Exercises in Clinical Reasoning: Beyond the Surface

In this series, a clinician extemporaneously discusses the diagnostic approach (regular text) to sequentially presented clinical information (bold). Additional commentary on the diagnostic reasoning process (italic) is interspersed throughout the discussion.

A 72-year-old man was sitting, awaiting his medications at a pharmacy when he suddenly collapsed. No preceding symptoms were endorsed. Bystanders were unable to arouse him and noted a blue tinge to his lips.

This older gentleman has suffered from sudden collapse/loss of consciousness while in a seated position. The lack of a prodrome and the fact that this event was not provoked by exertion make a sudden cardiac event, particularly an arrhythmic event, a can’t-miss diagnosis. In the absence of further data, I would approach this patient as if he had an out-of-hospital cardiac/respiratory arrest. As in many acute and critical care situations, diagnostics and therapeutics occur in parallel, rather than serially, and management often needs to be initiated in the face of diagnostic uncertainty. Assuming the patient was pulseless, I would immediately initiate cardiopulmonary resuscitation (CPR). I would simultaneously instruct bystanders to call 911 and locate the nearest automated external defibrillator (AED) for a rhythm check.

While initiating CPR and calling for help, I would run through the “Hs and Ts” (Table 1) until rhythm check. At this juncture, limited information is available to direct the differential diagnosis but the cyanosis of his lips makes hypoxemia a particular concern, and I would wonder if this was solely an acute process or if the patient had a chronic component of hypoxia and then suddenly deteriorated.

Table 1.

Hs and Ts—Potentially Reversible Causes of Cardiac Arrest¹⁸

Hypovolemia	Tension pneumothorax
Hypoxia	Tamponade (cardiac)
Hydrogen ion excess (acidosis)	Toxins (poisoning)
Hypo-/hyperkalemia/-calcemia/-magnesemia	Thrombosis (myocardial infarction)
Hypo-/hyperthermia	Thrombosis (pulmonary embolus)

At this stage, the discussant is operating under conditions of extreme uncertainty. Even so, she is able to recognize a potentially dangerous situation and strategize (both diagnostically and therapeutically) in the case of the most morbid possibility. How is she able to do so in a facile and expeditious fashion? While not explicitly stated, this is likely the product of reflection on her prior experience. Reflection has been identified as a fundamental tool in the early career growth of clinicians. ¹

The patient’s medical history included dyslipidemia, hypertension, coronary artery disease, and end-stage heart failure due to ischemic cardiomyopathy. He underwent orthotopic heart transplantation 8 months prior to this event. His transplant-related medications included tacrolimus, mycophenolate, prednisone, fluconazole, valganciclovir, and dapsone. He was also taking aspirin, bumetanide, clonidine, hydralazine, omeprazole, rosuvastatin, and sildenafil. He did not use tobacco products, drink alcohol, or use recreational drugs. He had worked in construction and lived his whole life in California, without any recent travel.

The patient’s post-transplant course was complicated by a right subclavian vein thrombosis. He was treated initially with enoxaparin followed by a 3-month course of warfarin, as the thrombus was felt to be provoked in the setting of his recent surgery (course completed 4 months prior to the event). Radiographic resolution of the thrombus was confirmed by ultrasound. His course was also complicated by moderate acute cellular rejection that was treated successfully with solumedrol and followed by ongoing prednisone taper. His most recent biopsy 1 month prior to the event demonstrated no evidence of rejection.

The additional history helps us reframe our problem representation—specifically, this is an immunocompromised older gentleman who is status-post orthotopic heart transplantation 8 months ago with a history of cellular rejection and provoked thrombosis.

Therefore, in this patient, the T for thrombosis in our “Hs & Ts” (Table 1) raises to the forefront of our differential. A coronary thrombosis is possible. Additionally, he recently had a subclavian thrombosis treated with enoxaparin that was presumed to be provoked. Because anticoagulation was recently stopped, we must revisit the assumption that this was a provoked thrombosis and consider DVT/PE highly on our differential as a cause for sudden cardiovascular collapse. An urgent ECG, troponin, BNP, and echocardiogram will help, as well as consideration for CT-PE once the patient has been stabilized.

In any solid organ transplant patient, we must otherwise systematically think of 5 main possibilities in our differential diagnosis: infection, rejection, medication effect, recurrence of the underlying cause, and post-transplant lymphoproliferative disorder (PTLD). Regarding infection, patients > 6 months out of the transplant window are at risk for typical community-acquired pathogens as well as fungal (e.g., coccidioidomycosis given his residence in California) or viral infections. In cardiac transplant recipients, rejection can manifest as signs and symptoms of new heart failure. The immunosuppressive and antimicrobial prophylaxis medications are common culprits of drug-drug interactions. We are suspicious for arrhythmia in this patient, and that suspicion rises even higher knowing that this is a cardiac transplant patient. The medication list should be assessed for any drug-drug interactions resulting in QT prolongation or other pro-arrhythmic states, although significant QT prolonging medicine exposure does not necessarily specify arrhythmic cause of arrest.² A tacrolimus level should be checked to evaluate for tacrolimus toxicity. Other considerations include recurrence of his underlying disease (in this instance, an ischemic event) or, rarely, a post transplantation malignancy like PTLD.

This additional information increases the complexity of this case, raising more questions than answers. Not to be deterred, our discussant deploys a schema (a framework or approach to a specific clinical situation) for problems affecting solid organ transplant recipients. Schemas are frequently employed in the diagnostic process and help clinicians to reduce the cognitive load associated with a given encounter. ³ If armed with a familiar approach to a problem (e.g., hypercalcemia being caused by parathyroid hormone–mediated or non-parathyroid hormone–mediated pathologies), then the clinician can focus her mental energy on evaluating how the current patient’s case fits within the established knowledge structure.

Emergency medical services (EMS) arrived on scene 8 minutes after dispatch. He was found to be pulseless and cyanotic, and bystander CPR was ongoing. An AED was attached with no shockable rhythm identified on initial evaluation. A laryngeal mask airway was placed and confirmed by end-tidal CO ₂ and breath sounds. Finger stick blood glucose level was 210 mg/dL. Epinephrine was given. Ventricular fibrillation (VF) was noted on rhythm check, and defibrillation was administered without return of spontaneous circulation (ROSC) and conversion to pulseless electrical activity (PEA, rhythm not specified) after shock delivery. Amiodarone and additional doses of epinephrine were given. The patient arrived in the emergency department (ED) 30 minutes after initial collapse with CPR ongoing.

By the time EMS arrived, he is in witnessed out-of-hospital cardiac arrest with initial arrest categorized as PEA. Correct management was performed by promptly starting compressions, securing the airway, confirmation with end-tidal monitoring, and checking for readily reversible causes such as hypoglycemia. On repeat rhythm check, VF was noted which then converted to PEA after defibrillation.

At this point, I would continue running through the Hs and Ts (Table 1). We have discussed hypoxemia and thrombosis as high-probability causes. Hypothermia seems unlikely and could be easily ruled out, and tension pneumothorax would be unusual spontaneously. Tamponade seems unlikely though I would quickly evaluate for a bounding JVD and obtain a bedside cardiac ultrasound as soon as possible. Hypovolemia should be assessed with a volume exam. Toxins are possible, as discussed with respect to medication use. Hypoglycemia has been excluded. Electrolyte abnormalities such as hyperkalemia and acidosis should all be appraised once labs are obtained.

The detection of VF is concerning for ischemia either as a sequelae of a non-cardiac PEA arrest or a primary ischemic etiology and is particularly pertinent in this post-cardiac transplant patient. Arrhythmias meriting special consideration in the transplant population due to postoperative sequelae such as suture lines, autonomic denervation, and graft ischemia and that may precipitate collapse include (1) sinus node dysfunction, (2) supraventricular tachycardia (e.g., atrial fibrillation, atrial flutter) with rapid ventricular response especially in the setting of depressed left ventricular ejection fraction, (3) ventricular tachyarrhythmias (e.g., in the setting of cardiac allograft vasculopathy), and (4) complete heart block. An urgent EKG should be conducted to assess for ST-segment changes concerning for an acute coronary syndrome, and particularly a STEMI. If the clinical suspicion for an acute coronary syndrome is high, after stabilization, he should be expeditiously evaluated in the cardiac catheterization lab with a coronary angiogram and endomyocardial biopsy to exclude acute ischemia and rejection, respectively. Unfortunately, as the patient has now arrived to the ED 30 min after initial out-of-hospital cardiac arrest without ROSC, neurological prognosis may be poorer.⁴ However, VF arrest does have a better prognosis than PEA arrest,^5,6 and again, ischemic cause of his arrest needs to be considered. Neurology should be consulted for consideration of cooling once there is return of spontaneous circulation.

Performance in high-stakes, high-stress scenarios can be improved through both anticipatory training and reflective analysis. Cardiac arrest is a prototypical example. Many healthcare providers are required to participate in advanced cardiac life support (ACLS) training sessions to reduce the cognitive load associated with these often stressful and chaotic scenarios. Similarly, post-code debriefs, in which the code team analyzes what went well and what could be improved upon, are recommended systems-level interventions to improve the care of critically ill patients and prevent diagnostic and therapeutic errors. At this point, our discussant calls on her practiced experience with high-stress situations to both comment on appropriate ACLS interventions, major diagnostic considerations, and prognostic factors for this patient.

In the ED, venous blood gas (VBG) was notable for pH of 7.06, partial pressure of carbon dioxide of 80 mmHg, and lactate of 11.8 mmol per liter. No pericardial effusion or pneumothorax was noted on ultrasound. The patient developed one additional episode of VF, and he underwent defibrillation followed by return of PEA (rhythm not specified). In total, resuscitation lasted approximately 50 min (30 min in-field and 20 min in ED). The patient received a total of 7 mg of epinephrine, 300 mg of amiodarone, 4 amps of sodium bicarbonate, 2 g of calcium chloride, and 1 L of normal saline. The decision to end the code was made by treating team consensus. The family requested an autopsy.

We are now able to synthesize the lab results that have returned the most quickly, notable for a VBG showing acidemia and hypercarbia, with at minimum a primary respiratory acidosis as well as likely a superimposed metabolic acidosis from lactic acidosis, consistent with a cardiac arrest and impending multi-system organ failure. After 50 minutes of resuscitation without ROSC, in this elderly immunosuppressed patient with multiple comorbidities, it is reasonable to end the code after these heroic efforts.

I am grateful that the family requested autopsy. Autopsies are frequently not conducted⁷ and yet they are incredibly powerful tools in providing closure to both clinicians and family members. This is particularly important when patients decompensate or die and there is high diagnostic uncertainty, as is often the case with code situations. Despite ostensibly excluding a number of diagnoses in this case (e.g., tamponade and tension pneumothorax), the diagnostic uncertainty frequently lingers long after a code for treating clinicians, and contributes to the “Second Victim” effect.⁸ Moreover, debriefing with a family after a code in situations where there is diagnostic uncertainty is understandably difficult. Best practices have been outlined on how to communicate with patients and families around areas of diagnostic uncertainty.⁹

How can clinicians improve their performance in diagnostically uncertain scenarios? One avenue highlighted by the discussant is the performance of an autopsy. Autopsies have been suggested to find potentially actionable causes of death missed by treating clinicians in roughly 10% of cases. ¹⁰ The information provided by autopsies, often regarded as a form of gold standard diagnostic test, is important to calibrate our own diagnostic acumen as well as inform performance of premortem testing.

At autopsy, a large branch of the right pulmonary artery contained an organizing thrombus adherent to the vessel wall with fresh thrombus layered on top, obstructing the distal artery. On microscopy, there was remodeling and recanalization of the thrombus base consistent with the thrombus being present for at least 1–2 weeks. A portion of the thrombus appeared fresh, however, suggesting an acute increase in thrombus size. In concert with chronic pulmonary emboli, the right ventricle of the heart was hypertrophied, with a wall thickness twice that of a normal heart. The cardiac allograft was unremarkable, with the anastomoses intact and coronary arteries patent. Histologically, there was no evidence of allograft rejection or myocarditis. Multiple small healed infarcts and focal areas of interstitial fibrosis within the hypertrophied left and right ventricles as well as septum were identified, reflecting remote injury of indeterminate cause. There was no evidence of infection or hemorrhage. The cause of death was determined to be acute on chronic pulmonary emboli with organizing emboli.

Since we have the autopsy results, we can further dissect our clinical reasoning throughout this case. Croskerry¹¹ and colleagues coined the term “cognitive autopsy,” a process which enables us to systematically and non-judgmentally analyze the decision-making process that could have contributed to diagnostic errors. In this case, we see on autopsy that the patient died from acute on chronic pulmonary emboli, with autopsy also showing right ventricular hypertrophy.

Revisiting my own clinical reasoning process, I had explicitly considered DVT/PE at multiple points throughout the case. One perennial question looking back with the “retrospectoscope” is whether systemic thrombolysis (e.g., tPA) should have been administered early during the code, and if so, would that have changed the outcome? The decision to administer thrombolytic therapy during cardiac arrest is often hotly debated and there are no good guidelines on when to empirically do so. While a placebo-controlled trial found that standardized administration of tenecteplase for all patients experiencing out-of-hospital cardiac arrest was not associated with improved outcomes,¹² one could argue that thrombolytic therapy warranted further consideration in this case due to the probability of PE in the setting of his prior venous thrombosis. Observational data have shown that thrombolysis during CPR has been associated with higher 30-day survival in PE-related out-of-hospital cardiac arrest.¹³ I did have some anchoring bias towards ischemia with this being a cardiac transplant patient who fluctuated between PEA and VF arrest, which is often an ischemic rhythm. A primary VT/VF rhythm has been identified as key predictor in witnessed cases of sudden cardiac arrest of having a primary cardiac, arrhythmic etiology (either ischemic or primary ventricular arrhythmia).¹⁴ In hindsight, because the initial arrest was described as PEA and the VF came later and could have been due to ischemia from a period of low cardiac output during PEA, perhaps more emphasis could have been placed on PE.

While our patient ultimately passed away, our discussant seeks to learn from their illness and improve her diagnostic skills so that future patients may benefit. Unlike the literal autopsy, the cognitive autopsy does not require the patient to have passed and is purely an educational exercise to systematically extract learning from prior patient encounters (reinforcing positive decisions and seeking to grow from mistakes). Through deliberative reflection, she identifies a potential therapeutic opportunity in the case—namely the administration of systemic thrombolytic therapy and discusses the rationale behind considering its use.

DISCUSSION

In this case, an elderly man with a history of orthotopic heart transplant experienced a cardiac arrest and ultimately died. Clinical and cognitive autopsies were utilized to help learn from this case and inform the discussant’s future practice and clinical reasoning.

A cognitive autopsy is a retrospective examination of one’s diagnostic journey through a case. It provides an opportunity to reflect on bias, error, and key pivot points in patient care as well as reinforce appropriate clinical decision-making. A cognitive autopsy is ideally conducted in close proximity to the patient care encounter, with the aim of examining both the cognitive (e.g., knowledge structures) and situational factors (e.g., distractors) that influenced the diagnostic journey. As clinical decisions are rarely made in isolation, openly performing a cognitive autopsy as a team may be an ideal strategy so that all members of patient’s care team can learn from their clinical course. An unbiased and open reflection on such factors has been suggested as an educational activity that can help mitigate future diagnostic error.¹¹

As the standard for postmortem diagnosis, clinical autopsy has long played a vital role in medical advancement and has demonstrated durable efficacy in identifying missed diagnoses in which treatment would have likely improved survival (consistently 10% of cases over the past half century).¹⁰ However, a number of factors including individual confidence in modern diagnostics and financial disincentivizing of autopsy have seen autopsy usage “vanish” recently.⁷

Without clinical autopsy, this patient’s demise would be classified as a presumed sudden cardiac death by the World Health Organization.¹⁵ The San Francisco POST SCD (Postmortem Systematic Investigation of Sudden Cardiac Death) study showed the extent of inaccuracy of this definition, demonstrating that nearly half of WHO-defined SCDs in San Francisco were nonarrhythmic and non-cardiac causes identified by postmortem investigation.¹⁶ Clearly, autopsy was incredibly useful in reflecting on this case, providing closure to family members, and informing patient care moving forward. With the knowledge that the patient died due to pulmonary embolism, the discussant was able to effectively address pivot points and potential biases.

In summary, systematic reevaluation of a patient’s case, either through cognitive or clinical autopsies, helps us grow as diagnosticians. Many valuable lessons lie beyond the surface of a given diagnostic journey, and such reflection and reconsideration can help us better care for future patients.

CLINICAL TEACHING POINTS

• There are many non-cardiac causes of sudden cardiac arrest, such as neurologic catastrophe, sepsis, and pulmonary embolism, with non-cardiac causes associated with worse outcomes compared to primary arrhythmic causes such as acute or chronic coronary artery disease.⁶ Non-cardiac causes are underrecognized without autopsy and may drive the high mortality associated with sudden cardiac arrest.

• In heart transplant recipients, arrhythmias to consider include (1) sinus node dysfunction, (2) supraventricular tachycardia with rapid ventricular response especially in the setting of depressed left ventricular ejection fraction, (3) ventricular tachyarrhythmias (e.g., in setting of cardiac allograft vasculopathy), and (4) complete heart block.¹⁷

• In the solid organ transplant recipients presenting with a decompensation, special attention in the differential should be dedicated to infection, rejection, medication effect (e.g., drug-drug interaction), recurrence of the underlying cause, and post-transplant lymphoproliferative disorder.

Funding

The work reported in this publication was supported by the National Heart, Lung, and Blood Institute under Award Numbers NIH R01 HL102090 (ZHT) and R38 HL143581 (JWS).

Declarations:

Conflict of Interest:

Dr. Tseng reports grants from NIH/NHLBI and CDC. Dr. Salazar reports grants from NIH/NHLBI. Drs. Santhosh and Minter report no disclosures.