The Second Cut: DeBakey, the Shah, and the Hubris of the Comforting Explanation

1. The Legend, the King, and the Hostile Abdomen

On March 28, 1980, in a military hospital on the banks of the Nile, Michael DeBakey removed a spleen the size of a football from a dying exile.

The patient was Mohammad Reza Pahlavi, the deposed Shah of Iran. The spleen weighed close to 2,000 grams, riddled with tumorous nodules from a lymphoma whose severity had been concealed from the Shah himself for years. The surgeon was the most celebrated figure in twentieth-century American surgery, a man who had performed tens of thousands of cardiovascular procedures and whose very name was synonymous with surgical mastery. [1][2][3]

DeBakey is one of surgery’s heroes.  Reading Gerald Imber’s Cardiac Cowboys illustrates the magnitude of what he achieved. He completed a general surgery residency at Charity Hospital in New Orleans in the 1930s, trained in vascular surgery under Leriche in Strasbourg, then built Baylor College of Medicine into one of the great medical centers on earth. He pioneered the Dacron graft, credited for early successful coronary artery bypass, developed the LVAD, and trained a generation of surgeons. But what Cardiac Cowboys also makes clear is the trajectory: by the 1950s, DeBakey’s practice had moved into the chest and the great vessels. By the 1970s, he had become chancellor of Baylor, an administrator-surgeon whose clinical work was almost entirely cardiovascular. His general surgery residency was 45 years behind him. [5][9][11]

By 1980, DeBakey was 71, running an academic medical center, and was reported by contemporaries to have had limited recent experience with this type of abdominal operation.  He was recruited for the surgery due to concern about operative risk. Several physicians on the case noted the mismatch. Flandrin, the French hematologist who had been covertly managing the Shah’s leukemia since 1974, had little information about DeBakey’s general surgery experience. Other surgeons were more blunt: DeBakey was no surgeon of the abdomen. [1][3][4]

Standard teaching for elective splenectomy is mobilize the spleen, identify the splenic artery and vein separately, retract the tail of the pancreas away, ligate and divide each vessel individually. DeBakey and his first assistant, Dr. Gerald Lawrie, did not do this, perhaps because the spleen was massive and it was impossible to safely isolate each vessel, they applied a clamp across the entire bulk of the splenic hilum. They clamped across the splenic artery, the splenic vein, and, because the massive spleen’s inflammatory adhesions had fused the pancreatic tail into the hilum, the pancreas itself. The hilum was transected and the stump oversewn. [1][2][6][7][8]

Surgeon and historian Leon Morgenstern cited Lawrie’s account: the pancreatic tail was embedded in the hilum, the vessels were controlled by clamping, the team recognized that pancreatic tissue had been divided, the stump was meticulously oversewn, and drainage was not felt to be necessary. Dr. Nour, the Egyptian surgeon present, later testified that he saw the tail of the pancreas in the specimen bucket. Flandrin found pancreatic tissue in the histology of the splenic hilum. [1][2]

No drain was placed. DeBakey announced to the press that the operation had been successful. This announcement matters for what follows. A surgeon who tells the world the operation went well has created a narrative that is now publicly staked. Every subsequent data point suggesting otherwise must overcome not only clinical uncertainty but the reputational cost of contradiction. [1][2][3]

Three days later, the Shah developed low left-sided thoracic pain. Flandrin and Nour suspected a subphrenic abscess from a pancreatic tail injury. DeBakey and Lawrie denied any pancreatic injury and stated that no drainage was necessary. DeBakey returned to Texas, then back to Cairo when the Shah worsened, and attributed the deterioration to chemotherapy toxicity. He reduced the chemotherapy dose. Weeks passed. By June, French physicians confirmed what Flandrin had been saying since April. Surgeons eventually intervened, but they entered a quagmire of established infection, an entrenched catastrophe that had festered for weeks. [1][2][3]

The Shah died on July 27, 1980. [2][3]

 A pancreatic injury during removal of a massive, tumor-riddled spleen is a known risk. What makes this case instructive is everything that happened after: the absence of a drain, the dismissal of early warning signs, the attribution of deterioration to a convenient alternative explanation, and the weeks-long delay before intervention. These forces are structural. They operate on all of us.

The question this essay is about is not “Should DeBakey have operated?” It is the harder question: once the trajectory deviated, why didn’t he take the Shah back to the OR?

2. The Decision That Hurts More Than the First

Every surgeon understands viscerally that the decision to reoperate is categorically different from the decision to operate.

The first operation is forward-looking. You have imaging, labs, a plan, and time. The mood is anticipatory and hopeful. The core tenet of a reoperation is temporal: you must look backward and forward simultaneously, and the two directions are in conflict. You are acknowledging that something has not gone as predicted, then intervening again under worse conditions: inflamed tissue, adhesions, a patient with less reserve, and a social environment charged with the implication of error.

Operating the first time feels like courage. Redoing feels like confession.

This asymmetry distorts judgment in predictable and invisible ways.  The distortions are invisible because they feel like clinical reasoning.

3. Post-op Care as Trajectory Monitoring

In my previous essay, I described surgery as a series of nested predictions. The framing applies especially to the postoperative period, where the surgeon’s primary job is surveillance: monitoring the patient’s trajectory against an expected recovery arc.

The expected trajectory is a map. Vitals, labs, imaging, drain output, the patient’s subjective report: these are all signal of the patient’s physiology.  Every map is a compression of reality with loss of data. The losses are systematically biased toward reassurance.

Each postoperative data point has a benign explanation and a concerning one. The benign explanation is almost always more probable in any given instance. It is also more comfortable, more consistent with the narrative that the first operation went well, and more aligned with the social incentives facing the surgical team.

This is where the map fails. Not because the data are wrong, but because the interpretation is filtered through a prior belief contaminated by motivated reasoning.

In the Shah’s case, the map read “successful splenectomy, chemotherapy toxicity.” The territory was a pancreatic injury producing a subphrenic abscess. The map was plausible. The territory was killing the patient. And the people with the strongest investment in the map’s accuracy were the people who had drawn it.

I regularly operate in map-poor territory. When I perform reconstruction of the proximal ureter after prior surgery with urine leaks, the tissue is completely fused: ureter, peritoneum, bowel, the renal hilum, and on the left side, the tail of the pancreas.  Everything is matted into a single inflammatory mass. There are no tissue planes. The CT shows a dilated kidney, a nephrostomy tube and maybe a fluid collection. It does not show where the ureter ends and the renal vein begins.

In that setting, I dissect very close to the structures I know: the ureter itself, the renal artery, the renal vein. Too close for comfort, sometimes. There is a physical quality to the moment when the tissue you are dissecting stops looking like what you expected and starts looking like something else, a subtle resistance to dissection that does not match your model. That is the moment you are operating in unmapped territory, and the only proper response is to slow down and re-examine every assumption about what you are touching. I build mental models of what I need to see and preserve to prove to myself that the gonadal vein is not the ureter, that the bowel is intact, and that I have not entered the pancreas. The strategy is to hug the known structures, because in map-poor territory the only landmarks are the things you can identify with certainty. 

But hugging the known structures carries its own devastating risk.  The ureter's microscopic blood supply travels longitudinally along its outermost layer of connective tissue. If you dissect too close, you strip that blood supply. You leave behind a visually intact, perfectly dead ureter.

I recently did exactly this. When the patient spiked a delayed fever on postoperative day two, I did not consider ischemia. I told myself it was atelectasis, that the patient was noncompliant with incentive spirometry. When clear fluid began draining from the incision, I delayed obtaining a CT scan, deciding the output was merely lymphatic fluid from the extensive retroperitoneal dissection. When the scan finally forced me to acknowledge a fluid collection and a frank urine leak, I retreated to hope: the defect was small, the tissue would granulate, it would heal on its own. It took weeks for me to admit the catastrophe. The territory was a necrotic ureter leaking urine uncontrollably into the flank. The map was my own ego, and I had defended it exactly the way DeBakey defended his. Unlike the tragedy of the Shah, I was able to resolve my error with a complete ureteral substitution, but at the cost of unnecessary pain and suffering.

4. The Bias Stack: Sunk Cost, Hierarchy, and the VIP Distortion Field

The reoperation decision sits at the intersection of several cognitive biases, each reinforcing the others.

Sunk cost and escalation of commitment. The first operation represents an enormous investment: hours of operative time, a patient’s physiologic reserve, the surgeon’s reputation. Reoperation acknowledges that this investment has not yielded the expected return. The social cost of that admission is so high that the default is to reinterpret deterioration as a temporary setback rather than evidence of a structural problem.

Outcome bias and resulting. “The operation went well” is a narrative, not a clinical assessment. Once established, it becomes a protective talisman. DeBakey did not merely believe the operation was successful; he told the international press it was successful. The public declaration created a reputational position that subsequent clinical data had to overcome, not just clinical uncertainty. Every concerning sign now carried two costs: the clinical cost of acting on it, and the reputational cost of contradicting a public statement by the most famous surgeon alive.

Hierarchy and authority gradients. The people with the least surgical status (Flandrin, a hematologist; Nour, an Egyptian surgeon operating in a foreign authority’s shadow) were the first to identify the complication. The person with the most status was the last to accept it. Status makes it costly for others to disagree with you, and costly for you to be wrong.

VIP syndrome. The Shah was a deposed monarch, the subject of an international hostage crisis, and a political instrument of multiple governments. VIP syndrome does not mean VIPs receive worse care intentionally. It means that stakeholders, scrutiny, and political consequences create a distortion field in which clinical judgment becomes subordinate to reputation management. [10]. While the syndrome affects VIPs most acutely, the underlying dynamics are universal. Every patient interaction contains stakeholders, scrutiny, and, in a healthcare marketplace increasingly shaped by Google reviews and online ratings, the pressure of reputation management.

5. What a Rationalist Surgeon Would Have Done

The critique of DeBakey is easy. The steel man is harder and more instructive.

The spleen weighed 2,000 grams. The splenic hilum in massive splenomegaly is a distorted, hypervascular trunk where the pancreatic tail can be fused into the pedicle, invisible as a discrete structure. DeBakey was operating on another continent, in a military hospital, with an unfamiliar team. He was 71. He was reported by contemporaries to have had limited recent experience with this type of abdominal operation. The world was watching. The perceived risk of catastrophic intraoperative hemorrhage was real and immediate. [1][2][3]

DeBakey’s presence in the room was politically inevitable but clinically misaligned. DeBakey’s hands on the splenic hilum was the coordination failure. But this failure was upstream of the operating room. It was produced by the same forces: hierarchy, VIP pressure, political urgency, that would later prevent the complication from being recognized. The right domain expertise was not matched to the operation. The authority gradient made it impossible for those who recognized the mismatch (Flandrin, Nour, the Panamanian surgeons who had offered to perform the splenectomy themselves) to override the selection. There were no Toyota way “stop the line” norms. The incentives to defer to fame, to proceed, to avoid the political cost of delay or handoff, were overwhelming. And after the operation, the same gradient that prevented the right expertise from entering the room prevented the right information from being acted upon for weeks.

6. When Data Lies: Drains, Scans, Fevers, and Comforting Explanations

DeBakey did not place a drain after the splenectomy. This is a decision with asymmetric consequences. If the pancreas is intact, the drain is a minor nuisance. If the pancreas is injured, the drain is an early warning system that converts an occult catastrophe into a visible, manageable problem. The absence of a drain did not cause the pancreatic leak. It eliminated the signal that would have detected it early.

I should note the obvious counterargument: drains are also imperfect sensors. They clot, they migrate, they provide low output that can be falsely reassuring when the drain is not sitting in the collection. A drain in the splenectomy bed could have provided its own comforting explanation: “The drain is dry, so there is no leak.” This is true, and it is a genuine feature of wicked environments. But noisy data is better than no data.

The decision not to gather information is itself a decision, and its consequences compound over time. A surgeon who does not place a drain cannot check the output for amylase. A surgeon who does not obtain early imaging cannot see a fluid collection before it becomes an abscess.

The Shah’s fever was attributed to chemotherapy toxicity. This is what I will call a comforting explanation: an alternative diagnosis that is plausible, consistent with the clinical picture, and, crucially, does not implicate the initial operation. Every postoperative fever has a comforting explanation. Atelectasis. Drug reaction. Catheter Associated Urinary Tract Infection.  These explanations are not wrong. They are dangerous, because they allow the surgeon to avoid confronting the more concerning possibility without appearing to avoid it.

The discipline required is to ask not “What could explain this?” but “What is the most dangerous thing this could be, and what would I need to see to rule it out?”

Comforting explanations are not unique to surgery. In the years before the 2008 financial crisis, rising default rates in subprime mortgages were often framed by policymakers and market leaders as likely to be “contained to the subprime sector,” a comforting explanation that delayed recognition of broader risk.  The structure is identical: a plausible benign interpretation that aligns with the interests of the people making the interpretation, accepted not because it is the best explanation but because it is the least costly one.

7. The Late Truth and the Necrotic Pancreas

Flandrin had been the Shah’s covert physician since 1974, flying to Tehran under the guise of a “nutrition consultant.” He knew the patient’s disease better than anyone. He did not perform the splenectomy. He was not responsible for the postoperative course. But he was the person in the room who kept insisting the Shah had a subphrenic abscess, and he kept being overruled by physicians with more surgical authority and more investment in the existing narrative. [2][3]

When intervention eventually occurred on June 30, 1980, Dr. Pierre-Louis Fagniez opened the Shah’s abdomen through a small left subcostal incision and drained 1.5 liters of pus mixed with necrotic pancreatic debris.  The injured pancreas had been leaking enzymes into the Shah’s abdomen for 90 days. Morgenstern concluded that the interval between the injury and the drainage was too long, and that the delay was the proximate cause of the rapid decline. [1][2]

8. Practical Doctrine: Build Your Reoperation Triggers Before You Need Them

If the Shah’s case teaches anything generalizable, it is that the decision to return to the OR is difficult to make well in real time, because real time is precisely when there is maximum bias.

I attempt to reduce the bias by pre-committing to decision rules before the biases activate. Not rigid algorithms. Decision triggers: conditions that override the default toward watchful waiting and compel a structured reassessment. 

Before the first operation, ask: “How might this fail, and what would early failure look like?” After the operation, ask daily: “Is this patient’s trajectory matching my prediction, and if not, what is the most dangerous explanation?”

The distinction between this question and “How is the patient doing?” is critical. “How is the patient doing?” invites a narrative answer. “Is the trajectory matching my prediction?” invites a specific one.

Kind features (clear signal):

- Anastomotic disruption on contrast study: binary, visible, actionable

- Hemorrhage with hemodynamic instability: unambiguous

- Drain output that is frankly purulent or feculent

Wicked features (noisy signal):

- Low-grade fever with plausible alternative explanations - Imaging findings that are “probably normal postoperative changes”

- Subtle tachycardia in a patient with pain, anxiety, and opioid requirements

- A trajectory that is “not improving as expected” but not frankly deteriorating

- The patient’s subjective sense that “something is wrong” without localizing findings

These practices help, but even pre-registered red flags are subject to motivated reasoning in the moment.  Thus, wicked features are where surgical judgment lives and where bias does its damage.

9. Concrete Practices

1.        Pre-register your red flags. Before a risky operation, document the specific complications you consider most likely and the earliest signs that would indicate each one. Write them down. The written record anchors you against narrative drift.

2.        Ask the daily deviation question. Not “How is the patient doing?” but “What would convince me right now that this patient has a leak?” This is not a pleasant habit. It is a daily exercise in looking for evidence that you have harmed someone. The temptation to skip it, to trust the narrative, to accept “doing well” as an answer, is precisely proportional to how much you need to ask it. The days when the question feels unnecessary are the days when the bias stack is winning.

3.        Force a disconfirming opinion. When the trajectory is ambiguous, ask a colleague uninvolved in the first operation to review the data independently. The uninvolved colleague sees data without narrative.

4.        Treat drains and early imaging as sensors, not as defensive medicine. A drain near a potential pancreatic injury is not overcautious. It converts an occult process into a visible one. Imperfect sensors beat no sensors.

5.        Discipline your language. Replace “the patient is doing fine” with “the patient is matching the expected trajectory for postoperative day two.” Replace “we’ll keep an eye on it” with “if the heart rate remains above 100 at 6 PM, I will obtain imaging.” Vague language permits vague thinking.

10. Coda: Reoperation as a General Problem in Wicked Systems

The reoperation decision is, in miniature, the problem of admitting failure in any wicked learning environment. The feedback is ambiguous. The stakes are high. The social incentives punish early acknowledgment that something has gone wrong and reward narrative continuity. The people closest to the original decision are the least equipped to evaluate it, because they are most invested in its success. This is the structure of narrative lock-in. It is the same cognitive trap that kept the Pentagon doubling down on body counts in Vietnam.[12]

In every domain, the original decision creates a gravitational field that bends subsequent interpretation toward confirmation. The "reoperation" is delayed not because the evidence is absent but because the people who would have to act on it are the same people who would be implicated by it.

The Shah’s abdomen was hostile territory. But the truly hostile territory was the decision environment surrounding it: the confluence of fame, hierarchy, politics, and narrative protection that delayed a straightforward surgical judgment by weeks. The pancreatic injury was a complication. The delay in addressing it was a coordination failure.

The question is never only “What is happening inside the patient?” It is also “What is happening inside the hospital, inside the team, inside the surgeon’s own mind, that might prevent the truth from being acted upon in time?”

That is the second cut. It is always harder than the first.

Full Source List

1. Leon Morgenstern. “The Shah’s Spleen: Its Impact on History.” Journal of the American College of Surgeons 212, no. 2 (2011): 260–268. doi:10.1016/j.jamcollsurg.2010.10.014. PMID: 21193329. https://pubmed.ncbi.nlm.nih.gov/21193329/

2. Ardavan Khoshnood and Arvin Khoshnood. “The death of an emperor – Mohammad Reza Shah Pahlavi and his political cancer.” Alexandria Journal of Medicine 52, no. 3 (2016): 201–208. doi:10.1016/j.ajme.2015.11.002. https://www.sciencedirect.com/science/article/pii/S2090506815000822

3. William Shawcross. The Shah’s Last Ride. New York: Simon and Schuster, 1988.

4. U.S. Department of State, Office of the Historian. Foreign Relations of the United States, 1977–1980, Volume XI, Part 1, Iran: Hostage Crisis, Document 215. https://history.state.gov/historicaldocuments/frus1977-80v11p1/d215

5. Gerald Imber. Cardiac Cowboys: The 100-Year Rivalry Between the Heart Doctors and the Heart Surgeons. New York: Kaplan Publishing, 2008.

6. Richard Davis. “Elective Splenectomy.” Global Surgical Atlas, Vanderbilt University Medical Center. https://www.vumc.org/global-surgical-atlas/sites/default/files/public_files/PDF/Elective%20splenectomy.pdf

7. Kenneth L. Mattox, Ernest E. Moore, and David V. Feliciano, eds. Trauma, 8th ed. New York: McGraw-Hill Education, 2017.

8. AHPBA. HPB and Transplant Surgery: Trauma (Section 5). 2020. https://www.ahpba.org/wp-content/uploads/2020/07/HPB-Dilemmas-Section-5-Trauma.pdf

9. Thomas Parke Hughes and John P. McGovern Historical Collections and Research Center, Baylor College of Medicine. “Legacy of Excellence: Michael E. DeBakey.” https://www.bcm.edu/about-us/our-campus/debakey-museum/legacy-of-excellence

10. Guzman JA, Sasidhar M, Stoller JK. “Caring for VIPs: Nine principles.” Cleveland Clinic Journal of Medicine78, no. 2 (2011): 90–94. doi:10.3949/ccjm.78a.10113. PMID: 21285340. https://pubmed.ncbi.nlm.nih.gov/21285340/

11. Encyclopaedia Britannica. “Michael DeBakey.” https://www.britannica.com/biography/Michael-DeBakey

12. U.S. Department of State, Office of the Historian. Foreign Relations of the United States, 1964–1968, Volume V, Vietnam, 1967, Document 375, “Draft Memorandum From Secretary of Defense McNamara to President Johnson,” November 1, 1967. https://history.state.gov/historicaldocuments/frus1964-68v05/d375