Heroes of Healthcare
Heroes of Healthcare

Episode · 1 year ago

Informatics: Data & Decision-Making in Healthcare

ABOUT THIS EPISODE

How much of your healthcare can you trust to a computer?

Turns out, it’s probably more than you think.

Health informatics play a critical role in everything from diagnosis to decision-making.

In this episode, I chat with Dr. Stanley Huff, Chief Medical Informatics Officer at Intermountain Healthcare and Clinical Professor at University of Utah in the Department of Biomedical Informatics, about the past, present, and future of informatics in healthcare.

What we talked about:

  • The history of computers in medicine
  • How informatics can help diagnose disease
  • How to harness data effectively to support better healthcare outcomes
  • The role of informatics in reducing medical error through advanced decision-support
  • How pandemics like COVID-19 underscore the need for more efficient data sharing

Heroes of Healthcare is hosted by Ted Weyn.

To hear this interview and more like it, subscribe to Heroes of Healthcare on Apple Podcasts, Spotify, or wherever you listen to podcasts.

You were listening to heroes of healthcare, the podcast that highlights bold, selfless professionals in the healthcare industry focused on transforming lives in their communities. Let's get into the show. Welcome to the heroes of healthcare podcast. I'm your host, Ted Wayne. On today's episode we have Dr Stan Huff joining us. Dr Huff is the Chief Medical Informatics Officer at Inner Mountain Healthcare in Salt Lake City. He is certified in clinical pathology and clinical informatics and has been working in the industry for the past twenty five years. He also is currently a clinical professor at the University of Utah in Biomedical Informatics. Thank you for being on the PODCAST, Dr Huff. Thank you. It's great to be here and you don't need to call me Dr Huff. Stand will be fine. Great. I'll refer to you as stand. Can you tell the listeners a little bit about yourself and a little bit about how you ended up with Inner Mountain and informatics? Yes, you know, I grew up in a small town in Utah and my father was an electrician. That's I was always sort of fascinated with electronics and electrical things and that sort of stuff. And I mean I was in high school, graduated in one thousand nine hundred and seventy one, and you know, sort of the first exposure I had was to an HP, the programmable calculator, and then in college there was a little bit of computer science. I mean there wasn't a department of computer science yet, but they had computers and we, you know, we had some assignments to do some computer work and it was back in the punch card days, you know, you might a programming put it, then the punch cards and then you go to the data center. They but anyway, as computers science started to mature, I really became interested in computers and I I don't know if it's a genetic defect or what, but I mean I'm just fascinated by the power and the possibilities of computers and I thought the best real life implementation of that would be to apply it in medicine. And so I, you know, I read about things and and read literature and it turned out that the University of Utah and that medical school had a reputation, even at that time, for use of computers in medicine. And that's what informatics is really is the application of computer science to healthcare in a broad sense and in a more specific sense. It really is about trying to make systems not only convenient but smart so that they helped permissions make the best decisions possible for for helping their patients. And so I went to medical school. Rule I took electives when I could with Homer Warner. During my elective time I did a year of internal medicine residency. Then I did clinical pathology residency. Again, during any free time I would work on computer science kinds of activities. My first and it was really at when I was a senior in medical school, I decided to go into medical informatics as a career. At that time it wasn't a career. I mean we're talking thirty five years ago, maybe forty years ago now. But I did my residency and my first job was with bell laboratories, who at the time they had gotten into some medical informatics activities. But after working a couple of years for bell laboratories I came back to a joy position with the University of Utah and Inner Mountain Healthcare. My primary job is as a chief medical informatics officer at Inner Mountain, but I also teach at the University of Utah in the Department of Biomedical Informatics, a clinical professor there at the university. And how much of your time is d is teaching...

...versus practicing? You know, my teaching and academic time is pretty small. You know, five percent probably over all my activities. That earlier in my career it was much larger than that. Probably thirty percent or forty percent of my activities were related to education. But my current situation I yea probably five percent. So tell me about the early years of informatics. So you said thirty five years ago. I kind of laughed with the team here and I remind them that when I got into business I was pre internet, pre facts, pre email, all those things didn't exist when we got into things. But in so in the early days of informatics, what was the focus on? What were some of the things you guys were trying to work on in some of the areas you were breaking into? So Informatics withinner mountain in the University of Utah really started in the in the late S, early S, really really at the start of when digital computers became available, broadly available, and at that time, if computers were used at all, they were used for patient administration, you know, setting up appointments or billing patients, just doing sort of the business part of medicine. And homer Warner, who was a cardiologist, got interested in informatics and he came back and he was really a pioneer in creating the field of informatics and his focus was on using computers to automate activities within healthcare and to to actually do decision support. He did it. One of the seminal studies was the application of base theorem to diagnosis of CO general heart disease, and homer set up a critical study where at that time, as you point out, the technology was very different. They wasn't an the ultrasounds, wasn't any cardiac calthorization, there wasn't any you know, you had to basically look at the signs and symptoms of the patient and determine what sort of heart defect a new baby might have and then go into surgery to try and correct that if it was a life threatening sort of situation. And there's an theorem called base Theorem, which is just a statistical method of looking at the signs and symptoms that people have and knowing that, for instance, that people have pneumonia, that eighty percent of them have a cough and seventy percent of them have a fever and other things, and so you can do a reverse calculation and basically ask the question, okay, if they have a cough, if they have a fever, what's the probability that they have pneumonia? And homer applied that Theorem to diagnosis in these congenital heart cases and what he found is that the computer algorithm did as well as, actually slightly better than, the n expert panel of Physicians. So it was one of those pioneering things. You know a lot of people would still dispute that, by the way, but it really points to the fact that we're probably on a course that almost any clinical decision we could create a program that would do a better job at that than they human would do it. And that doesn't mean that we don't need doctors and things, but that's sort of the premise of a lot of the work that I do. You just make sure I stood. So you're saying that, yeah, the computers are going to become better at the diagnosis or the predictability of certain situations and we are or it's already there. It's there in certain situations already. And it's not only the diagnosis that the management. So, for instance, we have algorithms that we've shown are better than people at managing ventilators, at managing chronic and coagulation, at detecting and treating sepsis in the emergency room, etc. I mean...

...those are that's, you know, taken all together, that's like less than one percent of things that you deal with in medicine, right. But any my bias and my belief is that any sort of activity that requires diagnosis or management, we can make programs that will do better than an individual person, for sure, and probably better than even a panel of clinicians, because they if they're biased, you can detect that and you can change the algorithm. People are not perfect information processors and they're going to make mistakes in their thinking and they're going to be up all night with patients and they could get confused between two patients or they could just be interrupted while they were doing something, and so, you know, I always characterize it is as sort of two things. One thing is that a single person can't know it everything or another way of saying it is that we all are ignorant, just in different places. It's sort of you have to pick what you want to learn. Nobody can learn everything and by definition, there's probably one best person in the world who could be taking care of you at a particular point in time, and the likelihood that that person is actually taking care of you as close to zero running. And so that's the first thing is just that we don't know everything. The computers can actually know a lot more than we know. Studies are shown that people can take into account somewhere between three and five variables when they make a decision, and with that number of variables they can they can be pretty reliable in the decisions they make. But if you get beyond that, people are basically random in the behavior in terms of what decisions they make. Interesting and computers, computers can can man you know, through artificial intelligence kinds of techniques and day's Theorem and other kind of decision processes. The computers can systematically consider many more variables when they make a decision. And that's very pertinent because medicine is one professor quoted. You know, medicine used to be, you know, simple and relatively harmless or ineffective, and now it's effective and dangerous because there are so many variables that have to be taken into account as we take care of patients. So I want to come back in a little bit later to the eye that you talked about. But when Homer Warner came out with the base theorem and started to evangelize and move this, this thought of medicine and informatics through the into the market, where there any certain case studies or proof of concept or things like that that he was trying to or your group was trying to promote that to help move this into mainstream medicine. There they're been a whole series of projects. Some of the earliest projects were, you know, sort of simple in a sense. We set up some rules, for instance, to to just watch laboratory results that were coming back from the clinical laboratory. You know, things like glucose levels and potassium levels and you know those sort of things, and designed to set of rules. That said, you know, if you see something this abnormal then you need to alert to clinician so that they can take immediate action. So someone who is on the Jackson and has an abnormal potassium level, that leads to a dangerous situation where they could have dangerous or fatal Ay with me, as you know, obviously, if you have too high a glucose level you can go into Keto acidosis, which is a life threatening situation. HM, and they're a lot of situations like that. If you're you know, you could detect with laboratory studies that you're going into renal fay or which you want to take prompt action or you're going to end up with with chronic reing will failure and could lead to death and other kinds of disabilities. So he did those and a lot of this work, by the way, was not only homer himself, for people who are working with homer, and one of the key individuals,...

...and that was a PhD medical and from Attics Descuy, the name of Scott Evans. So one of the individuals that worked extensively on these activities with Scott Evans, and he looked at detecting drug drug interactions, which is, you know, almost ideal situation for the computer because they're thousands of drugs and they can interact with each other and and nobody could memorize all of the drug drug interactions, and so the computers can do a much better job at that. It looked at also detecting allergic reactions to medications and that was interesting. I mean there's some, you know, and most of these things have some cish, some sort of social and policy overtones to them. I mean, one of the challenges see is that with if you're going to actually detect the old method, at least the old method of detecting them, your dependent upon people to tell you basically that they had times and symptoms and a clinician to say, oh, that could be related to the drug that we're giving you. You know, could be having some nausea or rash or you know other kinds of problems. You know, there there's a situation where if you're in the wrong kind of environment then there's punishment of sex for that, for noting the act that there was a drug drug allergy, you have to report it to the government, all kinds of other stuff. And so one of the things that's Scott did was an experiment where he tried to detect from from the electronic data who had an allergy. And instead of asking who had an allergy, he looked for treatments for out. You know, when when a drug was, if you will, stopped, you know, before the normal course of therapy would have done, when antidotes, steroids or an Arcan or, you know, other kinds of treatments for drug allergies were noted in the chart, or other kinds of signs and symptoms that are more or less specific for allergies. And what what happened is that literally, you know, ut from having maybe twenty or thirty drug allergy problems reported a year to having four hundred drug allergies reported the year because the computer could detect those allergies, when people either didn't detect them or were hesitant to report them because of the potential consequences of their reporting. When? What do you mean by the consequences? Well, what would have been the hesitation? You said that earlier punishment. I wasn't such meant by that. Well, in the early days, you know, there wasn't any protection for people who were essentially reporting quality outcomes and so on the drug this day. Yeah, so there were requirements that you would report to the government about drug interactions, but there was no safety in doing that reporting. In other words, you could still be sued by somebody because there was a drug interaction or the government might decide to take your license for for malpractice or you know all kinds of other sort of retribution, if you will, for for reporting. Sure with they've gotten better policy. So now, in general, as long as it's in good faith, you aren't negatively impacted by the fact that you report important outcomes that maybe negative outcomes but are nevertheless important to sure support so that you can make corrective action. Yeah, so the fear, the fear of reporting, has somewhat diminished and you can feel safe to do that. Right, but just not to go into so much detail, but just to say other kinds of things, I mean just to mention them. Help with people ordering the right kind of blood product, whether your order red cells or just give people plasma to improve their coagulation. Management of ventilators, actually detection of ventilators when ventilators are disconnected, detecting sepsis in the emergency room and treating treating septus early, detecting the Venus frombosis in...

...emergency room and Diagnosi and managing that aspect. The prevention of improper induction of Labor in in women prior to thirty nine weeks or just stay and you know there I never know how to count this, but I mean they're probably a hundred, what I would think of a maybe a hundred and fifty of these kind of major, kind of decision things, and they think they've all been reported in prayer reviewed literature and substantiated, you know, through the prayer review process to validate to these have been valuable improvements in outcome and in situations where we can, we report also on the financial impact of what we're doing. And with rare exceptions, you say money when you take better care of patients because you've prevent drug drug interactions. You you provide them the proper drug and so they get out of the hospital earlier, so you have shorter lengths to stay. You know a whole stories of things like that, and so those are all been, if you will, scientifically validated as being valuable improvements in the quality of care that we provide for patients based on that clinical decision support. Absolutely so, again, staying back to some of the early days, we are and how was the data collected? So, as you you know, I know today we have very sophisticated dm our systems and things of that nature, but back in the earlier days of this. How was the data collected and how was the data aggregat? So for laboratory day to you know that one of the reasons I went into clinical pathology was that it's one of the most automated parts of medicine. So you have a lot of in for months, electronic and some Mus to do the measure months. So the day the sort of electronic from the very start and we just worked on interfaces that would take them, that data out of the laboratory system and put it into what would now be called, you know, a really early version of electronic health record. It didn't have as much information in it, but you know, the important things sort of went in. First there was, you know, the laboratory data, then medication information and pharmacy. The pharmacy became pretty quickly automated as well, so that people were using computers to dispense medications in the pharmacy and to send them to the patient. So using the computer to order medication so you could check drug interactions. That information. How Science and symptoms were really entered, you know, by nurses at the bedside. You know, blood pressures and heart rates, those things are now they're automated devices, but in those days, you know, the nurses would take things and they would they would report the patient signs and symptoms and all of that stuff would be, you know, hand entered by nurses and then clinicians when they were doing history and physical exams as well. It's reminded me when you talked about the pharmacist. My father was a pharmacist and actually he's on our first episode, so he's ninety one, and I interviewed him as my hero when you talked about the pharmacy world. But I remember back in the late s when he got his first computer and the difference that that aid for, you know, for him to besides all the record keeping and everything, but the immediaccy on the allergy flag or, as you said earlier, the inner inner the interactions. If he put in a medication that was prescribed and there was a known interaction with another medication, it would flag it right away and allow him to either consult the physician or tell the patient you might want to go back and talk to your physician about this. So it's funny you mentioned it as well, because I saw that as a that was a big change in their practice. Yeah, we have a close tie there. That's fun. Yeah, so I assume with also informatics, there's a big part of it that is data analytics, which is the date is all coming in from varying areas and forms and the date is getting, as we said earlier, aggregated.

But then there's somebody who's analyzing it and coming up with different opinions or observations or opportunities to impact the medical community absolutely. And you know, as soon as you started getting sizeable amount of electronic data, people started doing that kind of analysis and you know, in our case we did have good names for it. Really even back then it was just, you know, historical records, if you will, that we had archives and but we still had access to they became known, as you know, enterprise data warehouses and and Ann Ale, the databases and and it's gotten progressively more sophisticated, and so I always think about this area of medical informatics in at least two categories, probably a lot more ways you could classify it, but one is is learning from what's happening and the second thing is is applying the knowledge in the actual clinical environment. So you know, you can basically watch how you're doing and watch outcomes and learn better ways to treat patients by looking at the data from taking care of them. But just knowing what's going wrong doesn't correct the situation. You need those programs that we talked about. Says you need the data, for instance, to learn what what is predictive about patients who have sets us in the emergency room. How important is it that they have low blood pressure? How important is it that they have a high lactic acid? What is the real association between those signs and symptoms in the disease? And so you're you're learning new knowledge. But then you need a program that when a patient comes in, you make sure you get there, you know their white count, their blood pressure, their heart rate all of the day to repeat that application. Then you apply the application and it says, Hey, this person's at a high risk of steps us, and that now changes the outcomes. You know. Now you're intervening. You not only know things, but you're applying that knowledge so that this a patient is going to get better faster and that you didn't miss a diagnosis changes your course of action as well. Right, exactly. Yeah, yeah, I would think so. So, as you progressing through in this world and continue to refine and reflect. A couple of things questions come to my mind. One was what was the effect the Internet had on this, and the second is how do health systems share this data? Is it all kind of proprietary within each health system, or is there a central database by which all this date is now being shared and aggregated and cross referenced? So there's there's a mix of things going on. So the highest most common kind of use case right now are people still using data within their own institution. Okay, and they're their laws basically that say you can't share patient data without their consent, except for their you know, for their treatment and care. So two physicians who are collaborating in the care of a patient can share data to take care of that patient. But if you can't just take data from inner mount and then combine it with data from, say, Mao Planic or guy see or somebody else, without patient consent or without institutional review. But there's a lot of that that's going on now because, you know, if you can prove basically and go through an institutional review board and show that what you're doing going is basically analyzing data, it's already been collected and that the patient is not personally identifiable, so you're not putting them at risk that somebody will find them, you know, expose publicly some past historical event or disease. That would be embarrassing. As long as you can prove that's not going to happen, then you can combine data. And so there's some really important sorts of activities that are going on that are large national databases, but they're their individual activities. That is there's no there's...

...no plan to put everybody's patient data into one database and most of the most of the combining the data, most common situations are that their designed for a specific purpose, designed to, for instance, understand the efficacy of medications or the complications that can come from medications or further focus on a specific disease. There's focused on lung cancer or colon cancer or, you know, other kinds of diseases or that sort of thing. And so there are a large number of, you know, what are called patient registries, which are combining data from many different institutions to learn about a particular disease and to understand better how we can care for patients with that particular disease or situation. So as long as the Pii is being protected, is being or or eliminated in the sharing of it for patient and care benefits, is is approved and allowed? Yes, and allowed if you follow the follow the right procedures. Why? That's that can be done legally and appropriately. And are there any data standard so that the way inner mountain is collecting the data is consistent with how maybe a tenant or a kaiser or one of the other big health systems is collecting the data, so that there is this ability to pull it together with data integrity? Yes, they're a number of standards. One of one of the most important standards bodies is health level seven international, which is a not for profit organization that creates standards for sharing of data, for exchange your data, and they're the ones who've created the fire standard, which is one of the most popular standards today, at least in terms of work that people are currently doing to try and implement that standard. And then, along with you, you can think of the standards into sort of parts that have to come together to create legitimate representation of the day that you have the structure of the data, and then you have to have some standard terminology of vocabulary that's used with that structure so that you know if you have a field in your database that said, well, what was the patients diagnosis? You don't want just spelling errors and other things in that field. You want you want to use a standard representation of that diagnosis, make sure that it's correctly spelled and that sort of thing. And so terminologies like Snow Edward stands for the systematic nomenclature of medicine, and another coding system called W which stands for logical observation identifiers, names and codes. Those two standards have been developed. Their other standards. There are standards for actually creating an electronic representation of a workflow or of a process, so that you can make an electronic representation sort of of a process like a diagnostic process, that where you have formal tools for representing that process so that you can visualize the process, where the decision points are, where data need to be collected from other outside sources, all of that sort of thing. So there are a number of those and those are in place. The challenge is that those things are not perfect or not complete in a way. If you say that you're using, for instance, hl seven fire and long and Snowmad. It still leaves different ways that you can legally represent the same data, and that's kind of the area of what people talk about is as semantic or plug and play interoperability, where you add some further constraints and information to the existing standards so that you're that you get the same day that truly represented exactly the same way every time across different institutions, and so that's been...

...growing. There's a different aspect of that, though, that's not so much technical as it is again, sort of social and political, and that is, you know, you can create models for these things, but you need expert clinicians to say what is the important data to collect and that happens that that that's not now technical people talking about that. You need to get expert clinicians saying, look, if somebody's coming in then and you're worried that they have a heart attack or my cardial and park and taking place, what are the things that you want to know, that you want in all blood pressure, you want to know, study cardiac NZIYES, you want to know you want to do an Akg and see if they're ekgchain. You have to have clinical experts to agree about what data is important to collect in order to drive and an important decision support process. You know, and and that makes perfect sense right. You need to know what you need to know when the situation is happening. So how is informatics being tied into the Mrs in a practical and a practical application? Is it literally I'm a I'm a physician, I'm walking into a room, I'm talking with the patient, I'm putting down information and checking it off, and informatics behind the scene is is listening and not listening, but taking down what I'm saying and making recommendations. Have we gotten to that point or is it more something where we still have to go take the symptoms of the patient and go look at what the data tells us? So people are experimenting with with a scenario like you described. I mean people talk about it. One one phrase that they've used is that, you know, the the hospital room of the future, if you will, where there's there's a computer running in the background listening get the patient is saying what the clinicians and saying. There's also artificial vision, so you can actually see, you know, that a blood pressure was taken or you can you can do interpretations of the actual image of the patient. You know, those sorts of things. But that's that's not in common youth. That's that's in an experimental phase. So we're in kind of a middle ground because there are lots of things that don't have to be said. So all of the betside instrumentation. You know now blood pressures, heart rate, oxygen. You know percent oxygen saturation pulled, talks information, all of those things. Respiratory rates are collected by instruments and go directly into the record where they're accessible. You'll also have automatic monitoring, for instance, of intravenous fluids and and IV pumps, because the computers know what kind of drug are infusing, they know the rate at which they're infusing that, all of that kind of information that can go in there. If you're talking about ventilators, ventilators are connected and so I know that. You know what is the what's the number of respirations per minute that the ventilator is giving and what percent of oxy each other they on. How much pressure am I using to be able to ventilate the patient? So you get a whole bunch of information like that. But there are things right now still either the patient and nurse or a physician has to say, you know, is the patient having pain? Where is the pain? You know, all of those kinds of other things, like proscilitation of the heart, you know, listening for a heart murmur or listening whether somebody may have sounds that are consistent with pneumonia in their lungs. Postings still have to be entered by a person. You know, you can make easier or harder sort of data entry screens to do that, but right now a lot of that sort of sign and symptom physical exam data still has to be entered manually. But if you do that real time or close to real time, then the computers are watching in the background for that and so, for instance, in the in the case of detecting sessis in the emergency room, the computer will have, will have data already in terms of the patients saying something like, you know, the front the admitting clerk, if you will, in the emergency room...

...will ask well, what's what's the reason that brought you to the emergency room, and they gets say well, I've had a fever and nausea and vomiting. They all immediately get vital signs information and if they've had a fever, probably get a white blood count other things, and the computer will see those things immediately as they're entered, as they happen. So you really are doing really close to real time analysis by the computer. Right, I suggest then. Okay, I've got enough data now to say that this patient, it is a high risk of sepsis. You need to give a high volume of fluid, you need to drop blood cultures, you need to start the biotics, etc. So it's happening, you know, sort of real time, with with data that's available to the computer real time. And when you think about the application of informatics over the last, you know, ten or fifteen years, where have you seen the biggest advancements? Where have you seen that either her you know, heroic work or that piece of thing that is really making a difference in the medical profession? You know, in some ways it's having the biggest impact on common and Munday and things. So one of the one of the things that it's been done was the created a program to monitor patients who are on chronic and cooagulation. So they're set of patients for, you know, for different reasons. A lot of them related to heart and heart failure or a rhythm is where our patient needs to be antecoalulated so that they don't have get plots and have those clots flow up and give them a stroke or cause other kinds of problems. And but the drugs that we have to to anacoagulate people are dangerous. Give too much of them. Then you can get a hemorrhage and they they can bleed in you know, you can get a hemorrhage and causes stroke or they can they can bleed internally and have, you know, terrible complications from that. And so it's important that you monitor it closely and we found that we could write computer programs that would monitor the patient. It would and the sort of many different dimensions. You know, what's The doose with medication? There on what was their most recent lap dated? That indicates how how fast their blood is clotting. But other other things, like one was the last time we tested them, and they are rules, basically to say you know you need to if you make a change in the medication, you you make that change, then you wigh a certain period of time, then you collect another sample and if that looks good, then you can wait another week and then you test again, and then you wait another period of time and test again and then you make sure that you're routinely testing. So the computer watches and says, Oh, not only am I helping you out with the dosing of the medication and saying what you should whether you should increase or decrease the dose, saying you need to check with this patient because they haven't been tested and so you don't know whether they're at you don't know what danger there it because they haven't that tested. So it follows up with the patient. You know to to do that and so that you know checking for drug drug interactions. I mean that's one of the first first things, preventing patients from getting medications they have a known allergy to. You know, again, management of community acquired pneumonia, knowing which patients you can just prescribe an antibiotic to and send them home, versus which patients need to be hospitalized, versus which patients need to go directly to the intensive care unit because of their situation. That kind of guide. So it's a combination of those things. It's not not like prevention of all heart attacks or the cure to cancer. It's improvement in care, sort of one situation at the time. Yeah, looking aproach it. Yeah, okay. I mean that brings back, though, the sort of the motivation for this at a grand scale. In general terms, you know, I went into this because I've got computers. would be a great application, or one of the best applications for that kind of technology would be for helping patients get better faster and learning...

...from what what we do. My real motivation, if you will, it's because, unfortunately, we're not taking very good care of patients. Multiple articles have been published that would indicate that there are at least two hundred and Fiftyzero patients who are dying from preventable medical errors each year. Wow, and that's that's a staggering number when you think about it. I mean the COVID covid now in a year is reaching that level. They're almost nothing else that's going on in our lives that it is at the same level of risk as medical errors. So, you know, you talked about nationwide. To put it in perspective, nationwide there are on average, fortyzero deaths per year from automobile accident. So that's that's a sixth of the number of people who are dying from preventable medical errors. You know, the OPIOID crisis is again about at that same level, that on any given day they're about a hundred and twenty people who are at one time, we're dying from narcotic overdoses. And again that's one six there less, of the patients who are dying every day from preventable medical errors. If you do the math, it says that there would be, you know, roughly between six hundred fifty and seven hundred people a day who are dying from preventable medical aires. And I mean one of the curious things to say why, you know, why isn't that more of an issue? Why isn't that from was going to be my next question. Why is that not getting more attention? Yeah, and I don't know the answer. I can speculate about the answer right, but my speculation is number one, people don't believe the numbers. They sort of say, well, you know, I've been practicing medicine and I haven't seen those right, you know whatever. Or you know, they say things like, you know, go look at the Center for Disease Control reports of Morbidity and mortality and you know, you see cardiovascular disease and asthma and diabetes and all of those kind of things. Those are the top, you know, the top things. And and there's no there is no code for, you know, preventable miracle are all of the preventable medical errors get classified in that kind of thing as whatever underlying disease the patient had, right, which was not the proximal cause of death. But you know, if somebody had a heart attack and they were given the wrong medication and died, they would go down as a cardiac death, not as a as a medical error death, right. And you think so some people. And Yeah, I'm sorry, but like in it's funny you talk about that because to me that's been a kind of question of mine, which is so with covid you hear about people saying it's something. So to me it's the opposite. They're saying they passed away from covid related symptoms, which is they might have had a cardiac underlying condition and they died of cardiac arrest but it was prompted by Covid. So there seems like we are coding things in that manner, but yet not this that you're talking about in that manner. We're not saying they died of cardiac arrest from a preventable oversight, medical error. Right exactly. It hasn't been treated uniformly. Is that needs a ends? Is that an insurance issue? No, I don't think no, I don't think it's insurance. I think not from insurance companies. It's probably an issue for physicians because they don't want to be sued for malpractice. That that's kind of what I was meeting to there. Yeah, yeah, yeah, so I think there's a big part of it that would be self protective, if you will, or, you know, trying to avoid any kind of legal action. But the other thing, I think. So part of it, I think, is people don't believe the numbers that I...

...have to tell you that I believe the numbers and I believe the numbers because of my own personal experience, not only with myself but with my family members and others. And I would like to share just one experience that back in two thousand and twelve I was at a an HL seven conference. Actually I'd participated in Hil seven and creating the standards, and a woman came from Singapore to the Ahl Seven meeting and I think she was thirty two years old and her you know, thirty something, in her early s. She had run a half marathon a couple of weeks before she came and when she left Singapore she was feeling fine. When she arrived in San Antonio, she has some fever as she had some nausea and vomiting. Just felt like she had the flu, quote unquote, you know, and she actually came to a meeting the next day but she just didn't feel well and went back to a room and so we didn't hear too much. And then the next day we heard oh she collapsed in the hall and that the emergency the EMT's had become take her by ambulance and by by law within the United States, if a person is unable to tell you to take them to a particular hospital, they have to take you to the closest hospital that has an emergency room, right, and so they took her to a community hospital. It was only three or four blocks from where we were meeting and she was seen in the emergency room. You know, they do blood for lab results. They asked her her history. She told him about the fee where she had a had a little grade fear she was obviously high contentsive. So they started giving her some fluid. Because she was from Singapore, they were thinking about Exotic Oriental Diseases, you know, yellow fever, Dang you fever, you know, tropical tropical diseases, etc. And they didn't know what was happening with her. They didn't couldn't make a diagnosis, but it was obvious that she was sick and so they they hospitalized her and she was up in the hotel room waiting for the physician to come and see her and do further work up as an inpatient and she went into cardiac arrest in the room. Well, they were able to resuscitate her and then moved her to the intensive care unit and I was called because she was from Singapore. Didn't have any other and at this point she didn't she wasn't conscious, so she could direct her own health care. She had another friend from Singapore who was there that called me. I came an unfortunately, dis watched her go into renal failure. They tried to start renal dialysis, they were trying, you know, ordered blood to be given, but all of those things were very delayed and unfortunately, at two in the morning, three in the warning. She died M and it ties back to some things that we've been talking about. It turned out that she had fain O Garden Variety Sepsis. She had group be Sepsis, and I mean the sad thing again is that all of the data to make that diagnosis was in her electronic medical record right her hypotension, the low grade fever, the nausea and vomiting, the low lactic acid. She had an incredibly low white count because all of her oliver white blood cells have been consumed trying to fight the infection by the infection. I mean it was absolutely a textbook case that was missed by the people in the emergency room because they were thinking about oriental diseases and other other diseases and I'm sure they were busy. You know, they had other people in the emergency room right that they needed to take care of, and so you know, that's a person who died. Now, in confession, I have to say you know, as an intern in or resident working in in medicine, I had patients die because I was ignorant of things that I didn't know what I didn't know and I didn't ask somebody else. I obviously didn't intentionally hurt anybody. Sure, I don't think physicians do either. But no, I had patients die from preventable medical errors that I...

...personally took care of. And I can go on because the other the other statistic is that other studies have shown that physicians do the thing, make the right decision, prescribe the right medication, if you will, other kinds of things, roughly fifteen percent of the time. Yeah, it's a point, costs a coin toss, you know. Again, to go to my brother. My brother had had hyper aldostronism, which the symptoms from that are renal stones, hypertension, you know, a bunch of a bunch of things, and he was treated and he was treated appropriately for kidney stones, he was treated appropriately for his hypertension, but those things got progressively worse and it was over a twenty year period before somebody diagnosed at and again it did. That's diagnosable. You know, it's based on chemistry that's been known for a hundred years. Right, it wasn't outside of our purview. Right. No, again, another situation. My you know, my father had rheumatic fever when he was young and and went on, but he was being cared for by a general internal medicine doctor who diagnosed him he was having some you know, having some sort of episodes where he had trouble breathing and other things, and he was diagnosed with asthma. And that went on for five or six years and then he was seen by a cardiac specialist who said, Oh, the real diagnosis here is congestive heart failure related to your rheumatic kind of fever, rheumatic fear right, and so I could go on that. Those are just some illustrations, but I believe the numbers. I believe the numbers based on not only patients I've cared for but others that I know about. And so the other part of that, I guess, is that I think other people who believe the numbers they don't think anything can be done about it. And so, you know, you know, so they might feel sad that those people are dying, but they if they don't think there's anything they can do about it, it's it really is, you know. There you can't sort of just be worried about things that you can't change. Sure they don't, you know, and that's my real motivation, because the kind of advanced decision support we're talking about. Is The answer to that? Well, that's what I was just going to say. So do you find that there is a resistance to the use of the data? I mean, in a sense are they are physicians who say the computers can tell me. I can tell better. You know, it's good old fashioned doctoring. Is there resistance to the use of the data in some places? And it's interesting, I mean it's sort of like any other kind of situation in life. Trust is developed by repeviously meeting, you know, a person or a situational expectation. So I trust people because, you know, I can tell you how much I trust people to show up for a meeting on time based on how many meetings I've held with them and how many times they came on time or didn't come on time exactly. It's sort of the same thing with with decision support. You use it. You know, clinicians use the decision support. If it's well done, in active it. Their confidence in the decision support builds over time and they use it. And so a lot of our experience, for instance, would show that when we first started using drug drug interaction reporting, the compliance of clinicians to that advice was somewhere around thirty percent, as I remember right, and a couple of things happen. One is we looked at you know, a lot of that was justified because the decision support was not that good. Yet we're reporting on drug drug interactions that were sort of theoretical, but all of the clinicians would say, yeah, that could happen, but what it's very unlikely and what we're treating in the patient, it's more important that they receive these two drugs, even though there's a theoretical interaction. We cleared those ups so that they were really only being told about things that they...

...really should take an action on. And if that happened and as they got experience with it, then compliance to the advice rose to like ninety percent, you know, of advice. But so it most clinicians think of it as a tool and if it's a good tool they're going to use it because they want to provide the best medical care possible. There are other clinicians, and it would be a minority of Colinissians who are, you know, arrogant and just arrogant. This arrogant. I'm trying to think about. That's the right that sums it up. Yeah, so in the time we have left, I just I have a couple of the things that have just kind of come come to my mind. So, as we see, medical and the health care system in the US is changing and there's a lot more data being used. The insurance companies are using a lot more data, sometimes right, sometimes wrong, and their calculation of treatments and costs and expenses managed. You see now more and more the insurance company saying we want treatment done the right time, right way. First, it's the patient returns, we're not going to pay for it, and that's being using data to make those decisions. Do you think you know, going back to what you just talked about about preventable errors, do you believe at all, are you beginning to track physician accuracy? So what came to my mind was, so I am a physician, I'm with a patient and I look at the symptoms, I make a diagnosis and the computer says you're right or I agree. All right, concur do you believe they'll be tracking of that? We're almost physicians might start to get scored in a way. I can you know, I can go to my Google look up and look up Dr Ted and it says Dr Ted is a seventy five percent accuracy in that. I mean maybe the I'm going a little far fetched with it, but do you think it, the tracking of the data, might get to that point? I think it will. I don't think it will stone just because we don't have decisions support for so many of the things that are going on. I mean, one of the things that we didn't talk about is that if you count that the things we did, it kind of dinner mountain. We had a hundred and fifty important kind of decision support applications. Without exaggeration, there's the possibility of tenzero decision support applications. Okay, because the things that we did with the hundred and fifty, we carefully selected them based on the potential harm or benefit to the patient, the number of patients that would be. You know how basically how prevalent the disease is. So what's the opportunity to do good? The cost of treatment or the cost of air, the actual money costs that could be saved by what we do. And so that you know that that's led to the things that we did and you know all of the things that we've done. You know would be in the top thirty kind of diagnosis that are going on. But we're not doing anything for fiber biologia, we're not doing anything for Ashimoto's disease, we're not doing anything for hyper Oupostin, is it? So? You know, it's only there's still so much ground to be gained that we can't be there. Okay, yeah, and so on. You know, if you're doing a physician report card today, you know, it's like a physician is taking a course, you know, once a year, you know, so you can give him a report card on one patient a year, you know, or something. You know. Okay, so the sense, but I think it'll come to what you're saying. And people speculate, like this is taking a little bit of a tangent, but they speculate that at some throucial point, and it's probably it's probably technically possible now, if we exerted some some effort, that they would come in instead of you doing what you do now, which is report bad outcomes to the government, you know, to see MS and other people, and you're required to do that reporting in order to be reimbursed by the insurance companies and by by Medicare and Medicaid,...

...etc. Instead of doing that, you actually come in the data available electronically and you use a program to evaluate the quality of care that they're providing. And it would be a different sort of algorithm because what it would do, basically, is it would look at the decision points, you know, it would look at the data and say, okay, here's there's the data on which this diagnosis was was founded, here's the action that was taken. Was that the most appropriate action? And if it wasn't, then you get it to merit for that, even if the patient didn't die or didn't something else. And so the quality of care being provided would be evaluated in a totally objective manner by a set of rules. Of course, the rules, the rules, wouldn't be perfect and you would want to fix them over time and sure, all of that kind of stuff, but what happens right now is it's a little bit like the hospitals. The hospitals in that kind of error reporting are not unlike us in our income tax form. I mean they tell us the rules about how, you know, if your self employed, do this, if you're you know, if you've got this much income, and and this is a tax exemptible thing and all that, but you have to figure it out and in fact we as individuals, we probably lose money every year because we're not as good at it as a professional would be figuring out. But there's not, you know, there's the government doesn't supply us with an objective way of saying hey, you know, give us all the data and will calculate the tax that you all they do that as a an experience for us. I would must rather have that kind of and that's a situation with with the hospitals, is they're given a set of rules and they interpret the rules to their own best interest. Yeah, and so the numbers that you see from different hospitals are not a reliable and objective measure of the quality of those institutions. Understand, yeah, exact. So, anyway, that I went off on a tangent, but yeah, so let's just spend a few more minutes as we wrap up here. First all, tell me a little bit about what are you seeing with AI, how is ai being applied to informatics and what's the future looking like? They're and changing it and how's it changing the face of it? So Ai and you know, closely assorty associated. You know, machine learning, deef, machine learning, all of those kinds of technology you have incredible. They're providing the trouble value now and it will just increase in the future. So, you know, getting algorithms that are smarter and smarter about detecting that and of course we're getting more and more data so that the algorithms have a better and better basis for making decisions and detecting otherwise unrecognized associations between events, etc. And so it's having a tremendous impact. You know, the things that are short of eye opening. I know of a publication where I think it was scientists with Google who took who looked at data for patients, for a patients who are admitted to the hospital, and they came up with an algorithm that was basically, at least for the for the set of patients that they were that they used in their test case, was a hundred percent predictive of patients who would die during the hospital admission. Wow, so that's that's pretty amazing. Yeah, it's pretty I can and there are other things like that that are going on and we'll just see more and more of that. But it comes back to the thing that I said earlier, there to two aspects to the impact of any of this kind of technology on medicine. One is learning things and the other thing is creating interventions or or creating direction and so that clinicians do the right thing. And so you know, the next question I would ask is, okay, so we know these four people are at risk of dying. What do I do so they don't die? Right, that's an entirely different kind of knowledge now that I need to incorporate in a program that I can apply to those patients and guide, guide their care so that they don't die.

And so that's is another area that we're going to see huge expansion as as people apply AI knowledge real time to be now the patients that have been identified for for management. And so I see that is being incredibly important area of research within medicine. Probably the only thing that's comparable would be what we're learning about genetics and the Association of James With with disease and condition to Tetra. Yeah, well one I would think that even some of this data will start to influence the insurance companies what they call the the normal course of treatment. Right, you hear about that where doctors are saying, well, I've got to do this test next because the insurance company says that that's the normal course of treatment. So exactly, you know. So maybe some of this day will start to change that and hopefully maybe for a better is it's a better but you know, when money gets involved, everything changes. So well, say something. You know in my situation that might be surprising and that is the biggest challenge here is not technical, it's actually social political. HMM. Okay, so the kind of change that you're talking about would happen. You would use more decision support, you would do it more cost effectively, as we get into more paper for performance, your payment for the quality as opposed to payment for volume, for procedures and medications, etc. Right because right now the motivation, that the monetary motivation incentive for for healthcare systems and is is to do more, to do more surgeries, to do more interventions, to do more tests, because they get paid for it and they make a margin on this, on the work that is done. Yeah, on the other hand, if you're in a situation where you say, look, we're going to pay you this amount of money for this this set of patients, and you take the best care you can of them and we're going to monitor the quality of care that you provide. MM. Then you've created a motivation for me to do the test in the right order because it costs less, right, and it causes that the patient less concern. You eliminate unnecessary procedures. You, you know all of that sort of stuff. And if there was that motivation, as that motivation grows, as people focus on value based care, that motivation will accentuate the importance of decisions support in care, because now it will become important for the financial bottom line of institutions, where right now it's not. No, it's got right. Then, once the money motivates it, the action will follow, right. Yeah, so exactly the once we shift that, that'll go. So let's talk a little bit. Number One. Thank you for all this has been awesome. There's so many places I can keep going. I don't know that everybody would listen to us forever, but it's very interesting, especially if we could talk about tell a medicine and all those things that we haven't even been able to touch upon. But I guess as we close out, when I think the listeners might love to hear is from your view of the over thirty years of your experience and the work that you've done? What's your view of Covid? It's a big broad question, but you know, what do you what do you see and what do you think is coming? I think a lot of people want to hear what do we think is coming with these vaccines and and where we've been? Yeah, well, I guess the first thing I would say is that we haven't been able to manage covid as quickly as we could because we weren't able to share data as quickly as we could. Should have been able to, and this is a it's a chronic problem the standards that are available for sharing data for something like covid or if you go historically, sharing data about Zeka, sharing data about the first the first covid right epidemic, sharing data about AIDS. You go into all of those, it's the same situation.

It was technically possible to share data, but we did we didn't have the systems in place. Yeah, and then, fortunately, what happens is everybody, basically as soon as the crisis is over, people say what we got? We've got other places we need to spend money, and they don't. They don't spend the money to create the infrastructure that prepares us for the next epidemics or pandemic. So I hope that we change that because we could learn so much, and I mean people are learning and of course people have been learning, but we would have learned so much and so many things faster. Yeah, if in fact we could, we could correlate data across all of the United States and do it in the same day and real time while we're taking care of patients, instead of waiting till the end of the day or the end of the week when we port statistics to the government. So we would be faster. We would have been better in providing covid care if we could share data better. The next phase, if you will, is exciting because now we're going to shift from, well, I don't know that we're going to shift so much as they're going to be an entirely new emphasis on the efficacy of the vaccine running and we need data for that too. We need to know who got the vaccine, when they got the vaccine, that they ever get the disease, where they exposed to the disease. You know, did you have any complications from the vaccine, you know, trivial things like, you know, just swelling in some and a low grade friever, or did you get something serious like theon Beret or some other you know, serious complication? And so there's they're already people rushing to say we need to create standards for how we shared data about the about the efficacy and the outcomes and the complications from vaccines. Yeah, and what do you and the thing I keep wondering too, is, at least in my mind, it's one of the first time is where we have, for lack of a better term, a gold rush and we have multiple players. Who and what is going to determine whether I get the fiser vaccine, the astras Nica Vaccine, the Maderna vaccine? I mean, what is going to WHO's going to track it? mean is it going to be? You know, I have pictures of the Big Board in Vegas and you can take the odds which person, which one of these, is going to be the bigger winner. And obviously big money is all involved. But how, I wonder. I wondered. I mean, you have any thoughts? I mean, how are they going to determine who gets the vaccine from what supplier. I don't have any great inside into that. I mean I think to a certain extent it's going well. It's certainly going to be a combination of factors. You know that there is scientific data behind the you know the trials that have been going on now, and certainly people are going to trust a thirtyzero person trial more than they trust the ninety eight person trial. Yeah, but they're all reported ninety percent, so they're yeah, odds are even again now. Well, well, there they're reporting ninety percent. That the are bars around the ninety percent that were tested on ninety eight people is a lot different than they are bars sure, around ninety percent for the people that were tested on Thirtyzero. So size still have a big IMPI side of past. You know all of the other pragmatics are going to need come into play. You know the things that have to be super cooled, essentially, and transport of that great expense and difficulty. Legis if it's something that can be logistically is easier. That's going to how fast they can produce it, how many doses they can get out quickly and get to people. And then politics. You know which countries did it go to and price you know, how do we price this so that we're competitive with the other people and Garner market share and, in today's climate, to socioeconomic aspects of the whole thing too? Absolutely, yeah, absolutely, so it'll be a combination of fact. It will be kind of fun. Like you said, maybe you can get up your own bedding pool come out.

Who's going to be the winner? Yeah, yeah, well, listen. I wanted to thank you for your time and your insight. It's been fascinating. It's an aspect of healthcare that I don't think many people even know about or have heard about, but obviously critically important and tremendous up sign opportunity to even make things better as we continue to to modernize and use technology to help us provide better healthcare. As I typically end all of my episodes with, I'd like to ask you question, which is, as you were either growing up or now, who is your hero? You know, I would I would say a couple. One one is, I think, following in your pattern, I I would say my father. My father was an electrician. He work to the steel mill. He, you know, fixed huge motors that will use, you know, in cranes and you know, all kinds of equipment and then he that he was also then use that electrical capability and he was a master electrician license by the State. So that, you know, we work together basically my whole life doing, you know, residential electrical works and also small commercial work, and I just learned so many life lessons, you know, from that and so much of, I think, the way I think about things came from working with him, working in that environment in many ways, being independent at an early age and doing a lot of things that he would just say go do this and was up to me to figure out the best way to do it, and I learned from that because I also suffered the consequences. I mean, if it wasn't right or it didn't work, people would call you up and say you need to come back and do this right, you know whatever. And certainly and then he would just so fun to be around. I mean I just have to say that. You know, a lot of life is about is about fun and enjoying life, and he certainly taught me that as well. Yeah, that's awesome. At a professional level, you know, they're that. They're been a number. I already mentioned homer Warner. Yeah, another medical informatitis that I work with and was probably my closest mentor his name was was alt prior, and he would just he was just a wonderful and had incredible understanding. But again, one of the thing he would just so fun to be around. So why you were learning all that stuff? You know, it was Everard dull when when Al prior was in the road, always had something provocative to say, you know, or something something to ask you about or ask you how you were doing or to give him the demonstration of what you were working on. So yeah, I would say Homeo Warner and then and then I'll prior especially just an incredibly bright, fun guy that taught me so much about informatics. That's amazing. That's amazing. Listen, this has been a pleasure. It's been a delight. Thank you for your time, thank you for your insights, thank you for your service. I appreciate, we all appreciate the work you've done, the pioneering work you've done in and informatics and how it's making the medicine world better. Thank you for being a part of heroes of healthcare podcast and we just look forward to continuing to follow up and talk with you and learn more about informatics and what in your mountain is doing. So thanks so much for being a part of this stand. So honored that you would ask me Ted. It's been a pleasure for me and I'm gratefully you would put up with so much technical jargon and and other things. So thanks so much for having any interest in this our pleasure. So all right, you've been listening to heroes of healthcare. For more, subscribe to the show in your favorite podcast player or visit us at heroes of healthcare podcastcom.

In-Stream Audio Search

NEW

Search across all episodes within this podcast

Episodes (40)