Learning Objectives

At the conclusion of this educational activity, participants should be able to:

• Discuss challenges associated with the dynamic nature of the diagnostic process, and the importance of evolving problem representation in that process.
• Identify ways that bias associated with medical conditions, such as obesity and its treatment, can lead to poorer outcomes.
• Describe how effective teamwork improves the diagnostic process and identify key barriers and enabling factors for effective teamwork.

The Case

About five weeks after gastric bypass surgery, a woman began experiencing nausea and vomiting when she attempted to eat solid food, but she could keep down liquids. A physician performed an outpatient dilation procedure for a suspected postoperative stricture. Two days after this procedure, the patient was treated for dehydration in an outpatient clinic. The next day, the patient went to the emergency department (ED) with continued nausea and vomiting, and she reported losing 100 pounds since her bypass surgery (i.e., several times the expected weight loss over this period). A computed tomography (CT) scan was performed in the ED, and she was admitted to the intensive care unit (ICU) for pancreatitis and dehydration. Her neurological and mental status examinations were documented as normal.

Within a few days after admission to the ICU, nursing documentation indicated that the patient had difficulty walking even with assistance, and she complained of a tingling sensation in her fingers and tightness in her shoulder. Additionally, the patient’s vomiting had not subsided, she developed fecal incontinence, and she appeared alert but would not respond to questions, leading the physician to question if she was depressed. During the next week, the patient was unable to tolerate either food or liquids by mouth. Although a nutritional assessment suggested total parenteral nutrition (TPN), TPN was not started and the patient continued to suffer from dizziness, vomiting, and an unsteady gait, suggesting pelvic muscle weakness and requiring use of a “tether” to keep her from falling. Nurses continued to report that she was intermittently uncommunicative with a “fixed” gaze.

On hospital day 12, the patient was discharged with diagnoses of “intractable nausea and vomiting,” “obesity,” and “obstructive sleep apnea,” with orders for TPN administration at home. The TPN order included glucose and standard nutrients but did not include any supplemental vitamins or lipids. Three days after discharge, the patient was readmitted for worsening confusion and profound motor weakness, which progressed to respiratory failure requiring mechanical ventilation. Laboratory tests revealed an extremely low thiamine level, and the patient was diagnosed with advanced Wernicke-Korsakoff Syndrome. Due to the delayed diagnosis, the patient suffered permanent brain injury requiring around-the-clock care.

The Commentary

By Andrew P.J. Olson, MD, FACP, FAAP

Background

One of the hallmarks of many patients’ diagnostic odysseys in modern medicine is complexity.¹ Diagnosis is not only a complicated activity – that is, one with significant amounts of data, multiple steps, and interrelated parts –instead, it is a complex activity.^2,3 Complex activities include not only significant amounts of data and multiple steps but also feature non-linearity, uncertainty, and even unpredictability. Complicated processes can be reliably carried out by carefully following the steps laid out for the process – like assembling a piece of machinery, building a Lego set, or solving a calculus problem. Complex problems, in contrast, cannot be solved by following a series of predetermined or knowable steps. Diagnosis is among the most complex cognitive activities to which humans are tasked, and it is important that we begin to characterize and discuss complexity in diagnosis to prevent diagnostic errors and forward the pursuit of diagnostic excellence.⁴ In the diagnostic process, new information is continually added and must be considered as part of the decision-making progress. If this new information is not incorporated into teams’ medical decision-making, there is a substantial risk for diagnostic error.

This complexity mindset contrasts with how we usually conceive of diagnostic and management reasoning, especially during medical education: we often consider a patient’s presentation as a problem to be solved at one point in time, rather than a recurring and evolving problem that is influenced by ongoing changes in a patient’s health. Diagnostic reasoning cannot be reduced to an equation to be solved, especially when considering new problems that arise as a result of treatment. We often consider side effects of treatments, but it can be very challenging when completely new problems arise after appropriate treatment for a patient’s health condition.

In the case presented, the patient underwent a procedure (gastric bypass surgery), aimed at treating a chronic health condition (obesity and its consequences). The procedure appears to have had its intended effect: the patient lost a significant amount of weight, although the rate of weight loss was more than intended or recommended. The differential diagnosis at this initial presentation of nausea and vomiting appeared to include mechanical/anatomic complications of the patient’s gastric bypass surgery, and an endoscopy with dilation was performed. At this point, however, the case begins to move from a complicated case to a complex one; the endoscopy with dilation did not improve the patient’s symptoms and she was admitted to the hospital for ongoing symptoms (nausea and vomiting) attributed to an apparently new problem: pancreatitis. She was managed supportively – as is the evidence-based approach – for acute idiopathic pancreatitis, including appropriate fluid resuscitation and pain control.^5,6 Her pancreatitis improved, but her overall condition did not.

She continued to be unable to eat and drink, but now also developed difficulty with ambulation, altered mental status, and other neurological symptoms. These symptoms persisted and she was again managed supportively – by being placed on TPN – and she was discharged. Upon readmission, her neurological status, including profound weakness and encephalopathy, had continued to worsen, necessitating intubation and mechanical ventilation. Alternative causes were explored and the diagnosis of Wernicke-Korsakoff syndrome – a feared condition resulting from thiamine deficiency – was made.⁷ However, this diagnosis was made after the point at which thiamine repletion would have substantively improved her clinical condition.

Quality Improvement Approach: Managing Complexity in the Diagnostic Process

How might we begin to effectively navigate complexity in the diagnostic process? We must make decisions based on shifting, incomplete, and imperfect information to move care forward. While the science in this space continues to develop, we will highlight approaches that may help improve diagnosis in the face of complexity:

• Practice iterative and dynamic problem representation
• Ensure that all team members have a role and voice in the diagnostic process

Practice iterative and dynamic problem representation

Problem representation is a fundamental and widely taught part of the diagnostic process in which salient features of a patient’s case are distilled down and translated into specific terminology to succinctly describe the problem to be solved.^8,9,10 Problem representation is a key activity in most diagnostic reasoning curricula in health professions education and is considered by many to be a hallmark of diagnostic expertise. From a psychological standpoint, problem representation can be conceived in two ways.¹¹ First, and most educationally relevant, problem representation can be considered a formal approach to using specific language to frame a patient’s problem in a standard way.¹² This process includes using terms – called semantic qualifiers – to allow clinicians to begin matching a patient’s presentation to one or more of their illness scripts, thus enabling the diagnostic process to move forward. The second way to conceive of problem representation is as an individual’s mental representation of a patient’s health problem; these mental models are specific to the individual diagnostician and shaped by their previous knowledge and experience. Although both approaches to thinking about problem representation have utility, we will focus primarily on the first type for this discussion.

In this case, the patient developed new problems superimposed on old ones throughout the case, adding substantial complexity and uncertainty. In addition, while some conditions persisted throughout the case (e.g., nausea, vomiting, recent gastric bypass, and neurological symptoms), others came in and out of the diagnostic and management frame (e.g., pancreatitis). This dynamic shifting of what is in frame and out of frame means that the problem to be solved is continually changing. This frameshifting can be managed by continually updating and refining the problem representation used in each case.

For example, a potential problem representation in this case upon the patient’s first presentation could be:

• A middle-aged woman with a history of obesity 5 weeks status post gastric bypass surgery presents with nausea, vomiting, and inability to tolerate oral intake. 

Based on this problem representation, a differential diagnosis was focused on postoperative, mostly anatomic, complications that can arise from gastric bypass surgery; an upper gastrointestinal endoscopy was performed based on this framing.

However, upon presentation to the ED after her endoscopy and dilation, the problem representation could be slightly different:

• A middle-aged woman with a history of obesity 5 weeks status post gastric bypass surgery and 3 days status post upper endoscopy and dilation presents with nausea, vomiting, and ongoing inability to tolerate oral intake with significant, rapid weight loss and recurrent dehydration.

While much of this updated problem representation is the same, new information has been added to the frame, including profound weight loss and lack of improvement after endoscopy and dilation. This subtle shift begins to change the differential diagnosis from focusing on anatomic, postoperative complications alone to include other conditions that could contribute to her symptoms. After many days in the ICU, the problem representation could again evolve:

• A middle-aged woman with a history of obesity 6 weeks status post gastric bypass surgery and one week status post upper endoscopy and dilation admitted with nausea, vomiting, and inability to tolerate oral intake with rapid weight loss and dehydration, now with possible pancreatitis, encephalopathy, limited eye movement, and ataxia.

The previous differential diagnosis focused on postoperative, anatomic complications as well as conditions leading to nausea, vomiting, and poor oral intake – such as pancreatitis – is no longer adequate for this new, updated, problem representation. In fact, there are components of this new problem representation that necessitate reconsideration of the entire diagnostic approach. In this situation, clinicians should recognize their uncertainty and invite new information and ideas.

Upon readmission to the hospital, with the patient even more severely ill, the problem representation again shifts:

• A middle-aged woman with a history of obesity s/p recent gastric bypass surgery and upper endoscopy and dilation, profound recent weight loss, with a recent hospitalization for nausea, vomiting, and inability to tolerate oral intake discharged on TPN now admitted with profound encephalopathy and motor weakness, necessitating mechanical ventilation.

As neurological symptoms become more prominent in the problem representation and other aspects shift away (postoperative anatomic complications and pancreatitis, for example), the diagnosis now comes into focus. It is with this lens, then, that the team appropriately recognizes the syndrome and sends the confirmatory test: a thiamine level. While this progression, in hindsight, seems relatively straightforward and linear, it was likely very complex when experienced in real time. Thiamine deficiency is a known and expected complication of gastric bypass surgery.¹³ As various clinical factors shift in and out of frame, the case evolved substantially and the final diagnosis, while a potential consideration, was likely not predictable from the initial presentation. This case demonstrates the fundamental nature of problem representation in undergirding the diagnostic process.

Ensure that all team members have a role and voice in the diagnostic process

There has been increasing recognition in recent years, reflected in the recommendations of the National Academy of Medicine’s Committee on Diagnostic Error in Health Care, that diagnosis is a team sport that requires effective collaboration.¹⁴ Even relatively straightforward diagnoses in outpatient settings are never made by a single clinician alone; instead, it is always a partnership involving patients, their families, clinicians, and other members of the health care team that leads to correct and timely diagnoses.¹⁵ One of the most widely recommended yet incompletely implemented strategies to improve diagnosis is to engage all members of the health care team as contributors to the diagnostic process. As this case progressed, various members of the health care team noticed clinical phenomena that, if fully considered, would have shifted the problem representation and differential diagnosis toward a fulsome and accurate explanation for the patient's health problem.

For example, members of the nursing staff noted limited eye movements during the patient’s ICU admission. Ophthalmoplegia, manifested by a fixed gaze with paralysis of extraocular muscles, has a relatively narrow differential diagnosis. Recognition of this sign could have helped steer the diagnosis toward neurological problems arising from nutritional deficiencies. It is very likely that physical and occupational therapists were involved in managing the patient's mobility prior to discharge, but it is not clear whether their description of ataxia made its way into the clinician's problem representation and thus the differential diagnosis. The dietitian could also have contributed to the diagnostic process, by virtue of their involvement in assessing the patient's nutritional status and TPN needs. A query from the medical team to the dietitian might have prompted earlier recognition of the patient’s severe malnutrition and resulting triad of Wernicke’s encephalopathy: ophthalmoplegia, ataxia, and altered mental status.

While teamwork is often recommended as an intervention to improve diagnosis, there are cultural and technical barriers that inhibit effective teamwork in the modern clinical care environment. These barriers include professional hierarchy, time pressure, challenging workloads, and inadequate structures for team collaboration. Implicit clinician bias against obese patients must also be considered,¹⁶ because this bias may interfere with full consideration of seemingly unlikely diagnoses in this population, such as severe malnutrition.

Fortunately, the modern environment also has enabling factors for teamwork, including an electronic health record to which all team members – including patients and their families – can contribute and view one another's thoughts and decision making, improved instant messaging technology, and prominent acknowledgment of the role of all members of the healthcare team in the diagnostic process.^15,17 In this case, we should consider how key information arising from nurses, physical and occupational therapists, dietitians, and other members of the health care team is considered in the differential diagnosis and how all these professionals should more actively participate as members of the diagnostic team, bringing their diverse perspectives, improving problem representation, and potentially countering implicit bias.

Take Home Points

• Diagnosis is a complex process that requires diligent consideration of clinical factors that are shifting in and out of frame throughout a patient’s journey.
• Problem representation is a key part of the diagnostic process that frames the clinical situation and helps inform differential diagnosis as well as management.
• Accurate problem representation requires input from all members of the diagnostic team, and it is imperative that health care systems identify methods to actively engage all health care team members – including patients – as contributors to the diagnostic process.

Andrew P.J. Olson, MD, FACP, FAAP
Associate Professor of Medicine and Pediatrics
Director, Division of Hospital Medicine
Director of Medical Education Research and Innovation, Medical Education Outcomes Center
University of Minnesota Medical School
apjolson@umn.edu

References 

1. Bordini BJ, Walsh RD, Basel D, et al. Attaining diagnostic excellence: how the structure and function of a rare disease service contribute to ending the diagnostic odyssey. Med Clin North Am. 2024;108(1):1-14. [Available at]
2. Choi JJ, Durning SJ. Context matters: toward a multilevel perspective on context in clinical reasoning and error. Diagnosis (Berl). 2023;10(2):89-95. [Available at]
3. Wilson E, Daniel M, Rao A, et al. A scoping review of distributed cognition in acute care clinical decision-making. Diagnosis (Berl). 2023;10(2):68-88. [Available at]
4. Yang D, Fineberg HV, Cosby K. Diagnostic excellence. JAMA. 2021;326(19):1905-1906. [Available at]
5. Mederos MA, Reber HA, Girgis MD. Acute pancreatitis: a review. JAMA. 2021;325(4):382-390. [Available at]
6. de-Madaria E, Buxbaum JL, Maisonneuve P, et al. Aggressive or moderate fluid resuscitation in acute pancreatitis. N Engl J Med. 2022;387(11):989-1000. [Free full text]
7. Eva L, Brehar FM, Florian IA, et al. Neuropsychiatric and neuropsychological aspects of alcohol-related cognitive disorders: an in-depth review of Wernicke’s Encephalopathy and Korsakoff’s Syndrome. J Clin Med. 2023;12(18):6101. [Free full text]
8. Chang RW, Bordage G, Connell KJ. The importance of early problem representation during case presentations. Acad Med. 1998;73(10 Suppl):s109-s111. [Available at]
9. Thammasitboon S, Rencic JJ, Trowbridge RL, et al. The Assessment of Reasoning Tool (ART): structuring the conversation between teachers and learners. Diagnosis (Berl). 2018;5(4):197-203. [Available at]
10. Schaye V, Eliasz KL, Janjigian M, et al. Theory-guided teaching: implementation of a clinical reasoning curriculum in residents. Med Teach. 2019;41(10):1192-1199. [Available at]
11. McQuade CN, Bonifacino E. Reasoning on rounds: summary statement or problem representation? J Gen Intern Med. 2024;39(4):714. [Available at]
12. Bowen JL. Educational strategies to promote clinical diagnostic reasoning. N Engl J Med. 2006;355(21):2217-2225. [Available at]
13. Oudman E, Wijnia JW, van Dam M, et al. Preventing Wernicke Encephalopathy after bariatric surgery. Obes Surg. 2018;28(7):2060-2068. [Free full text]
14. Balogh EP, Miller BT, Ball JR, Committee on Diagnostic Error in Health Care; Board on Health Care Services; Institute of Medicine; Improving Diagnosis in Health Care. Washington (DC): National Academies Press (US); 2015. [Free full text]
15. Sabin JA, Marini M, Nosek BA. Implicit and explicit anti-fat bias among a large sample of medical doctors by BMI, race/ethnicity and gender. PLoS One. 2012;7(11):e48448. [Free full text]
16. Graber ML, Rusz D, Jones ML, et al. The new diagnostic team. Diagnosis (Berl). 2017;4(4):225-238. [Available at]
17. Olson APJ, Durning SJ, Fernandez Branson C, et al. Teamwork in clinical reasoning - cooperative or parallel play? Diagnosis (Berl). 2020;7(3):307-312. [Available at]