Activity Transcript

Dr Mark Graber: Hi. Welcome everyone. My name is Mark Graber. I'm the founder and now the chief medical officer of the Society to Improve Diagnosis In Medicine, SIDM. We are a nonprofit group dedicated to reducing the harm related to diagnostic errors in health care. If you are a patient, you probably have already experienced a diagnostic error. That's why you're here. If you're a clinician, you know about these areas firsthand because diagnosis is an incredibly dynamic and complex process. We will all make many diagnostic errors in our careers, but from the SIDM perspective, we firmly believe that by understanding how these errors arise and taking some simple steps, the likelihood of error and harm can be dramatically reduced.

We got a gift recently from the National Academy of Medicine. Here it is. Improving diagnosis in health care. This is a landmark report released in 2015 and it's everything we know about diagnostic error. The report is free. You can download it from the National Academy website. The report gives us a new and very useful definition of diagnostic error. Here it is: "It's the failure to establish an accurate and timely explanation of the patient's health problems or communicate that explanation to the patient."

It's an interesting definition from a couple of perspectives. First, it's patient focused. It's all about getting it right for the patient. Secondly, it emphasizes communication. Earlier definitions of diagnostic error didn't include this component, but communication is so critical to the diagnostic process. I'm glad that it's in here and the definition includes some challenges for us going forward. For example, what does accurate mean? It used to be enough to say that somebody had lung cancer. Now we might need to say that they have a small cell lung cancer with a particular genotype. What does timely mean? We have very little guidance in medicine right now about how long it should take to diagnose a particular ailment. We need much more research on that. So the National Academy report is full of facts and figures, and I hope you take a look at it. Every diagnostic error involves a person and the possibility of harm.

So I'd like to start off by telling you a story. This is a story of Rory Staunton, a 12-year-old boy. He had a dream to grow up and be a pilot. He had fallen in love with the story of Sully and how Sully had landed his jet on the Hudson river. He had written to Sully, Sully had written him back. For his 12th birthday, he got a helicopter ride around New York City. At any rate, Rory was at school. He was diving for a ball at recess day and he got the ball, but he scraped his arm.

A couple of nights later, he awoke from sleep. He was feverish, chills. He vomited. In the morning. He was really sick, couldn't go to school. The family succeeded in getting him an appointment with their pediatrician. Here you see the vital signs. His temperature was 102, heart rate was 140 which is really up there even for an adolescent. Respiratory rate was 36. The skin was described as modeled and we don't have a picture of Rory's skin, so this is just a stock picture of modeled skin. The technical term for this would be livedo reticularis, which is kind of a striking finding. His abdomen was benign and the pediatrician trying to put this together, she'd seen an adolescent boy who's febrile, vomiting, she made the diagnosis of gastroenteritis, which is a perfectly reasonable diagnosis, and she recognized that Rory didn't look good, so she suggested that he go to the closest emergency room.

Now he is in the emergency room later that day, pretty much the same vital signs. Abdomen is still benign.

There is no documentation of what the skin looked like at that visit and again, they agreed, gastroenteritis. They gave him some medications for nausea, a liter of saline, and sent him home. His lab tests came back before he left, but from what we know, they weren't reviewed and they showed a minor elevation of his white blood cell count, but a striking elevation of his bands. 53% bands. This is atypical for a diagnosis of viral gastroenteritis.

The next day, Friday, he is home and he's worse. His parents say they had never seen anybody's skin look like this. It was all blotchy, sensitive to the touch with red, blue, and white spots. They tried to get back to the pediatrician, did not succeed. Somebody from the office suggested that they give Rory some Tylenol. The family could not stand it anymore, so late Friday night they took him back to the emergency room. He was immediately admitted to the intensive care unit ( ICU). The next day the diagnosis was made, streptococcal sepsis, which in all likelihood is what he had all along, but it was too late. He died Sunday the next day in the intensive care unit.

It's a sad story, but it illustrates many of the common features that we hear about diagnosis and diagnostic error. These are the main findings from that national Academy report. These diagnostic errors are common. They happen in every healthcare setting and may arise from the complexity of the diagnostic process and shortcomings in clinical reasoning and our healthcare systems. Here are some relevant facts and figures from the report that illustrate the magnitude of the problem. One of our colleagues, Hardeep Singh, studied the records of patients seen in primary care and found that 1 in 20 every year will experience a diagnostic error in this setting.

From autopsy studies, we know that 40,000 to 80,000 people a year in the United States die from a diagnostic error. And these errors are expensive. Estimates of the cost associated with diagnostic error exceed $100 billion per year. This is the major conclusion from the national Academy report:

"It is likely that most of us will experience at least 1 diagnostic error in our lifetime and sometimes these will have devastating consequences." We've learned a great deal about diagnostic error from studying malpractice cases. This slide is based on data from the CRICO database. Malpractice suits relating to diagnostic error are the most common reason for a claim. 34% of medical errors causing medical serious harms are related to diagnostic error. The number 1 rank. They're the most catastrophic and the most costly with a median payout of over $700,000 per case. Another thing we've learned from these malpractice suits is that 3 big categories comprise the majority of these malpractice claims, so these are the big 3: vascular conditions, infections, and cancers. You can see the top 5 diseases in each one of these categories.

Diagnostic errors happen in every healthcare setting. The malpractice data shows us that most errors arise in ambulatory care and it makes sense. This is where most health care is delivered. There are plenty of diagnostic error involving inpatients and patients seen in the emergency department. The emergency department is referred to as the Petri dish for diagnostic error. It is not a great place for diagnostic accuracy. You do not know the patient. The patients don't know us. We often have incomplete records. The atmosphere is dynamic, sometimes chaotic. There are distractions. It is just not a great place to be making a diagnosis.

One of the most helpful contributions from the National Academy report is its description of diagnosis as a process, a series of steps. They do not always occur in this order and some steps may be skipped in some cases. In general, this is a very helpful way to think about diagnosis and to start thinking about where in the diagnostic process errors can occur. So many errors involve the very first step where the patient decides to seek medical care. Many patients these days are trying to make their own diagnosis based on what they know or what they read, what they have looked up on the internet. Maybe they have gotten advice from family, friends, or the local pharmacist. Many of these diagnoses will be correct and many will be wrong.

Other errors occur in the clinical reasoning process. That is a big circle in the middle of the diagnostic process. We might miss something in the history or in the physical exam or we might not have all the right data we need. Or we don't put it together correctly. Our reasoning is a bit off. Many errors involve the last step. The communicating of the diagnosis to the patient. A Report might be lost or miscommunicated or the patient might misunderstand what we were saying.

Another one of our colleagues, Gordy Shift, collected reports of diagnostic error from over 500 physicians. And here's how they map out on the diagnostic process. It turns out we're not perfect at any step. 20% of the time there's an error in just doing the history or performing the physical exam. Most diagnostic errors involve the circle. The clinical reasoning process, our hypothesis was wrong, we didn't order the right test or the right consult and another 10% involves suboptimal follow-up. It's instructed to think about how the errors in the Rory Staunton case would map out in this way. We'll come back to it in a bit, but you can start to see how breakdowns in every step of the diagnostic process might've contributed to his death.

Let's change topics now and start to consider why these errors occur. What are the factors that contribute to diagnostic error? And the idea here is that if we clearly understood why these errors arise, we would be better equipped to avoid the next one. So let's start with this: diagnosis is an incredible challenge. It may be the most difficult cognitive tasks that humans face and I am not exaggerating.

There are over 10,000 known diseases and every year roughly a hundred new ones are identified and added to the list. I think back to when I was in medical school, we used the textbook of medicine which presented 750 of the most common conditions. So how good am I going to be in diagnosing the other 9,000 that I was never exposed to? Another problem is that the number of symptoms is limited. Roughly speaking, there's only 200 symptoms. So for every symptom there are roughly 50 possible diseases that might be the explanation. In addition, it gets worse because there is so much complexity and so much variability at play and this involves our patients, our healthcare systems, and even ourselves as physicians and clinicians trying to make the diagnosis.

One way to look at the factors that contribute to diagnostic error is to separate out those relating to the healthcare system from those arising from shortcomings in our clinical reasoning or our cognitive processing. These are data from 100 cases of diagnostic error that my colleagues and I reviewed and here's what we found. It turns out that most cases of diagnostic error involved both system related and cognitive factors and cognitive factors are identified in roughly three-quarters of the cases. So if you add up the red and the light blue. And roughly two-thirds of the time there's a system contribution to diagnostic error. So that's the sum of the red and the dark blue. We believe that both the system related and the cognitive errors are very appropriate targets for improving diagnosis.

But before we get to that, I just want to mention this very small piece of the pie, the gray zone cases, where it isn't even clear that there was a diagnostic error. Where there were no clear system or cognitive problems. So these may be cases where the symptoms and signs weren't specific enough or the disease was just getting started and hadn't progressed enough to allow for its diagnosis. In some cases the patients hadn't followed up with recommended tests or visits. So this is a small number of cases, but these are important in terms of reducing diagnostic error because in the future we'll have better ways to follow up with our patients and we'll have better tests that allow us to recognize things earlier in the course of disease.

We've seen that system related factors are present in most cases of diagnostic error and what are these problems? Well, here's a word cloud that illustrates them and if you've been involved in health care for any length of time, you will be familiar for every single one of these possible contributing factors. These are, again, data from the CRICO malpractice database illustrating that breakdowns in communication and coordinating care are by far the most common system related factors involved in cases of diagnostic error. Many of these reflect lost test results, tests or consults that weren't reviewed or didn't get done. Others reflect the physician not understanding the patient's concern or not clearly communicating the problem and its urgency. And what is normalization of deviance? These are the cases where the same system related problems just keep recurring. Like the consultant who is always slow in responding or the repeated delays in getting back send-out tests. We just get used to the way things are even when they aren't ideal, even when they aren't the way we would like them to be. This is normalization of deviance.

Let's switch over to the cognitive factors involved in diagnostic errors and I'd like to start with a famous paper from George Ford Dodge where he asked the question, "Why did I miss the diagnosis?" And the number 1 reason was simply, "I just didn't think of it." And this was certainly the case in Rory Staunton's care. They just did not think of sepsis until it was too late.

Thanks to the work of Pat Cross, Carrie and many others, we now have a much clearer understanding of the clinical reasoning process. This is known as the dual processing paradigm for decision making and it's how we all solve every new problem that we confront in life. It all starts with the recognition center in our brain. When we are confronted with a new problem or a new case, the recognition center decides really within milliseconds if we recognize what's happening or we don't and most often we do. We know the answer and it happens within milliseconds and we are almost always correct. This is system one, our intuition. The subconscious ability we have to solve problems and it's how experts think. If we don't recognize the problem or we're a novice, we have to stop and think. This is system 2, the deliberate conscious consideration of what might be going on.

So in reality, clinical reasoning is always some combination of system 1 and system 2. But this paradigm is especially useful when we start thinking about the cognitive contributions to diagnostic error and where things go wrong. For example, it's well known that our thinking is sometimes biased. If you look up cognitive bias in Wikipedia, you will find this cognitive bias codex that arranges all of the 150 or so biases that are now recognized and new ones are arising all the time. Some of these are own as affected biases where our emotions interact with our clinical reasoning. For example, a patient who is drunk, shouting, or noncompliant may negatively affect how that patient is diagnosed. Many other biases reflect the human tendencies we all have and I'd like to give you 2 examples of these. These play out in our everyday lives and they play out in health care and diagnosis very commonly.

First, there are context errors. We think we have made sense of the situation, but we are off base. It is like the pilot who thinks there's no ice on the wings, but there is and the plane crashes. We think that is our car in aisle 4 but it's a look-alike car. In the Rory Staunton case, the pediatrician and the emergency room (ER) staff heard the story of an adolescent boy who was vomiting. So they were in gastrointestinal (GI) context and if you are in a GI frame of mine, gastroenteritis makes a lot of sense as the correct diagnosis. However, it was not a GI case. They were in the wrong context.

Another especially common tendency is what we call premature closure. Our tendency to just settle on the first diagnosis that makes sense without considering other possibilities. We all do this. Whenever we're confronted with a new problem, we tend to settle on the first answer that solves the problem. We do not stop to think if there's a better answer. So in Rory Staunton's case, there was no differential diagnosis listed, even though all the findings, like his modeled skin, didn't really fit with gastroenteritis.

We have discussed the likelihood of diagnostic errors and how they arise. Let's finish up considering what can be done to address the problem. This slide summarizes the interventions that we think might be most helpful in improving diagnosis and reducing harm related to diagnostic error. The interventions falling into 5 big categories and they're all based on what we've learned so far. So we know that most diagnostic errors involved some problem with clinical reasoning and there are specific steps we can take to address these that we'll get to in just a bit.

We know that system related factors are common. There are many opportunities here and ideas on how to improve communication. The third item is about improving teamwork in diagnosis. It's another aspect of how healthcare systems impact the quality of our diagnostic work. And I know teamwork is kind of an overused concept in health care, but it's never really been applied to teamwork in diagnosis and it should be. We need to start thinking of diagnosis as a team based activity.

This was actually the number 1 recommendation from the national Academy report on how to improve the diagnostic process. Improve teamwork. What they meant was that we need more effective partnerships with our patients and with all the healthcare staff who interact with the patient, starting with our nurse colleagues. We know that teamwork works from aviation and it will work for us in diagnosis as well. Fresh eyes catch mistakes, they catch breakdowns in communication, and we need to take advantage of this.

The fourth idea is to focus on having more appropriate beliefs, attitudes and practices that could improve diagnosis. So we're thinking about things like being reflective, taking time to stop and think. Being vigilant to the possibility of error and being sensitive to what the patient's best interests are and whether we are meeting their expectations.

The last idea is to improve education about diagnosis. Right now in most professional schools, there is no course on diagnosis. We think there should be. This is an area where SIDM is particularly active in trying to help promote these courses for all healthcare providers. Let us get back to the value of teamwork. We are big fans of second opinions and we know that this works. If a second radiologist or a second pathologist takes another look at the images or the biopsy material, the diagnosis changes in a small but very important fraction of cases. In internal medicine the diagnosis changes 10 to 20% of the time with a second opinion.

Here is our suggestions for things that we can do as physicians and clinicians to help prevent diagnostic errors. All of these make sense based on what we know about how diagnostic errors arise. All of them are inexpensive and relatively simple. I'd like to emphasize the advice to use decision support resources. There are many different web based tools now available that can help us with differential diagnosis and these are especially important to help us think of the thousands of diseases we don't see on a regular basis.

They are easy to use. You put in the key findings and within seconds back comes a prioritized list of other things to think about. If you put in the key findings in the Rory Staunton's case, for example, fever, nausea, blotchy skin, streptococcal sepsis is the number 2 suggestion on the list. If you can only do 1 thing, here are our suggestions and they relate to all of the problems using system 1 or intuitive way of making diagnoses based on recognition:

Stop and think. Stop and think. The idea is to get system 2 involved. In an everyday practice this equates to making a differential diagnosis in every case, not just the complicated ones, and stopping to think, what else could this be? As part of our SIDM activities, we often meet with patients to talk about their diagnostic errors and think about what could have been done differently in their case. What if things have been done differently? This is what it might look like if we mapped out some of these what if's on the Rory Staunton case.

In terms of the cognitive factors, what if the clinicians had used 1 of these decision support resources to help with the differential diagnosis? What if someone had stopped to ask, what else could this be besides gastroenteritis? In terms of the system-related factors, what if the ER had waited to review the blood test before making the diagnosis? What if the systems had made it easier for the family to reengage when things weren't going well?

That is our advice for you. To wrap things up, I'd like to quickly summarize things that we're working on since the National Academy report. Our main efforts in SIDM are focusing on the coalition to improve diagnosis. This is a group of over now 50 major organizations, each of which has committed to do something in their own area to help improve diagnosis and address diagnostic error. So please think about the organizations that you belong to and encourage them to join the coalition and help us in this fight.

Second, we sponsor an annual conference on diagnostic error that is excellent and features the latest work in this field. You can find out more information about our on the SIDM website and we look forward to seeing you there.

Third, we have worked with the Healthcare Research and Education Trust, HRET, to develop a set of recommended changes for healthcare organizations. So if you are active in quality improvement work and you are a health care organization, please check this out.

Finally, there are many resources on the SIDM website that you might find useful and you can follow how diagnosis is improving by joining our list serve, our LinkedIn group, or by reading the articles and essays in our journal diagnosis. Please visit the SIDM website. We are at www.improved diagnosis.Org.

Let us quickly summarize. Diagnostic errors are common. They are especially harmful and in our opinion, many or most of these are preventable. Here's the final bit of advice from the National Academy report: "Improving the diagnostic process is not only possible, but also represents a moral and professional and a public health imperative. Thank you for helping us address diagnostic error." Thank you for participating in this activity. Please proceed to answer the post-activity assessment questions to receive credit. Please take a moment to complete the program evaluation.

This transcript has been edited for style and clarity.