See, Hear, Feel

EP92: Dr. Jason Lee on diagnostic error, cognitive bias, and how we improve

December 13, 2023 Professor Christine J Ko, MD / Dr. Jason B. Lee Season 1 Episode 92
See, Hear, Feel
EP92: Dr. Jason Lee on diagnostic error, cognitive bias, and how we improve
Show Notes Transcript

Dr. Jason Lee has written and presented on how we think, cognitive bias in medical diagnosis, and errors in the pathologic pathway of a biopsy from clinic to final report. The latter was recently published in the Journal of Cutaneous Pathology.  It is truly a pleasure to hear Dr. Lee talk with ease about errors, including his own. Dr. Jason B. Lee, MD is Professor, Director of the Jefferson Dermatopathology Center, Director of the Dermatopathology Fellowship, and Director of the Jefferson Pigmented Lesion Clinic at Jefferson Medical College of Thomas Jefferson University. He interned at the University of California, Irvine and then did his residency and fellowship at Thomas Jefferson University Hospitals. He has a recent article in the Journal of Cutaneous Pathology that focuses on error in the pathologic diagnostic pathway. 

[00:00:00] Christine Ko: Welcome back to SEE HEAR FEEL. I'm really excited to be with Dr. Jason Lee today. Dr. Jason B. Lee is a Professor and Director of the Jefferson Dermatopathology Center, Director of the Dermatopathology Fellowship, and Director of the Jefferson Pigmented Lesion Clinic at Jefferson Medical College of Thomas Jefferson University. He interned at the University of California, Irvine, and then did his residency and fellowship at Thomas Jefferson University Hospitals. He has a recent article in the Journal of Cutaneous Pathology that focuses on error across the pathologic diagnostic pathway, and I will put a link to that in the show notes. I also met Dr. Lee long ago when I was working in Philly. Welcome to Jason. 

[00:00:44] Jason Lee: Hey.

[00:00:45] Christine Ko: Would you share a personal anecdote? 

[00:00:47] Jason Lee: I can tell you about an error, one highlighted in the paper. In hindsight, it was an obvious diagnosis. In fact, it was a slide that I saved for teaching purposes. Usually diagnostic errors are discovered by the clinician on the clinical side, but this one I discovered myself. It was a slide I saved for teaching purposes, I looked at it, and I said, this is a classic, dermatophytosis or tinea capitis. Then I look up the diagnosis on my EMR, and it says lichen planopilaris, and I said, wow, that's totally different. And so what happened was that the clinician who sends me the specimen was looking for lichen planus. I looked at the requisition form, saw lichen planopilaris, saw a perifollicular fibrosis, and then I wrote lichen planopilaris as a diagnosis, totally missing tinea capitis. If you look back and trace some of the thinking processes. There was diagnostic momentum. I saw his diagnosis, lichen planopilaris. And then, the phenomenon of inattentional blindness, where it's really something in front of you, but you just don't see it. Those were things that could explain my error. 

[00:01:53] Cognitive processes can be blamed for some of our shortcomings in coming to the correct decisions and diagnoses. I don't think doctors have huge personal insights on diagnostic errors. Who wants to look at your mistakes and dwell on it? But I think it's probably important for our patients to get accurate diagnosis. 

[00:02:12] Christine Ko: Yes, absolutely. That is one of the reasons why I love your paper because I think it's really not typical of current medical culture to publish a paper like this, where you have a whole table where you've documented all the errors across the 20 plus handoffs in the pathologic pathway, when a biopsy comes from clinic to finally getting a report out, but also you documented essentially your own errors in a very open way, which I think is so needed in medicine and really a role model for how I could be. How did you get into this topic on error? 

[00:02:54] Jason Lee: I've been dabbling in the topic of diagnostic errors and errors in our clinic for a while. I was a residency director for over 15 years. Every year we would teach the same thing, and the residents would make the same mistake and same errors. And so I question, how can we skip some of these steps and get the residents get to a more accurate diagnosis faster? I ran into a book called Thinking, Fast and Slow by Daniel Kahneman. I didn't know who he was at the time, I didn't know how big of a figure he was. I read the book, and it was fascinating. It outlined some of the cognitive processes in decision making, some of the shortcomings. I started to lecture a little bit at the American Academy of Dermatology since 2014, almost every year, just talking about our cognitive process and their shortcomings. 

[00:03:41] Now we're in the era of patient safety and quality improvement. So every fellowship and residencies have patient safety and quality improvement projects. It's been published that there are more errors in this diagnostic pathway than any other pathway in dermatology practice. And so we just started logging all errors that occur, including errors that occur at the clinic and including errors that occur within our lab and also including diagnostic errors, the errors that I make under the microscope. We reported that in the Journal of Cutaneous Pathology

[00:04:13] Christine Ko: Do you think that there's more error in the pathology diagnostic process than compared to derm? 

[00:04:22] Jason Lee: Dr. Elston and Dr. Swerlick had published on this topic. They surveyed and reviewed the literature on it. When there's a decision to do a biopsy or perform a procedure in the clinic, there's over 20 handoffs that goes from the decision to do the biopsy and when the report actually gets to the patient. Every step has a risk of having errors. Our study confirms that one of the biggest errors that occur is at the clinical level, particularly incorrect site that's been miscommunicated to the pathologist, particularly if it's laterally right and left. And it highlights specific areas where we can do more quality improvement in dermatology.

[00:05:02] Christine Ko: Yeah. Really, any time there's passing off of information, there's always potential for error and miscommunication. 

[00:05:13] Jason Lee: Yeah. The rate of diagnostic or medical errors or diagnostic errors in the cognitive field is supposed to be about 10 to 15%. Whereas in perceptual field, like us, dermatology pathology, is supposed to be less than 5%. And most of it is blamed on faulty cognitive processes. But I think in perceptual field, knowledge deficit also contributes significantly. Because, if you don't know the entity and never heard of it, you will not be thinking of it. The things I do is I always try to learn more and learn as much as I can. 

[00:05:46] Christine Ko: Yeah. The statistics that are out there in the literature are that in perceptual fields like dermatology or dermatopathology and pathology, it's less than 5% as you said, but maybe one to 2%. In your paper, I think it was a little bit less than 1%, right? 

[00:06:03] Jason Lee: Yeah. 

[00:06:03] Christine Ko: It was like 0.9 something percent of actual diagnostic error. Still, it's a little scary for me when I think about that because say I'm average, in dermatology clinic and in dermatopathology practice of reading slides, if I have a 1 percent error rate, that means 1 out of 100 cases or patients, I'm making an error. You can probably tell from my voice, it makes me really uncomfortable. I don't want to be making an error one out of 100 times. 

[00:06:36] Jason Lee: I know we're not perfect. To err is human, as Benjamin Franklin said. Error rates are probably underestimated. We don't know all the errors that we make.

[00:06:45] If you read some of these papers, most errors do not end up in harm and so probably not discovered. It's quietly absorbed by patients, the payers, and everyone else. Error rates are probably much higher, it's just not something we study.

[00:06:59] Christine Ko: Yeah. You seem very comfortable with this, more comfortable than I am. I'm trying to talk about these things because I think by talking about it and staying curious and interrogating my own discomfort, it will help me face error, and be better about it, and be a better teacher, and a better example for my own colleagues around me and trainees.

[00:07:21] Because I know I am definitely not perfect. When I say I'm uncomfortable with a 1 percent error rate, I don't mean that because I think I'm above error, or I even want to be perfect. I actually don't. I don't want people to expect that I'm perfect. The 1 percent error rate just makes me uncomfortable because we don't talk about it. And, I agree with you. It's probably even higher than that,. We act like it doesn't exist. And I think that's the thing that bothers me the most and why I really appreciate your paper. 

[00:07:52] Jason Lee: Okay. I'm glad to hear that. Again, I think errors are unavoidable. It happens. It's part of human nature. I think you're right. We need to face it, and we need to figure out why it's made, and try to minimize it. Going back to diagnostic errors, that's mostly due to cognitive processes and faulty processes, and part of it is our heuristics, the shortcuts that we use to come to a diagnosis, are subject to cognitive biases. Being aware will help, but as these are subconscious or unconscious, it's hard to control and hard to be aware of it as you're doing it. Just like my mistake about dermatophytosis, it's an obvious diagnosis, but I was subject to that unconscious bias that was occurring. We just have to be aware of it. I think that'll help. And when the stakes are high, like melanoma or not, Daniel Kahneman says, it's unavoidable to make errors. And all you can do is just try your best. 

[00:08:45] Christine Ko: Yes. I love that book as well. Daniel Kahneman's book, Thinking, Fast and Slow. It's hugely influenced me as well. And really his idea of this fast processing, fast thinking versus slower processing, slow thinking. It changed the way I look at my slides. Because what I'll do is I'll go through once, "quickly". It's not really a time thing, but I'll just be trying to use System 1. What do I really just think right away? Sometimes I will pause and take a little longer, even in that first pass with slides, or I'll even set certain ones aside. But my idea is just to use that System 1. Then I always go through all of them again, at least a second time before I sign them out and just ask myself then really in a more analytical way, what else can this be? Am I wrong? What makes me think it is that, and what are the things that I am maybe not paying attention to. And I also will ask like, is there anything else? Because usually there's only one diagnosis in a slide, but every once in a while there's more than one. And I think there is a tendency to miss it. That's another bias called search satisficing. Where you do end a search cause you think you're done. 

[00:09:54] Jason Lee: I'm glad you're into the subject because I think, as I said, most physicians, they don't have personal insight, how we make decisions. You're putting this in the foreground, and I'm glad that you're talking about how System 1 and 2 works, and how the interplay with the two, you're using both, leveraging both to increase your accuracy. I wish there would be more studies to do this, but it's hard to measure what we're doing, our cognitive process. And that's probably a reason why we don't have that many studies out there, but people like you will hopefully encourage some of the audience to think about the topic and maybe do some studies. 

[00:10:28] Christine Ko: Yeah, it is really hard to measure it. Like your lichen planopilaris /tinea capitis story. I do catch my own sort of errors, I'll call them, when I do these two passes. Even like really simple things for example, we know in dermatopathology, if something looks like a seborrheic keratosis, there might be a dermatofibroma underneath that, and sometimes when I am going that "quick" way, sometimes I do miss that. It's just faulty processing, I forget to think about it. And then on my second pass, Oh, no, this is a dermatofibroma. It's not a seborrheic keratosis. So I catch my own error. But it's also made me think about, is that really an error when the diagnosis goes out correctly? I still count it mentally as an error because I didn't notice the first time. It's a way for me to become more comfortable with, my mind is making these mistakes. 

[00:11:21] You do come across as having a comfort level with this, and I wanted to ask you if you think there's a component of shame when we make errors in medicine.

[00:11:32] Jason Lee: So we don't wanna flaunt our knowledge deficit and also that we make mistakes. Maybe I'm more inclined to do this stuff 'cause we had Philly derm society meetings every year and we present cases. I still run it for our department, and every time we present something there's an error or mistake in diagnosis, I would represent it the next year and try to let them know that this wasn't correct prior. An example is a case which I thought was keratoacanthoma centrifugum marginatum, but patient was African American. In the end, it turns out to be verrucous sarcoidosis, which can look like squamous cell carcinoma. So that diagnosis made more sense. It's a culture that I developed within myself over time. And also, with teaching the residents, I learned with them, I admitted my mistakes with them when I was a residency director. 

[00:12:19] Christine Ko: I love that. I love the way you put that because it goes to deliberate practice. That's true for myself too. Anything that I do more than once and then over and over again, I get better at it, and I get more comfortable with it. The way you described it made it sound like you started talking about error in Philly Derm Society and around the scope. And, the more that we do it, the more we all get comfortable with it. Is that kind of true? 

[00:12:43] Jason Lee: Yeah, I think so. I know I make mistakes, so let's try to figure out, you know, ways to improve it. 

[00:12:48] Christine Ko: That segues into my next question. How do you continuously improve?

[00:12:52] Jason Lee: So there was one moment where I thought knowledge deficit had more impact on my diagnostic error. Again, we only see what we know. I had a case of levamisole induced vasculopathy I did not recognize in 2004. I knew it was a vasculopathy, but I didn't know what it was from. The distribution was perfect for it, but it was like, two years prior things were published about it and the distribution, and I didn't know about it. I realized I need to make sure that I stay on top of things. That's one. So I try to keep up with the knowledge for sure. The second thing is I do try to get feedback. You need deliberate feedback. And so when a patient I treat, I make the diagnosis, and I'm not 100 percent sure, and I give him some medications, and if the patient never comes back, I don't know if the patient got better and never came back. Patient went to another doctor, wasn't happy with me, or just went away. So I try to get feedback for everything that I do, and that's important, and I think that helps you to improve your, your diagnosis and accuracy. So those are the two main things that I do. 

[00:13:55] Christine Ko: I love it. Do you have any final thoughts? 

[00:13:57] Jason Lee: The diagnostic errors, most were discovered by the clinicians giving me feedback and saying, Hey, this doesn't make sense. So I will say that the pathology report is not the answer key. It's an opinion on a specimen by a pathologist with varying experience. So it's important to make sure that the diagnosis makes sense. Give that feedback to the dermatopathologist. Work closely and communicate with your dermatopathologist. Help our patients by keeping in close contact with their dermatopathologist and making sure things make sense rather than taking the what's on the report as the answer to what you biopsied. 

[00:14:35] Christine Ko: Thank you so much for spending time with me. It was really fun. 

[00:14:38] Jason Lee: Thank you very much for having me. I enjoyed it.