between 44,000 and 98,000 people were dying each year in hospitals due to errors, this is a really large number. In fact, there have been considerations that a way to emphasize this number is to consider how many 747's crashing does this really indicate, you know, is this equivalent to because, you know, just to emphasize that almost 100,000 people are dying each year. Now, the question is, how large is the variation and is this the issue that is most important to tackle in terms of healthcare reform. Well, if you solved every one of these errors, if you avoided, if you put in a system that avoided every single error, 100,000 errors each year, bear in mind that this a pretty severe error because it leads to death, but what does that really change in terms of the healthcare system? Well, the total population of the United States is around 300 million, total hospitalizations per year is 38 million, so the actual influence of fixing this problem is about a fourth of a percent of all hospitalizations, so even though it came out as a very important result, and, and, and considered a very large number, it really only represents about a fourth of a percent of all medications. Now, for those 98,000 people, it's a very significant event clear. But just changing, changing errors only as a solution to the healthcare problem that we're addressing is not going to be sufficient, and we understand that because we look at the magnitude of the problem and then the variation. Another unit where there's, where there's considerations of variation is in practice patterns, this is from the Dartmouth Medicare Atlas, which is an analysis that was done about local variations or regional variations in healthcare delivery. So, they, you know, the blue represents the highest -- higher cost areas, and the green represents the lower cost areas. And if you look at the numbers, there's almost a two fold, two to three fold variation in terms of costs in different areas that is not associated with the improved quality. So, we look at that and, you know, we recognize that there's some areas, areas of the country it's just more expensive, and, and life is more expensive, but even when we count for differences in the prices paid for similar services, and account for differences due to illness, there still remain two fold difference. In other words, the differences in spending are almost, almost entirely explained by the differences in the volume of healthcare services that are received by similar patients without actually improving their quality of care. Personalized medicine. Personalized medicine also represents an area where there's an opportunity to reduce variation in healthcare and to make improvements, so what does that really mean? Well, personalized medicine means we customize the treatments according to what the genetic characteristics of the patient, so imagine you have a population of patients and there's one group of patients for whom a certain medication that would be otherwise effective in other in, in that population really just doesn't work. In fact, it could be -- cause toxicity, so if we can identify the people who should not be taking this medication that we normally prescribe then we can prescribe a different drug in that population, so, and you know that means that we customize the medications that are given to the genetic makeup and the ability of their body to respond to that. So, you know, and that has a real good opportunity because it identifies where the variation is and then it has the specifics. Unfortunately there has been recent articles that have focused on the value of personalization and have shown that in reality, while there has been some promising first results that it seems to be limited value over time that the first results seem to be the low hanging fruit and it's much more difficult to get greater value, that's what this slide on -- the image on the right shows that with Type II diabetes, you know, for the first genes, they were able to find a lot of potential for reducing genetic variation, but it's not a linear relationship. Over time, those results seem to be really decreasing. Efficiency is another area where there's been, where this is variation, so here we have two images just two graphs showing the average spending U.S. health per capita. First of all, the United States is more than twice as high in terms of spending compared with the other countries. One of the arguments is is the United States totaled expenditures and wealth is much higher and so there's a reason it may be higher just in terms of that, and so if it's, if you account for that, and you just do total expenditures on healthcare as a percentage of GDP, which is the second slide, the variation reduces some, but you still see a lot. You know, 14% versus, you know, what seems to be an average around 9%, so there's, there's not a 50% variation possible in terms of efficiency, meaning that if the United State's efficiency in healthcare spending was equivalent to other industrialized countries, then, then you'd have about 50% of the efficiency that could be reduced, or 50% of the spending that could be reduced. And if 50% of efficiency could be improved. And quality is another area. Here we show potential years of life lost to the diseases of the circulatory system. Again, the United States, we're number one, but not in a good way. The industrialized country median is 550 potential years of life lost, whereas in the United States is at 125, so even with this increased spending per person and as a percentage of GDP, we still have a 50% variation just in terms of quality in the United States. Here's another table from the Commonwealth Fund showing the United States ranked relative to other countries in a lot of different characteristics that -- so, the United States, you know, on the far right side on all of these issues, almost all of these issues in terms of quality of care, access efficiency, equity, and long healthy and productive lives, we rank 5 or 6 out of 6, in terms of overall spending. We're twice that of the other countries. Canada, just to the north of us, they're 5 or 6 as well on every -- almost everything except for long healthy and productive lives where they seem to be doing a little better. What's interesting though is I said almost all of them, there is one area where we're not in 5 or 6, and [inaudible] 4, we could kind of just assume that the United States happened to, to do a little better in that area, we're 3 even. With Canada, that long healthy and productive lives, it seems we may be able to explain that away with other factors, but for the United States, the one where we're not ranked the lowest is we're ranked number one, which seems interesting, and if you look across, what that characteristic is, where we're ranked at the highest is right care, meaning that when we finally get the patient with the physician and evaluate what type of care is actually delivered in that certain, in that space, the United States is the best healthcare system in the world. So you step back and you say, okay, what, what does this really indicate is going on? Is the problem with the clinicians in this healthcare system? No, because when we get the clinicians interfacing with the patients, we're ranked number one. The problem seems to be more systemic, and seems to be a problem more of the system itself that it doesn't facilitate -- well it, well, right care can be delivered, it doesn't facilitate within that environment, save care, or coordinated care, or patient centered care acts as sufficiency equity, all of those things that we don't have structures to specifically ensure. But the clinicians, at least, are doing a good job in, in the care that they're delivering. This is the healthcare system in the United States, and, and the challenges that we have in front of us. So, we could ask, how does informatics really relate to all of this? And how does informatics relate to this overall discussion that I just went through about efficiency? Well, first of all is that if we want to make a difference, we need to understand where the differences can be made, and the size of the difference that can be made is really the context by which our contribution should be judged. For example, medication errors, there have been studies that have shown that physician computer order entry has led to up to an 81% reduction in medication errors, which is a pretty impressive improvement. However, as I mentioned earlier, that doesn't solve the whole system, just fixing medication errors, while it's important for each of those errors, 'cause each of those errors could be significant, it -- that is not the problem we have with our system. What we need is actual transformation, and when we look at how informatics tools could be used, we need to consider how informatics tools could be really used to transform care. ^M00:13:47 [ Background noise ] ^M00:13:52 >> Now, I'm going to take you through one more discussion, kind of a historical discussion about how medicine has been changing over the past couple of decades, and informatics has been changing in relation to that. The reason is is because we want to understand all of the different contributions of, of informatics. In the context of what the underlying issues were in medicine so we know which things need to be carried forward, and which things maybe are relevant more to a different scenario than what we're dealing with now. So pre 1990, medicine was really -- had a lot of freedom to pursue the scientific aspects of medicine, there was a drive for deeper understanding in health care, there were a lot of randomized controlled trials, there still are a lot of randomized control trials. But before that the concept of cost wasn't that -- wasn't as significant as it, as it is now. The main driver behind a lot of informatics, or a lot of medical research was does it make some form of improvement, and there was a push this time in, in terms of informatics to support this drive for deeper understanding, and a lot of research in artificial intelligence. Now, with the explosion of research and lots more publications, it was for the first time where the publications became unmanageable for the human mind, and so there was need for bibliographic systems to manage all of this information, and then bibliographic retrieval systems, this led to some of the first issues with vocabulary, with medical subject headings, and med line, that vocabulary concept was relevant later as EHR's were more developed, but in bibliographic retrieval systems they were very important. And then there were some issues with clinical invent monitors, for the first time that we considered how computer may be able to monitor things that the clinicians weren't doing directly and improve their level of attentiveness on, on specific issues that, that had to be followed. And then there was also research done around diagnostic [inaudible] support systems, this is related to artificial intelligence, but we did find some challenges with the data that needed to be used for that artificial intelligence, and there were data completion issues that were discussed. In the 1990's, medicine was seen more as a service and not just as, as a science. For the first time in the early '90's we had the considerations around healthcare reform, and medicine as it went into this current path, may be potentially unsustainable, not to the degree that we're seeing now, but there was this first consideration. A lot of this came from the Dartmouth Medicare Atlas where they saw this, this variation in medical care practice, and this led to research in patient outcomes research teams that were began by the precursor to the AHRQ in 1988. In informatics, the result of this was figuring out ways to support the outcome of the patient outcomes research teams, which were a practice guidelines, and how these guidelines could be more automated, so for the first time, the concept of guidelines were real. At the time it was criticized as cookbook medicine, but rules where, whereby medicine could be practiced in a more standard fashion were created, and there needed to be support for how do you, how do you automate these guidelines and a way to make it easier to follow, and easier to do the right thing. Then in the early 2000's, based on the report that I talked about earlier from the Institute of Medicine on To Err is Human, and then the later report from the same group in 2007 on specific type of errors, medication errors. Medical errors became a big focus in medicine, and the first place that informatics really had an influence was on drug prescription assistance, so there was a lot of work in developing these standard drug interaction databases that could be used by systems to say whether or not, you know, as you enter a drug on one patient, hey, this -- there could be a direct interaction. And this was, this was important because it was addressing specifically actions by clinicians that could be dangerous. And then there was a lot of modeling about alerts and how alerts should be delivered, and how the responses should be measured, so that's one of the informatics contributions were at this time. Here we have an example of the drug, drug interaction display, I think most people have seen these from some demo or, or experience with a clinical information system, but the patient is just being prescribed aspirin, unfortunately -- well, for whatever reason, fortunately or unfortunately, it doesn't really matter, it was a demo patient anyway, but it had already been listed on this patient that they had Coumadin, and so that leads an interaction alert saying that please reconsider this prescription for aspirin, or at least acknowledge that you've already considered this, that the patient is also on Coumadin, there could be a severe drug interaction between these two medications in part because they, some of the things they do are the same and they could lead to the patient bleeding to death internally. Then in the mid 2000's, we had the rise of personalized medicine and genomic medicine, this was in part due to the breakthroughs of the human genome project when we were finally able to map the entire human genome, and the excitement around that, that led to the new [inaudible] road map from [inaudible] focused a lot on discoveries in genomic medicine and personalized medicine. It seemed almost that the informatics support at the time went back to that medicine is a science, and, and the discovery of components of the pre 1990's where informatics tools were around trial recruitment and data analysis. Trial recruitment such as adaptations of clinical event monitors to alert that certain patients may quality for [inaudible] trials, and data analysis in modeling the genetic information and how that could be accessed. With trial recruitment, there was also consideration about what data were available and how they could be used to monitor. And for the first time data warehouses then would search for information on patients prior to them coming in, and, and that may be relevant for their care, but just in terms of their presence for trial recruitment was used. And then in the later 2000's, though much of the 2000's was leading to with the Institute of Medicine Report on crossing the quality chasm, there was the recognition of the problem with the U.S. healthcare quality and costs, and I've -- as I've already discussed in terms of that efficiency at the lower quality, and this is part of the overall drive around healthcare reform. Medicare and Medicaid, as we see they're on a path to insolvency without some type of reform, that's led to some significant actions at the federal level. In terms of informatics involved in this, there have been informatics applications around preventive care reminders, specifically because it's an opportunity for some level of transformation that recognizing the, the healthcare system is that it's best when the clinicians are in front of the patient in delivering the right care. What may be helpful is getting the patients to that point where the right care can be delivered before they need more serious care, and so there's been some efforts around preventive care, some very positive efforts around preventive care and informatics. Challenges around that are with data acquisitions and modes of alerting, there has been work done according to that. ^M00:22:28