Blackford Middleton, MD, MPH, MSc
Director Clinical Informatics Research & Development,
Chairman, Center for Information Technology Leadership,
Professor of Medicine, Harvard Medical School
Hour Long Video - Please be Patient
Dr. Eric Fishman:Hello, this is Dr. Eric Fishman speaking to you from EHRTV and today we’ll be speaking with Dr. Blackford Middleton. Dr. Middleton is clearly one of the world’s experts in health information technology. His current position is Director of Clinical Informatics Research and Development for Partners HealthCare. However, he also holds a number of other titles. He is past chairman of HIMSS; he is currently on the board of American Medical Informatics Association and many other organizations. Dr. Middleton, tell us a bit about your experience at Partners at this point in time if you would.
Dr. Blackford Middleton: Good morning, thank you Eric. It’s a pleasure to be here. At Partners HealthCare in Boston, Mass. my work focuses on the implementation of clinical information systems for the physicians and nurses and patients at our teaching hospitals; the Brigham and Women’s Hospital, Massachusetts General Hospital and nine other hospitals in our network, our growing network of Partners HealthCare. Partners HealthCare is the partner entity, the parent company, for Brigham and Women’s Hospital and Mass. General. And itself is approximately a $7 billion system. We have about two million outpatient visits per year, about 200,000 hospitalizations per year. And we have a history of building our own software for our hospitals and clinics. Not everything, lots of the ancillary departmental systems are vendor supplied. We have the good fortune and the fun, usually, of making our own electronic medical record system –
Dr. Eric Fishman: That would be the Partners Longitudinal Medical Record?
Dr. Blackford Middleton: Correct, our LMR. And the patient portal, we call patient gateway, as well as the infrastructure and informatics to support clinical decision making across those applications and some others.
Dr. Eric Fishman: I understand that you have in your 80 people in your department at this point in time working in this area.
Dr. Blackford Middleton: Correct. The whole IS group is led by John Glasser, the CIO for Partners HealthCare, and he’s been very supportive of all of our informatics work over the years as well as the leadership of Partners HealthCare, Jim Mongan and all of the clinical leadership. We’ve been very lucky. In many ways I feel like a kid in a candy shop because we have a lot of support to do what we’re trying to do in improving patient care, reducing medical error, improving patient safety and doing so by building our own technologies.
Dr. Eric Fishman: Now Partners HealthCare obviously an affiliated group of physicians that are largely patient-oriented. But I’ve read your writings and you seem to know an awful lot about the ambulatory side as well. Tell us about your experience on the ambulatory side.
Dr. Blackford Middleton: Sure. Well I trained in medicine and I’ve always practiced until just this past summer. I’ve got too much going on so I let it go. But I’ve trained in internal medicine, practiced in out-patient care myself for 23 years. And at Stanford when I finished training I became the Medical Director of IS there. And we had both in-patient and out-patient systems being developed and vendor purchased. Then I spent many years, six years, with Mark Levitt at Medical Logic, now part of GE, in building the logician EMR package, now GE Office Centricity. And that’s probably where I focused first and focused most on ambulatory care, clinical information management. The GE EMR, logician, or now GE Office Centricity, is basically a full comprehensive application for ambulatory care. I don’t have anything to do with it now, it’s been under GE’s control since 2001, but we really had a lot of fun building that product through the end of the 90s up to 2000.
Dr. Eric Fishman: You learn a lot building an EHR as both of us know at this point in time. Now we’re all familiar with the high-tech apple, we’re all familiar with the $44,000 apple that’s hanging out there ready to be eaten. Tell us what you would suggest a physician should do at this point in time to be prepared for taking advantage of that.
Dr. Blackford Middleton: It’s an excellent question. Most important I think is that physicians just have to be aware of what’s going on. Read the newspaper; follow the RL legislation, the American Recovery and Reinvestment Act. And talk to their colleagues, their professional societies, the groups in the professional societies who are involved in Washington. Every society has somebody paying attention to advocacy in Washington and helping to digest the various bills and proposals. But bottom line we’re very fortunate. President Obama, and his administration, really has taken a keen interest in healthcare information technology and healthcare reform overall. In many ways Barrack Obama I think sees this as a prelude to healthcare reform. If we don’t have the right kind of instrumentation and automation and support for clinical practice in place we’re never going to get to healthcare reform. And if we don’t get to healthcare reform frankly, in my own opinion, we’re going to see economical troubles the likes of which we’ve never seen before in this country potentially.
Dr. Eric Fishman: Sixteen, 17% of GDP is certainly more than we can afford and it’s not going down without some changes.
Dr. Blackford Middleton: Correct.
Dr. Eric Fishman: And so I understand that in your activities, CDS or clinical decision support is one of the main areas of interest of yours. Tell us about that.
Dr. Blackford Middleton: Sure. You know it’s interesting, as a practitioner I know how it feels to be overwhelmed by the amount of data I’m expected to have access to and familiarity with for all my patients in outpatient care. Seeing a patient in 12 or 15 minutes with four of five complicated problems in an academic medical center or in a community-based practice is simply becoming extremely difficult. I’ve felt all my life in 20 years of being in this business that we should look to the computer to help us manage the information and support clinical practice, clinical workflow and even support clinical decisions so that the doc can focus on the patient encounter. Some people say the computer actually interferes with the doctor-patient relationship. In many ways, in my own practice, I’ve always used the computer to support and enhance the doctor-patient relationship. I’ll turn the screen when writing a new prescription and show the patient, “Well gosh, I’ve done a drug interaction check for your seventh med on top of your six other medications. And there is no warning. There is no alert. This medication should be safe for you.” And the patient will say, “Well gosh, what a smart doctor.” And of course it’s not me; it’s the computer that I’m using on behalf of the patient. So I think what we’re trying to do now with the computer is to help the physician access all the information that needs to be accessed for each and every clinical encounter at the time of that clinical encounter. We know for example that in careful studies of information needs physicians often times don’t have access to all the information they may need to have. Paul Tang did one very interesting study in a West Coast academic medical center where he found 84% of the time one or more pieces of information was missing relevant to that clinical encounter. So what do we do? We invite the patient to come back. We have an extra visit because we have to gather all the information or even find the chart. Or we repeat tests and procedures because we haven’t found the results that we thought were there, they’re not findable or they can’t be located. So that kind of decision support is very simple. Making sure that all the relevant information is available to the doc whenever he or she needs it to make a decision, to care for the patient.
Dr. Eric Fishman: And that’s over and above the statistics that show that in a significant percentage of office visits the physician doesn’t know something. Not with respect to just capturing the data but doesn’t have the knowledge, and if you would address that.
Dr. Blackford Middleton: That’s excellent, thank you for the lead-in. The second part, after information access, is really about knowledge access. And you’re right, careful studies have shown – maybe the most famous recent study is Elizabeth McGlynn 2003 in the New England Journal of Medicine – reported that looking at physician practices all across the country, surveying office practice. She found that only 54.6% of the time are we applying the best evidence that we know to each and every clinical encounter. So that’s a major problem. I don’t fault the doctors, I fault the system. The system that doesn’t make that best evidence available to us while we’re making these decisions and makes it easy to do the right thing.
Dr. Eric Fishman: So some of the low hanging fruit, thank you, is drug-drug interaction, drug-allergy interaction. I wonder if you could give us a list of other pieces of clinical information which could be made available to a physician in an ideal system.
Dr. Blackford Middleton: Sure. The National Guidelines Clearinghouse that’s run by the Agency for Healthcare Research and Quality has hundreds, if not thousands, of guidelines in it. In those guidelines is all kinds of detailed knowledge about the right thing to do for the patient based upon a constellation of symptoms or a diagnosis or a condition. Regrettably however, most of that knowledge is not translated into the healthcare information technology clinical decision support systems. The kinds of things that we hope to include as we work through this field are not only the right kinds of diagnoses, standard codes, medication standard codes, laboratory standard codes. Kind of the nuts and bolts of the informatics infrastructure, the DNA if you will of our data. But further things that define, what’s the right thing to do for this patient at this moment? Are there order sets we can use for basic conditions or standard care processes? Are there care protocols we can use for complicated patient care processes or chemotherapy, for example, over time is very easy to understand. Or are there even other things like the standard report for measuring quality and reporting quality to the CMS or other payers who will be interested in that. And of course simple things like alerts and reminders for preventive care services and other forms of guideline knowledge. This whole array of knowledge needs to be encapsulated, encoded, represented and then delivered to the EMR, or embedded in the EMR, to support clinical decision making at the time of decision making.
Dr. Eric Fishman: And so the information is available but in some cases is available in books in a library. And as you’ve just mentioned, it’s critical that it be available to the physician at the bedside in the exam room in the electronic health record that they’ve chosen. It’s 2009 now, is the CDS that’s being used most frequently a separate stand-alone system or are there beginning to be signs that the information is embedded into some of the higher end EHR’s?
Dr. Blackford Middleton: There’s both and I think it represents frankly an emerging marketplace for this business of knowledge-based tools and services. There certainly are a number of very credible free-standing decision support type tools, things which you can access to look up a drug monograph or a drug dose or do a drug interaction. It requires however of course repeating the data entry from my EMR system to another free standing decision support system and that’s often unattainable. From the doctor’s point of view and the clinical workflow, I can’t afford the time to do duplicate data entry. So it’s much more advantageous to have the decision support embedded in the EMR as you say.
Dr. Eric Fishman: How long do you think it will take until that’s the standard in this country?
Dr. Blackford Middleton: I’m going to suggest there’s even one more step which might be the more desired standard. In a way embedding the knowledge in the EMR certainly is becoming much more doable given the sophistication of the EMRs as they’re coming forward. We just finished an analysis, for example, and this will be reported in the Journal of the American Medical Informatics Association comparing the clinical decision support capabilities of nine commercial EMR systems. We found them actually to be pretty good. Not as good as they might be but pretty good. I say not as good as they might be because we compare them against home-grown applications that for example we have at Partners and the Brigham and Women’s Hospital where after 40 years we have very sophisticated decision support tools and what not in place. And some other notable sites, such as Vanderbilt, Columbia and VA for that matter as well. But the sad fact I think still is that getting the knowledge into the EMR is still not possible for the average doc. So the doc then has to really trust the vendor to make sure the vendor is putting the right knowledge into place or take on this responsibility him or herself. And this knowledge management exercise if you will, the translation of guidelines knowledge or other professional society knowledge all the way into EMR is an extraordinary burden. We did an estimate of the cost of this burden for this country at the Center for IT Leadership and probably if we were doing it the way it needed to be done, it would be about 25 billion –
Dr. Eric Fishman: It sounds like another government initiative might be appropriate in a situation of that nature.
Dr. Blackford Middleton: Thank you, I agree whole-heartedly. So the other model then, after thinking about knowledge that’s embedded in EMR, is to think about knowledge which is made publically available as what’s called nowadays as a Web service. So for example, when you do your taxes if you’re using a tool that for example allows you to hook up to your bank and download your bank balance that’s happening via a Web service. You give your permission, you give your logon and credentials so your Turbo Tax application or whichever one you might be using can access your Bank of America and download that information. And maybe even some knowledge about your accounts to your Turbo Tax. The same type of approach I believe is going to be the next generation of decision support. And I’m not alone in this belief; this idea has been around for awhile. But some work has demonstrated for example, where we can build a Web service at a reputable institution where the knowledge has been validated and tested and developed, et cetera. And then deliver it to a remote EMR around the country or even around the world.
Dr. Eric Fishman: In real time.
Dr. Blackford Middleton: In real time.
Dr. Eric Fishman: You’re discussing I guess they call it SAS software. It’s a service with cloud computing, some of the newer methods of gaining access to voluminous amounts of information in real time from anywhere.
Dr. Blackford Middleton: Absolutely. The bottom line is the small office practice – in my own experience practicing in the Providence healthcare system, Portland, Oregon – small office environment. You know, there was four or five docs basically, full-timers – and there’s just no way that the right amount of knowledge engineering can be done to keep up with all the new guidelines and protocols as they come forward. So I think that there is a federal role for establishing potentially a national knowledge repository which all vendors could use and ideally free of charge. I mean after we pay our taxes, free of charge. And what this might look like is, for example the National Guidelines Clearinghouse run by the AHRQ with this Web service capability for guidelines anointed as being worthy of doing this with. So that then a remote application of any kind, but one that has accepted this responsibility of subscribing to the software as a service, making that work in the application. I think that’s going to be the way to go so that we can alleviate this knowledge management burden from the average doc, give him the best evidence via his EMR investment from a federally supported, federally funded and managed national knowledge resource.
Dr. Eric Fishman: Fascinating, it’s an interesting direction to go in. Stepping back a few feet, there’s quite a number of acronyms; CDS, CDSS, EBM. I’ve mentioned just a few. I wonder if you could go over a list of them so that we could have a pretty good understanding of the scope and breadth of this industry because I think it’s a bit more complex than an expert such as yourself may be leading on to believe.
Dr. Blackford Middleton: Well sure. And gosh, I don’t know where to start. Informatics in a way has almost prized itself as an informatics soup – I mean as an acronym soup. But you know when I say CDS basically that refers to clinical decision support. And CDSS is a clinical decision support system. And –
Dr. Eric Fishman: So what’s included in the system that isn’t included just in the support?
Dr. Blackford Middleton: Basically, they’re almost equivalent. The small distinction I might draw would be that a CDSS is an implemented CDS if you will. So this is what you have to do in the software technology, in the EMR or the CPOE Computerized Provider Order Entry or the EMR to keep going, the electronic medication administration record. To implement the knowledge becomes a CDSS. And the CDSS is a component of a broader application, a module if you will in a broader application like I just described.
Dr. Eric Fishman: And will the CDSS include a KM, or a knowledge management component?
Dr. Blackford Middleton: Absolutely.
Dr. Eric Fishman: And that’s what you were discussing possibly being a cloud computing. We’re not there today but it may get there. Now talk about if you would please EBM, Evidence Based Medicine, and where that fits into this soup.
Dr. Blackford Middleton: Sure. For many, many years the idea about EBM was to take the world’s collective wisdom as demonstrated through clinical research trials and randomized controlled studies and the like, and to synthesize it into a codified piece of knowledge if you will that says here’s the best thing to do for community acquired pneumonia. And the challenge always with EBM is to say well here’s what the best evidence might suggest from the world’s trials. But how do I actually fine tune that then to the patient before me? Mrs. Smith who’s 67 years old presenting to the ED from the nursing home with her progressive cough and fever may not exactly be like what the guideline from the EBM says, or the evidence based medicine says. So the challenge with EBM always is to say here’s the best evidence how do I apply it to my patient right before me. And in fact that’s where CDS may come in, clinical decision support or decision support system, to say let me help you fine tune this knowledge for the patient before you; Mrs. Smith who’s 67 from the nursing home with her cough and fever.
Dr. Eric Fishman: Thank you. And then there’s EBG and EBID and you probably know more about these than most.
Dr. Blackford Middleton: You may have to help me with EBID.
Dr. Eric Fishman: Evidence based individual decision making.
Dr. Blackford Middleton: Ah. Same idea, fine tuning the EBM if you will to a patient before you. And the tricky part here is, is we often times will think about how do we actually apply the evidence to the patient before us? But the next level of sophistication is to say, what are the patient’s preferences for whatever the evidence might suggest? The classical example in this regard is how men approach the prostrate surgery decision. Certainly men will have different preferences and different utilities if you will about their sexual function or side effects or other types of complications of surgery versus medical therapy. And lots of analyses have shown that these preferences may actually sway the decision one way or the other depending upon the man’s preferences. There are many other examples like this but that’s kind of the next level of sophistication where we take the evidence based and then fine tune it not only based upon patient characteristics, but their preferences. And further one might suggest that there are preferences that pertain to the provider and contextual issues if you will about his or her coverage, or the patient’s coverage et cetera, that can further be added to the decision support to make it not only the right decision but the doable decision, based upon coverage and insurance and et cetera.
Dr. Eric Fishman: If a physician wanted to learn more about CDS there are a number of educational entities including your own, the CITL. Can you name some of the places for further research that a physician could go to?
Dr. Blackford Middleton: Sure. A number of things come to mind Eric. The first thing I guess of course is many of the professional societies that we’re already a member of are taking a renewed interest in informatics and decision support and HIT, healthcare information technology. So I’m familiar a little bit with what’s going on at the American College of Physicians and the Society for General Internal Medicine in my area. I know many of the surgical professional societies have active groups in informatics or healthcare information technology, pediatrics as well active in the American Academy of Pediatrics, et cetera. So I would start with the professional societies. I think the next level of inquiry might arise - I might point people toward the informatics societies, American Medical Informatics Association you mentioned already. I have the great privilege of serving on the board of this organization by way of full disclosure. But it offers comprehensive programs and symposia and informatics as well as a number of very accessible training programs. The 10 by 10 program for example from AMIA is a distance learning program that docs can take and do from the comfort of their own home with a pc. There’s some other types of certificate programs. There are master’s degree programs at National Library of Medicine funded training programs. HIMMS, the Healthcare Information and Management Systems Society has a number of certification programs and professional interest groups in both in-patient healthcare IT and ambulatory IT. So it’s growing by leaps and bounds. One of the things we recognize with the stimulus bill and the high-tech component of the stimulus bill is that to get all these EMRs installed we have to launch and manage a massive educational campaign, not only for the providers but also for the workforce to make sure that we have enough people to install the EMRs in the providers offices. If you will, a research oriented or someone doing a development if you will. But even HIT professionals, you know, someone who’s qualified to come into a clinic, analyze the clinical practice, the workflow, the layout and configuration of the examination suites and the consultants offices and to figure out how to use their company’s technology or the best company’s technology to support the workflow and implement all the things we’re talking about in a successful way. We don’t have enough of those folks either.
Dr. Eric Fishman: And I stand corrected I would call it an HIT expert as opposed to an informaticist.
Dr. Blackford Middleton: Yes.
Dr. Eric Fishman: And I’ll thank you now for your segue taking us from CDS to workflow. And if we could discuss workflow first from 30,000 feet if you will and then we’ll discuss workflow as it might pertain to a specific physician’s office most particularly in primary care if we could.
Dr. Blackford Middleton: It is probably the most challenging thing to get right. It’s not like Microsoft Word. When I buy Microsoft Word I have maybe one workflow in mind. I’m going to write an email or I’m going to write a letter or I’m going to write a paper. It’s basically a pretty clear workflow requirement. Workflow requirements in ambulatory care are just innumerous or legion. And it’s subtle to make sure that we understand the workflow carefully and adequately so we can map the clinical information system, the HIT, to support that workflow. What often times happens is there’s a gross mismatch between the vendor’s technology understanding the workflow, and what you and I might like to do in clinical practice. And when there is that mismatch it’s an impedance mismatch because then workflow may be impeded. And what we have to do is try to understand the existing workflow, map healthcare information technologies to support the existing workflow yet not ossify them because we in the end of course want to do workflow redesign and try to improve our clinical practices and workflow to get a bigger bang for the buck.
Dr. Eric Fishman: Thank you. And let’s get a little more granular and let’s take a hypothetical one physician family practice office. And let’s take a hypothetical electronic health record that has been designed, at least to a significant extent, for that exact environment. Not all clubs fit the same, not all practices are the same. Can you give us some specific examples of the types of workflows that a physician who is in this prototypical but maybe not so prototypical one physician, family physician, office that they should be thinking about when making a decision as to which EHR is most appropriate.
Dr. Blackford Middleton: Sure. There’s many, many things to consider and there’s probably going to be more than I can possibly get through with you here. I might point the interested reader to a textbook that was recently put out by the American College of Physicians. My only interest in it is I did write one of the chapters. But Jerome Carter wrote a book called Electronic Medical Records with all of us helping on chapters and it’s really super. It goes through from stem to stern kind of, lots of the different issues for the doctor to consider and it’s written for the physician. An example that comes to mind in this regard, in a solo office environment, a solo practitioner office environment, that physician may have two or three rooms running simultaneously. Busy family practice seeing lots and lots of patients per day, the record has to consider the multi-room environment. If the record system doesn’t consider the multi-room environment the doctor may be locked into a one chart at a time access model. That simply doesn’t make sense for the workflow.
Dr. Eric Fishman: And so the one chart at a time would be good possibly for psychiatrists but not for a family physician.
Dr. Blackford Middleton: Possibly. I don’ know of any psychiatrists running multi consultation rooms at a time –
Dr. Eric Fishman: No exactly so –
Dr. Blackford Middleton: Yeah, I know.
Dr. Eric Fishman: So the one patient at a time is appropriate and if it’s ossified in that fashion then a family physician couldn’t use it. They may buy something that has a lot of features but it’s not useable for them.
Dr. Blackford Middleton: It doesn’t fit the workflow. My workflow may be to have three rooms running and if I have to open and close charts in each room rather than opening a chart, documenting some, maybe asking for a procedure, going to the next room, interviewing a patient, documenting some, maybe looking for a sample med, doing to the next room, opening a chart, documenting some or what have you. And then if I have to open and close those charts sequentially or serially it simply doesn’t fit my workflow. What I’d rather do is have all three charts open at the same time, do what I need to do in the idiosyncratic manner in which I might do it which could be very non-linear but that’s what I need to do. Another example is the idea of having a chart open in the office and asking the nurse to start the chart and I come and finish it. If the chart doesn’t allow a multiple-author kind of scenario like that my nurse may open the chart and take vital signs, maybe even take some of the history or review the medications, what have you. And then I come along and do the physical and verbal examination and then do my assessment plan. If I have to start a separate node and can’t use her data entry in my node it simply doesn’t fit the workflow. Now the records are becoming much more sophisticated in this regard and the key thing for the practitioner is to really have a clear sense of your own ideal workflow and many doctors don’t think like system engineers. That’s not what we do.
Dr. Eric Fishman: That was exactly my point and I’m trying to draw from you the information that a specific individual physician, in looking at five or ten or 20 electronic health records that may be recommended to them, what they need to look at specifically as it relates to them. You’ve given us a couple of examples of having multiple charts open. Do they need to be thinking about what laboratory or x-ray equipment they have in their office or other idiosyncratic workflows that they have individually?
Dr. Blackford Middleton: You’ve mentioned a very good one. It’s really incumbent upon the doc to look at the record and figure out where he or she wants to get all the data for the record in the record. Some of that will come from our own documentation, some will come from the care team’s documentation, nurse, some may even come from front desk staff or other ancillary staff as well. So that describes a number of different roles that the record has to be able to accommodate with appropriate security and confidentiality and document privileges. So that’s kind of human data entry. The other part though is machine data entry as you alluded to. Can I make sure that the laboratory interfaces to my office based equipment if I have laboratory analyzers in the office or in my building, or from the regional free-standing laboratory facilities; Quest or other commercial laboratory facilities. The interface capabilities or records today is much more sophisticated than it’s ever been before but it’s still not trivial.
Dr. Eric Fishman: And how does a physician ascertain that in fact the software program that they’re about to spend $10,000, $20,000 on the license and $30,000 or $40,000 implementing that it fits each and every one of the specific workflows that may be required?
Dr. Blackford Middleton: Here again I would make sure I was knowledgeable enough as a consumer of this technology to understand how to describe my workflows to the vendor. The vendor will have an idea about my workflows. They will say, “Well I understand what family practice is and I’m going to tell you how you might want to do things.” But at the same time I think the physician as a good consumer has to know what their requirements are. Certainly when we go shopping for a car we have an idea of our requirements. When we go shop for a house we have an idea about our requirements. The same thing has to be true; the knowledgeable consumer is the best way to be protected and to get the right technology.
Dr. Eric Fishman: Is that part of what you might call a readiness assessment?
Dr. Blackford Middleton: Absolutely.
Dr. Eric Fishman: And if you’re prepared to sit and chat about that for a few minutes I’d be delighted.
Dr. Blackford Middleton: Sure. It’s another way to understand how ready a clinic and a physician group is for technology. Readiness assessments actually are often done by the vendor who will say this clinic is ready in the following ways or it’s not based upon installed hardware or interfaces or workflows and patterns and whatnot. But I would suggest it’s equally good, as I was alluding to already, for a physician or a leadership of a clinic to do their own readiness assessment. Because the other part of this is kind of the socio-cultural change which will come inevitably using these technologies. Are my docs in my group really ready to use the QMR – or the EMR.
Dr. Eric Fishman: Is there such a thing as the QMR.
Dr. Blackford Middleton: I’m sorry. That was a – there is a technology actually, quick medical reference system, I worked on years and years ago. It’s a decision support system for differential diagnosis, lots of fun to talk about that too. But to be ready for EMR some pundits have said this adoption of EMR is 95% socio-cultural change and 5% technology change. And what that typically refers to is how ready is the physician’s group and the practice at large, ancillary staff included, ready to take on this new technology and do things in a different way. It’s different obviously, charting in a record than it is charting on paper. The information flow in the office is different when the chart’s not moving around and one is accessing information rather on the screen if you will or in the computer. And there’s lots of subtleties about readiness for technology change that are very subtle. But I think you can look at the clinic and look at the practitioners and say, do they have a laptop at home or a pc at home already? Do they have a laptop or a pc on the desk at work already? Do they use email? Are they willing to type a note in email on the keyboard? Are they dictators versus paper-based charters in the paper-based environment? Do they scribble a note or do they dictate it away? And all these issues apply if you will or pertain to the adoption of EMR because some of them will be directly impacted obviously. Getting the physician’s data into the record probably is the most challenging thing. How do we actually get the physician to type a note or dictate a note or use dictation recognition software to get the data into the record, not only in plain text form but also in a form which is then useable for decision support. What that means is it has to be structured or at least coded in a way that if I say the cholesterol was done elsewhere and it’s 240 milligrams per deciliter that then goes into decision support somehow in a flow sheet or in a table so I can see it and use it in my reminder algorithms. So that’s probably the most challenging part of EMR implementation and the socio-cultural change is to figure out how to get the doctor to engage with the record.
Dr. Eric Fishman: I’ve frequently stated that the history, the historical portion of the record, is the least critical as it relates to data. And that maybe that should be placed in in free text and then the review of systems, past medical history, the physical exam should all be codified and structured. Would you have any disagreement with the fact that the historical portion of the record is best done in a unique, individualized, fashion.
Dr. Blackford Middleton: Now Eric, my professor of medicine from years ago is probably turning in his grave. Because he would have suggested to me, and the way we were taught, at least in internal medicine is, that the history exactly is 95% of the diagnoses can usually be made by the history, careful history. So that might be a difference between a surgical approach and a medical approach –
Dr. Eric Fishman: They taught me listen to your patients, they’ll tell you what’s wrong with them.
Dr. Blackford Middleton: Exactly. So I think therefore the history is important. But your point is whether or not the history has to be all coded and codified as data if you will structured in the EMR and that’s a fair question. It may not need to be actually today. There’s a very small number of critical data elements relatively that must be coded for informing decision support, for making the bill obviously and for keeping track of chronic care conditions over time. If we attempt to boil the ocean and code everything, that’s absolutely a recipe for disaster. What we should do is try to define the critical data elements we need for our profession, whether it’s orthopedics or internal medicine or diabetes care versus total hip, what have you. Because there’s a smaller number which then can be actually addressed in a more detailed way for data entry and monitoring and decision support, rather than trying to code everything.
Dr. Eric Fishman: Full disclosure, as I think you know I’m involved in Dragon Medical. And I’m a big proponent of speaking the history, and maybe a few sentences of the plan so that the record understands your thought process and how you came to that decision. Can you think of any specific pieces of data within the history that should be codified in various situations?
Dr. Blackford Middleton: Well certainly – yes, I’m a fan of speech recognition when used appropriately, in this way actually. The way you’re describing I think is to distinguish what might be entered in a free text form in dictation or voice recognition, versus what might be actually necessary to be captured in a structured way. In my own experience with voice recognition software and I’ve used it for 20 years from some of the very experimental systems, early Dragon actually at Stanford and whatnot. I think actually a useful approach is to make it a multi-modal approach. To use voice recognition where it makes sense but to not be resistant to, or to be willing to put the hands on the keyboard and capture a few pieces of data that need to be structured and in structured form. So what are those? Well the key pieces of data that really need to be structured or captured in a coded way. Certainly we believe, I would suggest the evidence from informatics literature would suggest, the problem list is very important. And this is not sometimes easy to swallow, depending upon the size of the practice and what not. In a solo office environment do I need to maintain a unique problem list when it’s only me? That’s a fair question. But I think of course we do make a bill and in the bill we do delineate a number of problems, at least in ICD9 coding. And I think that when we are increasingly sharing data with potentially personal health records the patient will expect to have a problem list, even from a solo office environment. In a large academic medical center or in a large group practice of course, the problem list is essential to help coordinate care among a wide array potentially of caregivers for the same patient. So there will be the orthopedics problem as well as the internal medicine problem and we have to have them both so we can understand the full gamut of things going on with this patient. Medication list is another cardinal data type if you will. We have to capture the meds to do all the decision support in a clinical decision support system to avoid drug-drug interactions, avoid adverse drug events to the extent we can. After those two principle data types obviously allergies must be online so we can avoid allergy reactions. And then in the last category certainly laboratories nowadays almost come for free because they are often times coded, in most of the laboratory systems, at the point of analysis or at the point of origin. So if they come over in what’s called a LOINK code, this is a laboratory data code system that gives a name and a number to all of the cholesterol results for example. So the doctor doesn’t have to do anything but the EMR can still use those data in decision support. Now beyond problems, meds, allergies, laboratories, there probably are some specialty specific things and other data types that – obviously USPSTF, the United States Preventive Services Task Force, alerts and reminders for screening and healthcare maintenance, those probably should be coded as well.
Dr. Eric Fishman: Thank you and very good analysis. I’ve frequently stated that the physical examination should be entered by clicking or through this systematic approach that the electronic health record provides. And I say that because it’s probably easier to get it in rather than speaking. But is the physical examination necessary for the data analysis?
Dr. Blackford Middleton: Frankly, often times not. Usually what you do in your head in making your assessment and plan based upon the physical and the historical information is the most important thing. And capturing that assessment and plan in free text is actually extremely helpful because that’s where the thinking is really revealed of the provider, the physician and the care team. I think in a physical examination the main problem right now is we don’t have a sophisticated coding schema for capturing, for example, two over six ejection murmur left sternal border radiating to the axilla. You know, that’s very hard to capture in a structured way. It may be more minimal to a graphical or a clickable way in a comprehensive schema if you will of physical findings. My concern there would be though that – in at least a typical internal medicine survey of the body that there would be way too much clicking rather than just using shorthand for physical exam findings. Within normal limits or negative for the usual string based upon the category, what have you. And those data are not often used in CDS anyway, yet, so we don’t have to worry about coding them so far.
Dr. Eric Fishman: And certainly there is this problem with too many clicks in electronic health records and I think we’re all trying to get away from that. There is a substantial amount of data that is drawn from the assessment and yet I see a dichotomy here. I believe in dictating the assessment as I’ve said because it tells you what’s going on. And then would you agree that after dictating the assessment that one or two clicks to put in the ICD9 code or some other similar method of putting computer readable information should be there?
Dr. Blackford Middleton: Absolutely. And CMS is going to increasingly require us to put an indication for each test and procedure. So that’s why in fact it might be from a workflow point of view advantageous for the – even the solo doc to maintain a problem list. If he’s ordering the CBC DIFF or the MRI or whatever, to easily apply one of the items from the problem list as the reason for the test and procedure may make that a single click kind of exercise. The other thought though I’ll share with you and I’m you’re well aware I’m sure, the other way to get data into the record – I mean there’s the inbound way if you will from the human, structured data entry and dictation and all the rest of it. The other thing though is once the text is online is to subject it to natural language processing. So this is actually very interesting technology. In a nutshell what it does is scan text, scan words, and look for known words and relationships to try to interpret the text. For example, we now subject all the notes in our EMR system to a small NLP program, natural language processing, to extract the blood pressure if it’s entered into the note but not in our blood pressure field. Blood pressure of course is used for monitoring any number of different things in lots of interesting ways. So it’s essential that we have blood pressure documented on all of our records. And where the physician or the care team, nurse and MA, what have you, if it’s not entered into the blood pressure field we scan the note and try to abstract it from the note. The same technology can be used to look at other kinds of information or derive other information from these notes. Even problems in meds that are not mentioned and things like that.
Dr. Eric Fishman: And so let’s talk about meds. And this is a problem I’ve had in understanding. Someone can be taking penicillin and they’ve mentioned penicillin and that’s the medication they’re taking. Similarly somebody could be allergic to penicillin. And how in your experience has the natural language processing been at differentiating between patient is taking penicillin and patient is allergic to penicillin if put in in a non-structured fashion?
Dr. Blackford Middleton: It’s an excellent question. And this is all about the sophistication of the algorithm that is being used to read the text. And it’s complicated because you could imagine all the different scenarios. Where the patient is allergic to penicillin, patient is not allergic to penicillin. Patient’s aunt is taking penicillin is not the same as the patient is taking penicillin and what have you. And these algorithms and our compute power frankly are getting so powerful now that they are better able to assess even distant relationships in the text. If not occurred at the beginning of the sentence but penicillin is at the end of the sentence, we’re getting better with the algorithms and the masters who are writing these things at figuring out how to draw the association correctly between the words and the text. It’s not a done deal and it’s not perfect but it’s getting better.
Dr. Eric Fishman: And that’s an important issue when a physician is practicing alone and he or she is using that information. But it’s a critical issue when somebody else is using that information. So let’s now discuss the interoperability, the IAG, and how information is getting into and out of a chart. So let’s speak for a few minutes now about a few more acronyms; IAG, HIE, RIO. All having to do with moving data around the country. And if you could discuss that please.
Dr. Blackford Middleton: Sure. Basically the IHE refers to the integrating the healthcare enterprise effort that’s been spawned or supported by both HIMSS, the Healthcare Information and Management Systems Society as well as RSNA, the Radiological Society of North America and I think some other standards development organizations. And IAG is focused on just that, integrating the healthcare enterprise. So how do we make sure that the laboratory system and the blood bank system and radiology system and the record system – both electronic medical records as well as potentially patient portals to those data. How does all that fit together and work? And that’s been very, very helpful. Lots of credible work there. For folks who are interested, a very fun thing to do is to go to the IAG showcase at the annual HIMSS meeting, Healthcare Information Management System Society. And they do a thing called the connectathon which is fun. It’s kind of like marine boot camp I think for the vendors. They’re asked to come together and make sure that all their technologies can interoperate. What I mean by interoperate is share data from point A to point B. And sharing data can happen in a number of interesting different ways which is important. It is possible to simply send information. I’m sending information to you verbally and you’re storing it at a level in some ways and we’re storing it on tape. But from a computer point of view if I send an email in unstructured form it doesn’t actually enter into a database or get used by the point B system. A more sophisticated way to share data is to put it into a structured message. And this is where HL7, health layer 7, comes in. HL7 defines an envelope so we have a reliable transport mechanism if you will from point A to point B. But in the envelope we’re still allowed to say whatever we want. You may be French and if I send you an English message, an English letter inside the envelope, you may not understand it. And then the third part is to think about the content of the envelope. How do we actually standardize the message within the envelope with controlled medical terminology or standard codes and terms so that not only is the message received in a reliable way but then it can be interpreted by system B. And this is important because the value of the message changes dramatically with how well it can be understood. We did a study at the Center for IT Leadership analyzing the value of healthcare information exchange or HIE, healthcare information exchange, between doctors’ offices and clinics and hospitals and retail pharmacy and the insurance plans and payers. And if you look at where information travels from a doctor’s office after a clinical encounter from the small amount of work we do, relatively in ten or 15 minutes, hundreds of transactions follow. From the retail pharmacy to the laboratory for orders to the payer to the plan to the quality organization locally or nationally, public health potentially for communicable disease reporting, what have you. So hundreds of transactions follow from even a single out-patient clinical encounter. And when we looked at the value of healthcare information exchange and inoperability at the Center for IT Leadership we found that the basic level of information exchange has value certainly. It’s useful for you to know about my patients when we share patients. But the real value comes when that information is exchanged in a transparent manner with both semantic and syntactic encoding. What I mean by that is semantic refers to understanding the word. So using a controlled medical terminology which the systems can both understand. And syntactic refers to the message, the architecture or interface standard for the messaging. If we had full semantic and syntactic interoperability across all the systems I just described, we’d save about 78 billion in this country from reduced duplicate tests and procedures and administrative costs and managing healthcare information exchange. So IHE focuses on all of this information exchange within a hospital or within a system. And HIE, regrettably the same letters, focuses on all this information exchange outside of hospitals and between hospitals and clinics and pharmacy and laboratory.
Dr. Eric Fishman: Blackford, thank you for that. Now if we could take all of this information which is of critical importance to the healthcare and the budget of the United States, and bring it down to an individual physician or a small physicians group what kernels of advice, what specific advice would you give to a small medical group that may not have an IT staff and an office manager or an administrator who was able to look at this. What would you tell the physicians who were actually going to now be looking for and implementing an electronic health record? What would you suggest that they do at this point in time?
Dr. Blackford Middleton: Well number one, I think you’ve already suggested Eric that they know there’s an issue, that they are thinking about EMR. I’m concerned about physicians who aren’t even thinking about EMR yet. So number one I’d say to all of our colleagues that EMR is here and it’s going to help improve practice and save costs and in the end all my work, I hope in the end results in medicine being fun again. So that we actually can see patients and focus on patients and not have to worry about the bill and information management and paper pushing. So we all have to begin thinking about this. Number two, the EMR really is going to be our modern day stethoscope. We won’t be able to take care of patients as well as we did before – we’ll be able to take care of patients much better than we did before using this EMR technology just like the stethoscope helps us oscillate the heart and other organs. The modern medical practice can’t be done with our unaided mind if you will. There’s simply too much to know, too much to remember and I wasn’t trained in medicine to be a repository of knowledge. I’m much more interested in being a patient advocate and being a broker of knowledge to help my patient achieve the best outcomes they possibly can. So what do you do? I think the number one thing to do is of course just begin to notice, read and observe what’s going on around you with your friends and colleagues in the hometown as well as in the professional societies. Take advantage of every free piece of literature and professional society service that might be offered. Every professional society is getting into it and has an informatics track or an HIT track. And when we go to our medical meetings we can attend a few of those sessions and begin to learn about things. Certainly there are articles appearing in the journals, especially the society journals as well as the general clinical journals and what not, and take not of those and observe. Third I guess there are of course billions of dollars at play now. President Obama has taken this hook, line and sinker. He fully believes and subscribes and understands I think how we need to have healthcare information technology in place to begin really addressing the fundamentals of healthcare reform. We didn’t talk about this earlier but an important part of HIT I think is that it will really enable us to do comparative effectiveness research, in practice in ways that we’ve never been able to do before. So that we’re not kind of guessing about what’s the right thing to do for our patient. We can look at the evidence, look at the experience from our record systems and understand what’s the right thing to do. Fourth, when you’re ready to take the leap and actually consider EMR for your environment, again talk to all your friends and colleagues, look at what the professional societies are suggesting and recommending. But then find the right vendor for you and that’s a complicated process. The EMR has to fit like a glove; it has to be sized to you. And that’s challenging because it’s a remote alien technology if you will but bring the vendors in. Let them come and who you their wares. Understand though before you bring them in, get together with the group and really decide what are your top priorities for this EMR. Is it automating the workflow, is it generating a better bill, is it improving documentation. Or at a minimum, and this will be a requirement, is it complying with the meaningful use criteria. What are the meaningful use criteria? This is what the administration is putting forward as the means by which we’ll be judged if our EMR is an appropriate EMR and being used appropriately. Certainly the EMRs must be certified that we wish to adopt or that we will adopt. And certification now will have a couple of flavors most likely. The certification commission for healthcare technology led by my old boss Mark Levitt and probably a few other different and perhaps less onerous ways of receiving certification for EMR. But we’ll be paid from ARA for adopting a certified EMR which meets the meaningful use criteria. And what they pertain to is are we actually demonstrably improving our care and reporting quality measures to CMS. And if you do then you get the ARA money back. And if you don’t after five years you may actually be penalized and receive less income from seeing patients without EMR.
Dr. Eric Fishman: Thank you. Do you believe that most of the top level electronic health record products out there – and there’s many now that are guaranteeing that physicians who utilize their technology will meet the meaningful use criteria. Do you believe that most of them already an adequate quality level that when the meaningful use criteria are announced that they will be met by these?
Dr. Blackford Middleton: I think most is a fair way to characterize it, but I would suggest the buyer beware. To be very careful and not only take what the vendor will say, and the vendor will be credible and honest, always to the best of their ability. But make sure you see it for yourself in a good old medicine kind of way. Look at the primary data, examine it yourself and understand it from someone that you understand and trust that it’s working okay in your environment. And yes, all the major vendors will meet these criteria and be able to report on meaningful use. The target is moving however. The first year is going to be this level of meaningful use criteria, the second year’s this level, the third year’s a higher level And thus with a moving target you have to have a vendor who’s really going to go the distance as well.
Dr. Eric Fishman: Blackford, anything to add to this?
Dr. Blackford Middleton: Well I’ve had great fun, thank you very much. And happy to help any time. And although it seems like a big mountain to start climbing and a big challenge, really the view from the top is fantastic.
Dr. Eric Fishman: Blackford, I’d like to thank you for your time today. It’s clearly obvious that you are a true expert in health information technology and we appreciate your expertise. This has been Dr. Eric Fishman for EHRtv. We’ve been speaking with Dr. Blackford Middleton, Director of Clinical Informatics Research & Development for Partners Healthcare. Until next time, thank you.