EXPERT’S CORNER: Evolution of uncertainty in surgical pathology. Part 2: Management

Paul C. Stromberg DVM, PhD, DACVP
Professor Emeritus, Ohio State University

The management of uncertainty begins with your acceptance that errors in judgment occur and that you will eventually make them. Dr. Groopman has told us we all make these errors. I hate it when I make a mistake and that is what keeps me centered on avoiding them if possible. There is voluminous literature about this in human medicine but very little in the veterinary literature. None of us wants to confront our mistakes or suffer incorrect diagnoses. If you are in denial about this, you are still in “The Danger Zone.” Management of uncertainty begins with the desire to manage it.

“Uncertainty Should Stimulate Description”

We teach our students to consciously step by step deconstruct the image pattern before them into its component parts. Find the elements that, when assembled, make the lesion pattern associated with a diagnostic entity. What are the elements that define a mast cell tumor, cirrhosis, blastomycosis, cyclical flank alopecia? Look for those features and when identified, make your diagnosis. However, with experience, we abandon this step by step process and immediately, subconsciously “see” or “recognize” the pattern of a disease entity in an intuitive way cognitive psychologists call a “Pattern Recognition” or “Gestalt” Diagnosis.” This is a real skill that marks the expert pathologist many of you will eventually acquire. The analogy is what you perceive when you meet an old friend or former student. You subconsciously observe their face without enumerating all the various parameters that make it up. Unlike the facial recognition software on your smart phone which rapidly compares the various elements identified as “You” to those stored in your phone, you see your friend and instantly perform a pattern recognition identification. Your phone does what we teach our students to do with pathologic processes but you don’t.

I gradually began to experience this about 5 years after I started to study pathology. It’s incredibly helpful but it’s also inherently dangerous. You should trust this skill (and eventually you will learn to) but always take the time to verify your first impression by looking for the unique combination of elements in the pattern that define the particular pathologic entity before you. Ironically a pattern recognition diagnosis without verification is an error made by experienced pathologists not students because students lack the experience and awareness of pattern recognition and so rely on deconstructing the image as they are taught. “Cogent pathologic evaluation combines the first impression in pattern recognition with deliberate analysis.” Research in human medicine indicates experts form an opinion in about 20 seconds. I often look at a plasmacytoma and make a pattern recognition diagnosis in about 2 seconds! But then I pause and force myself to verify it by searching for and finding the critical elements which define it. Resist the temptation to make the diagnosis and go on to the next case without at least some verification. Powerful forces may impel you to take the shortcut. What else is going on in your life that could distract you? Are you in a hurry to finish your cases? Are you tired, sick? Yes, this slows you down and we all understand that time is the most precious commodity in medicine. None of us have unlimited time in our diagnostic tasks but if you try to rush things, you risk making an error. How many times have you approached someone you thought was an old friend and quickly discovered it was a stranger? Pure pattern recognition identification is sometimes wrong. Remember accurate diagnosis is the most important goal in surgical pathology. Rapid report generation is secondary.

If I look at an image and cannot “see” a familiar pattern or have trouble finding the essential elements right away, I set the case aside and come back to it later. Sometimes a few hours is sufficient. Often for me it’s over night and occasionally a couple of days. Sometimes I begin to write out a description of what I see similar to what we do on the certification examination. That helps me clear my mind, start over, find the critical elements of the pattern and unlock the diagnosis in my mind. Adopt the aphorism that “Uncertainty stimulates description.” This forces you to slow down and go back to first principles of deconstructing the pattern before you make a diagnosis. Frequently when I do this, I see the pattern elements quickly and wonder why I had difficulty with it before. It also decreases my uncertainty and gives me confidence that I made the correct diagnosis. Secondly it reinforces confidence in my judgment. Usually I do the pattern deconstruction mentally. Personally I try to avoid lengthy descriptions in biopsy reports because it’s wasteful of time and effort which can generate fatigue and lead to errors in judgment. Only in cases of excess uncertainty do I engage in writing out my observations. When I do that it is less for the clinician’s benefit and more as a mechanism to slow my thought processes and ensure I see all the diagnostic elements of the pattern. I used to tell clinicians that “The certainty of my diagnosis is inversely proportional to how much I write.” When I generate a biopsy report, I remember that I intend to communicate with five people; 1) The primary care clinician who sent me the biopsy, 2) The specialist to whom the clinician may refer the case, 3) The pet owner who has a right to see my report; 4) An attorney who may be involved in the case later and finally 5) Myself. If I have to relook at the case at a later date it reminds me what I saw and what I was thinking. The guiding principle is give each of them what they need to manage the case.

“Be Aware of the Cognitive Traps That Can Impair Your Judgment”

Confirmation Bias is a well-known principle of cognitive psychology. It is the tendency to search for or interpret new information in a way that reinforces our first impression (“Pattern recognition” diagnosis) and avoids or ignores information that contradicts or would lead us away from our prior belief. The common aphorism for this is “We see in the data (pattern) what we want to see.” It’s a constant threat to accurate diagnosis in surgical pathology. This is why it’s important to remain uncommitted in our diagnosis until we have seen all the critical elements and what makes “Gestalt” diagnosis without deliberate analysis dangerous. Put succinctly, “Describe first…then interpret.” This is why I set aside difficult cases for a short time. It clears my mind so I can return to it later and reduces the risk of committing confirmation bias. I teach our students preparing for the histopathology cases on the certification examination to briefly look at the slide to get oriented first but do not make a pattern recognition diagnosis. Then start by noting the various processes until you think you have seen everything important. THEN make your morphologic diagnosis and interpret a cause or name of the disease entity. A common mistake on this part of the examination was seen among the candidates who began their essay by stating the diagnosis followed by a description of the critical elements of their diagnosis. Sometimes their initial conclusion was wrong but their description would be an enumeration of the elements of that misdiagnosed entity when in fact those elements were not present on the specimen before them. It seems such an error is not possible but I saw it many times. They made a diagnosis and so were predisposed to see its essential elements even though they weren’t present. “We see in the data what we want to see.” A classic manifestation of confirmation bias. You can do the same thing on biopsy cases if you are not careful. “Describe first… then interpret!” Again for most of your cases you do the image deconstruction in your head and only write out a description for the confusing cases.

Anchoring occurs when we do not consider multiple diagnoses but quickly and firmly latch on to our first impression and ignore discrepancies that would argue to reject it. This often follows confirmation bias (we only see what we want to see) and so become “anchored” to our diagnosis. Once anchored, it is very difficult to change your mind. It is vitally important to remain uncommitted in your opinion until you are sure you have seen all the elements in the pattern.

A third cognitive trap is Search Satisfaction or the tendency to stop searching when we find something important. Often there is more than one important pathologic process in biopsy samples which we may overlook if we stop after we find something important and make a diagnosis. Histopathology essays on the certification exam are a test of a candidate’s ability to find all the elements in an image pattern and assemble them into a diagnosis (es). Occasionally the certification examination contains a few “Two-fors” (a slide or gross photo with more than one entity present in the image pattern) because there is often more than one pathologic process in surgical biopsies. Analysis of the histopathology essays revealed a high correlation between successful candidates and the ability to see all the changes in the cases. So thoroughly evaluate the entire biopsy and don’t stop looking when you find something important. “Two-fors” are placed on certification examinations because they occur in biopsies and they are very discriminating and predictive of success.

“Total Patient Evaluation is the Way to Go”

Total Patient Evaluation is a foundational principle in medicine. Clinicians do not make a final judgment and formulate a treatment plan until they have all the facts possible. It should be the same for pathologists but we are 100% dependent on clinicians to provide this information and they are failing us too frequently. Generally this is not a problem in academic veterinary medical centers but mostly in those private practices that submit biopsies to commercial diagnostic laboratories. Often I am not told what species the patient is, much less a good history, description, location, distribution of the lesions and what the clinician thinks etc. Such information orients us by framing the clinical problem and adds objective data to the case that may unlock a thought pathway leading to the correct diagnosis. It can also make you aware of a diagnosis that you may not have considered. Clinicians must frame the case for us. “Don’t tell the pathologist anything, you will bias them” is wrong headed. Research in human medicine shows that without proper framing the potential for an error is considerable. It is the pathologist’s responsibility to control bias. Clinicians should share with them what they know. I often tell clinicians to “Help the pathologist to help you!”

Promote the “Diagnostic Biopsy” concept to clinicians. What is a diagnostic biopsy? It’s a sample submitted to the pathology lab that has the highest probability of generating an accurate, unambiguous diagnosis in a timely manner that is useful to the clinician. The essential elements of such a sample are 1) an adequate amount of tissue 2) representative of the pathologic process 3) free of artifacts that obscure the pattern of the lesions 4) accompanied by a signalment, history, description of the lesions 5) and what the clinician thinks (DDx) and wants (Rule Outs). In other words give the pathologists what they need to make a total patient evaluation. In 2 to 4 percent of cases I see, I am not told what species the patient is. “I can’t diagnose FIP unless I know the patient is a cat.”

There will be cases when your uncertainty is fueled by external conditions that can’t be mitigated by more time such as insufficient sample, crush artifact, poor fixation, freezing, all of which can fuel uncertainty. This is beyond our control but I explain this to clinicians so they know why I cannot be certain but only “favor” a diagnosis. It also gives them feedback that what they do during the biopsy can impact our judgment. I frequently tell practitioner groups in evening CE seminars, “What you do during the biopsy matters. What you do NOT do after the biopsy matters even more!”

“Make the Surgical Biopsy a Multiple Choice Quiz”

It also helps to know what your differential diagnosis is for the group of patterns you are observing. We assembled these lists when preparing for the certification examination. Keep these in your head or in notes somewhere. Life (and pathology) is not a multiple choice test but it’s helpful if you know the range of choices to consider when looking at lesion patterns. The mass is a round cell tumor or a spindle cell tumor, superficial perivascular dermatitis, or interstitial pneumonia. As new literature defines new entities, add these to your lists. Proper framing by clinicians may add new possibilities to your list that you may have left out or forgotten. If your uncertainty is such the pattern does not seem to fit into your DDx list, it’s time to stop and reconsider. Slow the perception and analysis process. “Time opens the mind!”

“Judgment! How do I Teach Judgment?”

“Primum non nocera (First or above all do not harm)!” You have to know how you can cause harm to avoid it. For us mostly it’s an incorrect diagnosis. It starts by knowing the ramifications of your diagnosis, whether it’s correct or incorrect. You should understand this for most of the diagnostic entities you are considering. What if you diagnose lymphoma when it’s really lymphoid hyperplasia? What if you diagnose lymphoid hyperplasia when it’s really lymphoma? What is the impact of your mistaken diagnosis? Often it’s a radically different prognosis that can result in emotional distress, financial loss and delayed treatment. When the difference has potential to cause “significant harm”, I set more stringent criteria for my interpretation. That is good judgment. As your experience grows you will learn to “juggle” contradictory bits of data in your mind while seeking other information to bolster your confidence and help make a decision. (“Judgment difficult”). This is where the science of pathology gives way to the art. “Little competes with honesty in the biopsy report.” Tell the clinician when you have significant uncertainty and why so that he/she can manage THEIR uncertainty in the best interest of the patient or owner. There is no escape from uncertainty. Judgments in pathology have to be made every day in the face of uncertainty and we need to adjust to it. Don’t be paralyzed by your uncertainty. Accept it and learn to manage it. Dr. Groopman tells us that

doctors learn best by recognizing their mistakes and keeping them close at all times. The problem for most surgical pathologists is we often make a diagnosis and never learn if we were right or wrong. Take every opportunity you can to get feedback on your cases and remember your errors. (“Opportunity is fleeting.”). I never look at spindle cell proliferations without thinking twice about that nasal granulation tissue case I had many years ago.

So, on the back side of our exploratory journey through pathology we indeed arrive where we started and know, perhaps for the first time, the nature of uncertainty. It doesn’t go away. It can be managed by application of our skills blended with an awareness of potential errors and the methods to minimize them. At this juncture surgical pathology becomes as much an “Art” as “Science.”