Curing Our Diagnostic Disorder
The disease has a variety of symptoms: medical errors, poor utilization management, professional misunderstandings. Could the diagnostic management team prove to be a panacea?
Michael Laposata |
We are all aware of the key issues that pathology faces in patient care – diagnostic errors, failures to communicate, poor test utilization, and even miscommunications and misunderstandings between the bench and the bedside. And, of course, we are also all in agreement that these issues need to be fixed. Unfortunately, that’s where the consensus ends. How do we address these overarching problems? How can we improve communication, education and patient management across disciplines and specialties?
We already have a tool that allows us to discuss cases with clinicians, review treatment decisions, and reconcile them with existing test results and patient outcomes – the case conference. But those meetings take place long after the relevant decisions have been made. Why? Perhaps it was necessary in the past, when specialists had to travel to consult with one another, or wait for days while test results were mailed from one site to the next. But nowadays, we have the technology to assemble a team of experts from various specialties, either in person or remotely, and discuss a patient’s treatment not after, but as a part of, the decision-making process. I call this the “diagnostic management team,” or DMT.
Three key features make the DMT unique:
- It takes place in real time.
- All of the information is written into the patient’s medical record.
- It results in clinically valuable information.
What exactly is “clinically valuable information?” To provide an example, let’s say I spot a band of abnormal hemoglobin on a patient’s capillary electrophoresis. To get paid, the only notation I need to make is, “abnormal band seen.” But that’s not informative; it won’t help with the patient’s care. Instead, I should say, “There was an abnormal band of hemoglobin; it was in this region of the gel; it was later identified as hemoglobin Köln; and the patient is (or is not) likely to require transfusions.” The DMT gives us the opportunity to provide value-added data – and because we’re doing it in real time, the clinicians get that information in time for it to be reflected in their treatment decisions. All too often, the pathologist has information that could be useful, but because he or she is never asked, the doctors never hear it. That’s what I want to change.
“Do it my way!”
There has been one barrier after another to implementing the DMT in hospitals – and most of those barriers can be explained by the incentive that all doctors have at this particular time and times in the past: money.
My startling first experience with colleagues who didn’t know how to evaluate partial thromboplastin time (PTT) – a blood clotting test – was in 1984. I was a resident in clinical pathology, and I received a call over my pager from an internal medicine resident who was particularly smart, and I admired him for seeming to know everything. He asked me a simple lab test selection question: what subsequent tests should he order for a prolonged PTT? I would have assumed he knew the answer – but when I discovered that he didn’t, I realized that I had a body of knowledge about making a diagnosis that he and most residents in other specialties did not have. I asked him what he had done to evaluate prolonged PTT prior to our conversation, and he said, “I did what every other doctor did – I guessed which test to pick and then guessed what the results meant.” His response reminded me that there was very little teaching about test selection or result interpretation in medical school, even though doctors do it every day.
I went straight to my professor, who dealt with coagulation, and said, “We have to develop an interpretive service similar to the one offered by radiology.” They don’t just hand back a film and say, “You figure it out; call me if you have any questions.” They automatically interpret their results, so I recommended that we do the same in pathology. His answer to me was, “I have to become an associate professor or I’ll lose my job. And you don’t become an associate professor by taking care of patients – you do it by publishing papers!” As a trainee, it was a shocking revelation to hear my mentor admit that patient care was not the primary responsibility of a hospital physician. But at that stage of my career, what could I do?
Well, I got my first job: director of the coagulation laboratory at the University of Pennsylvania. Now that I was on the faculty at a major medical institution, I designed and implemented a new diagnostic test result form: a page to be inserted into a patient’s chart that had not only blanks for the lab test results, but also a box at the bottom in which to write a patient-specific narrative. It was incredibly well received – so much so, in fact, that the hematology fellows were no longer receiving as many consults.
Unfortunately, the response to the new service also prompted the Chief of Hematology to come and ask me to stop providing interpretations. Of course, I asked why; after all, the information had obvious benefits, especially for clinicians. His answer was, “My hematology fellows aren’t seeing as many patients anymore, so my revenue is down.” Naïvely, I asked if that wasn’t less important than an immediate diagnosis – I didn’t want to delay treatment. Nevertheless, he said, “Do it my way!” That was the end of it. He was a full professor and I was only an assistant professor, so I knew I couldn’t win.
“I don’t need this!”
The only solution I could devise was to keep moving up the ladder. Eventually, I joined Massachusetts General Hospital as director of clinical laboratories and associate professor at Harvard Medical School. Another step upward; another attempt to implement diagnostic interpretation. I initiated an interpretation program for complex coagulation cases; it was an even bigger success this time around, because the field was becoming rapidly more challenging with the introduction of new anticoagulants and other coagulation-related therapies.
Not everyone cottoned on right away, though. There was one hematologist in particular (subsequently my best professional friend, I have to add!) who came in with an interpretation I had created and said, “I don’t need this!” I said, “Yes, I know, but most of the doctors in this hospital do need it,” and he crumpled the paper up, threw it at me, and left my office. Yet again, I could have backed down – but I didn’t. It was becoming obvious that the few hours of education medical students receive in lab test interpretation was never going to be enough. Somebody was going to get hurt.
So we persisted with the interpretive service, and eventually created a program that became a source of national and international attention because of its clinical value in establishing a rapid, accurate diagnosis. In fact, the interpretations in the chart – attributed to us as pathologists – actually led to patient referrals to our pathology service from other doctors in the area, and we developed a widely respected clinical practice for patients with bleeding and clotting disorders.
The patients themselves were very confused to be referred to a pathologist, though, with some even asking if they had been referred in preparation for an autopsy! I assumed they were joking, but I’m afraid that just highlights the need for us to educate our patients and the public about what exactly it is that we do. Every pathologist should be enthusiastic about spreading the word – and we should all embody the three ‘As’: affable, accurate and available.
A technological leg up
In 1996, I gave a talk to my colleagues about the DMT and what I hoped to accomplish. I happened to mention that I used a Dictaphone to record my cases; then I sent them out to be typed, and when they returned, I edited them. A time-consuming process, especially with 20 or so cases per day! But that day, for some reason, there was a software developer in the audience. He took me aside after the talk and said, “There’s no need for all those steps. I can make this much easier for you.”
We ended up working together to develop software that provided a customized bank of pre-written comments. That way, I could just select the comments I needed for any individual case, and then add any unique information afterward. We still had to individualize it for each patient, but the software gave us a head start. In fact, it made us fast enough to sign out 30 or even 40 cases in a single session, enabling many more interpretations of patient records.
Our ultimate goal is now to complete what we call a “what’s in the box?” project. We’d like even the smallest community hospital to be able to contact us and say, “We’re interested in setting up or connecting to a DMT.” And then we’d send them a box. What’s in the box? For one thing, the software I’ve just described – something to allow them to enter comments, connect with other experts, address billing and payments, and anything else they need. If we could assemble a DMT starter kit like that, then anyone could set up a team. That’s our goal – and I think it’s well within reach.
From concept to reality
I joined the faculty at Vanderbilt University as pathologist-in-chief in charge of both anatomic and clinical pathology. The role came with a wide range of opportunities, but most importantly, the opportunity to truly begin building a diagnostic management solution. That’s when we came up with the name – “diagnostic management team” – and it’s when we began implementing them in microbiology, hematopathology, coagulation, and transfusion medicine. Each of these areas claimed a triumph in improved patient care and decreased cost. It became abundantly clear that, with thousands of increasingly complex laboratory tests available, diagnostic expert consultations were necessary to help our bright and well-intentioned physician colleagues.
About five years ago, there was an unidentified epidemic in the United States. We had a patient fall victim to it in Nashville; she presented with symptoms of meningitis and was transferred to Vanderbilt. The infectious disease doctor referred it to the microbiology DMT because she had no idea what the causative agent might be. It took the DMT about a week, but they figured it out – Aspergillus fumigatus, a common disease-causing fungus in immune-deficient patients. But our patient wasn’t immune-deficient. We eventually discovered that she had been caring for her invalid husband, injured her back, and had an injection of prednisone – and it turned out that the injection was to blame. There were birds flying around in the facility that manufactured the prefilled prednisone syringes, and the syringes had become contaminated with Aspergillus. Thanks to the DMT’s identification of the problem, the company stopped production of the syringes and the epidemic ended (2).
The economic angle
After we started the DMT program at Vanderbilt, they brought in a business expert who knew nothing about medicine to determine whether or not it provided an economic advantage. Specifically, they wanted to know whether the DMT process reduced length of stay – because it costs about US$2,000 a day to keep a patient in the hospital. So if an extra $20 lab test can discharge a patient even one day earlier, the overall cost is massively decreased.
It’s amazing how many people say, “I think I’ll just see the results of the first lab test, make a decision, and then order the second test if necessary.” By doing that, they’ve just added another day to the patient’s stay – and another $2,000 to the bill! So why not have a set of relevant tests and order them all at once? Not every doctor feels confident enough in his or her laboratory knowledge to do that – and that’s why you have an expert team to help!
So the hospital asked the business expert to check every one of the codes for a diagnosis-related group and see if the length of stay changed for any of them between August and December of 2011, immediately after we started the coagulation DMT. He said, “There’s a significant reduction in the length of stay for two things – but because I’m not a medical expert, I don’t know if either one is related to coagulation.” I sat there thinking, there are thousands of codes – please say something coagulation-related. And it turned out that the two things were pulmonary embolism and intracranial hemorrhage – our bread and butter!
We then realized just how much impact the DMT was having, not just on patient outcomes, but on the hospital’s bottom line as well. But unfortunately, the gains weren’t coming directly from the laboratory budget, and hospital managers think of budget success or failure by department. They’ll look at the laboratory budget and say, “What? Ten percent more tests?!” and ask us to reduce that number. It doesn’t always matter that those relatively inexpensive tests save thousands of dollars, if the managers don’t make the connection. You have to look at the budgets globally to see the difference, but too many managers don’t understand that a little extra in one budget can result in saving a fortune on another. Clearly, resources like clinically effective DMTs can play a significant role.
A friend of mine was an ear, nose and throat surgeon at the Massachusetts Eye and Ear Infirmary while I worked at Mass General. He performed a lot of tonsillectomies on eight- to ten-year-old boys and would often request coagulation tests to ensure that they would not have bleeding issues after the procedure. He said that he often found a prolonged PTT.
When he received an abnormal PTT result, he had two choices: i) call a hematology consult, which would usually take two days, or ii) transfuse two units of plasma. Most of the time, he opted to transfuse so that he could complete the procedure and discharge the child. Unfortunately, he was doing this between 1981 and 1984 – a time when one in every 20 bags of fresh frozen plasma was infected with either HIV or hepatitis C. We didn’t know that at the time, of course – but these children were put at risk because their surgeon didn’t know what test to order.
My friend admitted to me that he was overjoyed when, in 1995, we began providing interpretations along with our test results. But then he considered how many people he had unnecessarily transfused. His “error” didn’t come to light when his patients were eight years old, but many years later – when these children had become young adults and needed liver transplants or received diagnoses that would change and abbreviate their lives.
And yet, nobody (until now) has pointed out the root cause of this problem – the fact that somebody didn’t know what laboratory test to pick because of an inability to interpret a prolonged PTT. In short, a major diagnostic error led to patient deaths.
I wish I could say that this was the only example of such an error but, unfortunately, it’s just one of many.
I’m currently Chairman of the Department of Pathology at the University of Texas Medical Branch (UTMB). That was a calculated move – Texas was once its own country and still believes in itself as the strongest state in the union. It has a strong sense of “can-do”, which is exactly what we need to fully realize the potential of the DMT. We also have a powerful medical system – eight campuses, for which we can share diagnostic experts to ensure that each campus has access to every possible area of expertise while working remotely. We have four DMTs in place right now, but if we can get all of the campuses linked via telemedicine, we’ll have dozens!
We’re piloting the telemedicine approach at the moment. We have a designated DMT room with screens and cameras so that we can see our long-distance collaborators and they can see us. If we gather all of the experts from all of the medical schools involved, we can cover almost the entirety of diagnostic pathology in a single team. We’ll have experts in virtually everything! How do we help a patient in a town without a university hospital? We may offer a subscription service or a “pay for what you use” model as we move forward – so if a clinician is uncertain of the correct diagnosis, or in some cases is unsure how to treat the patient, they can dial a number and reach the DMT.
We’re lucky that we have people here in Texas who are willing to put enough money behind the project to get us through testing and put the full system into operation. It doesn’t cost a lot, at least on a healthcare scale (maybe a few hundred thousand). Health economists have referred to DMTs as “the missing link.” Texas is a great proving ground; with a population of nearly 30 million, you can imagine how many hospitals and primary care clinics need diagnostic support on a regular basis. The endgame is to provide accessible services to all of those institutions to obtain better patient outcomes. We should never allow our patients to die because there is no expert in their immediate environment.
Once we have developed DMTs for dozens of clinical areas and can bring diagnostic experts electronically to the bedsides of patients all over the world, we will have achieved a goal that started with a simple query from one resident to another in 1984.
I’ve been invited to speak at more than 50 different medical schools and over 20 different major societies about the DMT. Recently, the Institute of Medicine convened a panel of 21 experts to put together a report on improving diagnosis in healthcare (1); it’s a summary of all of the literature on diagnostic error, and it speaks loudly that there is a major safety problem in healthcare that must be addressed. So, with everybody saying that the DMT is a great idea and highlighting how badly our patients need it, why has nothing changed?
The biggest reason, in my opinion, is the lack of incentives for pathologists in the United States. Right now, we are paid handsomely for anatomic pathology work – recent statistics from the College of American Pathologists state that the average pathologist works about 48 hours a week and makes US$250–400,000. If you’re interrupted to answer diagnostic questions that don’t involve a microscopic diagnosis, you lose money, because those events are only worth about $25 each. If you go to the emergency room and stop a patient from losing a unit of blood an hour through a chest tube, it’s worth $25. So obviously, pathologists here are incentivized to focus on anatomic pathology.
I recently visited a prestigious university to give Grand Rounds. They had a Department of Laboratory Medicine with experts from many specialties. I asked, “Why aren’t you taking an interpretive approach to report development?” and they said, “We don’t have the right IT infrastructure to do it.” I was astounded! But here’s something that surprises me even more: to this day, nothing has changed. The department is just throwing the test results over the wall like they’ve been doing for the last century, even though they have all the necessary people power to get it done properly!
Despite obstacles like these, I haven’t lost any of my enthusiasm. It may take a while for the idea of the DMT to become commonplace, and even longer for it to be widely implemented – but after all, Rome wasn’t built in a day. I’m going to keep working on this until I can’t work anymore.
The external environment in the United States is changing. The pay-per-click model is going away, and different methods of payment for healthcare delivery are taking its place – and that’s good news for the DMT, and good news for patients. Why? If anatomic pathology alone isn’t as lucrative as it was before, and you as the diagnostic pathologist are responsible for getting involved in microbiology, transfusion, endocrinology, coagulation, and other areas, then you’ll have to adapt quickly. If you can’t, you will be ill-prepared for this new environment. I’m looking forward to that day, because I truly believe – and the Institute of Medicine agrees – that DMTs can solve many of our existing issues with diagnostic error and improve outcomes for all of our patients.
Will pathologists go for it?
Not the old school, no.
And I’ll admit that it’s not an easy change for a pathologist to make. Instantly, you’re being put into the firing line. You are now accountable for a much greater medical challenge – integrating large amounts of diagnostic data. Not only that, but it’s a big change to your day-to-day routine as well. For example, I have to take phone calls to discuss my recommendations and interpretations. I have to attend team meetings. It all adds hours to my day. My medical liability has changed now that I’m in the middle of this team sport called “making a diagnosis.” It’s far easier to sit on the sideline and do what you’ve done for years.
A Dangerous Differential
At a diagnostic management team meeting in the late 1990s, I was presented with the case of an infant who developed a subdural hematoma. Her father was accused of shaking her to produce the hematoma; despite his defense that a minor fall had caused the observed bleeding, he was convicted of child abuse and imprisoned. Later, the same patient came to my institution – this time with meningitis. Our expert DMT clearly showed that she suffered from a common bleeding disorder known as von Willebrand disease, a diagnosis that had been missed during the previous evaluation because of a flawed interpretation of the lab test results. The fall the father had described in his defense could indeed have produced the subdural hematoma in a child with an underlying bleeding disorder. After two years in prison because of a diagnostic error, he was finally released.
A subsequent case arose when an attorney read about the first story and contacted me about his client. Again, a child had suffered a subdural hematoma – and, as in the previous case, our DMT found that he and both of his siblings had severe, previously undiagnosed von Willebrand disease.
Since that time, I have been involved in more than 30 cases associated with a question of child abuse. Approximately five percent of cases of bruised children turn out to be child abuse mimics, rather than true child abuse. Such patients require a thorough coagulation evaluation to determine whether a bleeding disorder is present. I am pleased to say that, in all but two cases in which I have explained a bleeding disorder to a judge or jury in support of falsely accused caregivers, the child has been returned to a loving home.
To me, this is the perfect illustration of the need for DMTs – a team of diagnostic professionals can look at something like suspected child abuse, find evidence of a bleeding disorder that might otherwise have been missed, and realize, “Holy cow! This was not child abuse at all. It was an underlying disease.”
I think new trainees are the key to real change. The residents at my hospital have never worked in the “old” way. They have responsibility for providing a diagnosis – and not only that; a recommendation as well. One told me, “During my training, I’ve seen and been involved in hundreds of cases. I’ve learned how to do my job and how to help patients. If you had just lectured at me for 14 hours over the month of my coagulation rotation, how would I have learned anything?!”
You can’t just put all of the responsibility for change onto the shoulders of the next generation, though. Yes, it has to fall on fertile soil, but somebody has to provide an example. The residents who come through our program realize that this is the only way to practice pathology – you must make real-time contributions and learn from real-life situations. Classroom lectures alone are not enough. And that’s why our teaching programs are successful: one domino tips another. The residents go to the laboratory in the morning and ask the technologist, “What do you have for coagulation cases?” and the technologist gives them the day’s cases. Then, they have the rest of the day – from about 8:00 AM to 4:00 PM – to call the doctors, find out more information, and stop tests that aren’t needed. At the end of the day, they show up with an interpretive paragraph on each of those cases. It gives me great pride to know that they can walk out the door and have that kind of impact on all of their patients.
- EP Balogh et al., “Improving diagnosis in healthcare”, Institute of Medicine (2015). Available at: bit.ly/1KxJPO4. Accessed September 1, 2017.
- AC Pettit et al., “The index case for the fungal meningitis outbreak in the United States”, N Engl J Med, 367, 2119–2125 (2012). PMID: 23083311.