By Craig Smith, MD
Article first published on Dr. Smith's Substack, Long Incision.
First, a pretentious epigram:
“The facts belong to the task, not to the solution.” (Wittgenstein)
Something to ponder on the path.
A fictionalized/composite patient of mine needs her aortic valve replaced. The valve sits inside a prosthetic graft I used to replace her aneurysmal ascending aorta 12 years earlier. Sticky fibrous bands of scar called adhesions will have formed around everything we touched back then. The adhesions of greatest immediate concern are those that form between the inner wall of the sternum (breastbone) and important underlying structures like the aorta, the prosthetic graft, the innominate vein, and the right ventricle. How do I prepare for this reoperation?
The goal of preparation is to maximize the probability of a good result and prevent complications by anticipating problems. Is anticipate a synonym of predict? Not quite. The hockey player who anticipates where the puck will be doesn’t just place a bet with Polymarket pinpointing that spot on the ice, he’s there to take the puck when it arrives. Definitions of “anticipate” coalesce around “to foresee and forestall.” Foreseeing what can possibly go wrong is the first and most essential step; foresight becomes a kind of prediction when expressed as a probability. To forestall is the next step. If there is more than one possible maneuver that might prevent (forestall) the problem, the various options can be ranked based on probabilities—another prediction. To that end, in the case I’m describing the first step of preparation is a CT scan, done to show the anatomic relationships between all the structures in the thorax.
I study the CT scan and decide to approach this case as a relatively routine reoperation because the probability of injury to any critical structures appears low, and if an injury does occur, I think I can deal with it and still achieve a successful valve replacement. Asked to state the probabilities, I say the risk of injury is less than 7-8% and the risk of an injury that compromises the larger goal is less than 2%. What is the basis for these estimates? Decades of experience, my own and that of others.
On the day of the operation other elements of routine preparation include the simple choice of an oscillating sternal saw rather than a reciprocating saw. I manage the personnel assignments to assure that I have a first assistant skilled enough to be useful if things go seriously wrong. I start the operation by cutting through the previous incision to expose the sternum. Now my preparation is put to the test, and as I was reminded recently, the step from preparation to action is always a leap into the unknown. I divide the sternum with the oscillating saw, free the innominate vein and the right ventricle from the sternum uneventfully. I’m about 80% through the period of greatest hazard when I find a spot where the prosthetic graft is tightly adherent to the sternum. How tightly? I’m working my around it when I make a hole in the graft and arterial blood is suddenly pumping into the field. The patient’s blood pressure begins dropping almost immediately.
A new menu of actions immediately clicks in place for everyone in the room. My first assistant is highly experienced after five years of cardiac surgery training. He’s already moving towards the patient’s groins before I tell him to “get her on bypass” (connect her to the heart-lung machine) as quickly as possible. The Physician Assistant takes over as my first assistant—now all that’s required of my assistant is to pull on the retraction hook with one hand to help me see the problem, and suck blood out of the way with the other hand.
Under the general heading of “aortic injury during dissection through adhesions,” this is still a very manageable situation, and I decide to approach it by rapidly mobilizing the graft so that I can see the injury more completely and control it while my resident is getting us on bypass. In my experience that almost always works. Almost in this case is where I conceal my working knowledge that a hole in a prosthetic graft is less forgiving than one in an aorta, because the biological aorta is thicker-walled and more elastic. That reality comes out of hiding when I see how torrentially this innocuous little hole in the graft is bleeding, once fully revealed.
I manage to achieve control after several minutes of rapid dissection, but her blood pressure is still very low. We transfuse as fast as we can. The team has the patient on bypass about fifteen minutes after the bleeding starts. The operation proceeds as planned from that point on, very smoothly. She is transferred to the ICU with her heart and other major organs functioning normally—save one. Her brain died during the period of low blood pressure while we were getting control, and she becomes a non-cardiac organ donor. Generously setting aside causality as a complicated question for another day, at the very least I failed to prevent a tragedy for her and her family.
Was I unprepared? I have freed an aortic graft from a sternum many times, and I have dealt with injured grafts before. With all due humility, based on my decades of experience it would be hard to find anyone better prepared for the general category of structure-stuck-to-sternum challenges. Could I have prepared better for this specific case? Certainly, in retrospect. I could have interpreted the CT scan as showing more dire adjacency of graft and sternum. With or without that cautionary insight, I could have chosen one of several options that can speed up connecting the patient to the heart-lung machine, options that include putting the patient on full cardiopulmonary bypass before even reopening the sternum. Why would that help? Because on the heart-lung machine the pressure is lower in the aorta, the atria and ventricles are collapsed and less easily injured, oxygen delivery to the brain and the rest of the body is not dependent on blood pumped by the heart, and even massive bleeding can be much more easily managed by capturing shed blood with suckers and reinfusing it through the heart-lung machine. At extremes of risk, the patient can be cooled to 18°C before reopening the sternum, because cold temperature makes the brain more tolerant of low blood flow.
If you’re wondering why we don’t simply put every patient having a reoperation on the heart-lung machine before reopening, it’s because there are trade-offs to consider. Bypass circulation is a time-limited miracle. The negative effects of low body temperature, continuous small-vessel bleeding, consumption of the body’s clotting factors, and transfusion are ongoing and cumulative. Once disconnected from the heart-lung machine bleeding can still be a problem for some of the same reasons; clotting reactions are less suppressed, in part because temperature is higher, but the ability to form clot often remains impaired. Persistent diffuse small vessel bleeding makes it harder to distinguish one structure from another and to locate specific sites of bleeding. For myriad reasons, bleeding begets bleeding. Prolonged circulation on the heart-lung machine and hours spent stopping the bleeding afterwards also take a gradual toll on the function of the liver, kidneys, and lungs, even when the brain remains unharmed. A policy of maximal-caution preparedness that used prophylactic cardiopulmonary bypass in every case would do far more harm than good. A more tailored approach to the use of that preparedness tool depends on the surgeon’s ability to anticipate—to foresee and forestall—by estimating the probability of injury in advance. In the case described, I underestimated, and the patient died.
A constant of preparation for many surgeons, including some excellent surgeons, is an elaborate mental rehearsal of the steps involved in the operation ahead, no matter how many hundreds of times they’ve done that type of operation. Me? Usually not. I prefer to think I’m not too lazy, too cavalier, or too overconfident to rehearse. I do a thorough mental rehearsal before any new operation, or before one I expect to be unusually challenging. To be effective, the possible twists and turns (“if this, then that”) must be part of the rehearsal. But even the most elaborate rehearsals are limited to situations you’ve thought to rehearse.
Obsessive mental rehearsal can also become mechanical and ritualistic, like running through one’s Rosary beads, which can make the shock of unanticipated surprises even harder to handle. I’ll offer a more personal analogy. Quite a few years ago I started memorizing piano music, thinking that a combination of muscle memory and deep familiarity would somehow transcend the notes and overcome the performance aversion that has limited me to an audience of one (me). What I’ve learned is that the memorized music flows out of me like fine wine—until something interrupts my trance. The interruption can be anything—an unexpected listener, a noise, a sticking key, or the sudden awareness that I’ve just powered through several bars on muscle memory alone with my mind elsewhere. Whatever it is, I’m instantly, hopelessly lost, and my audience of one can only cringe.
Whenever preparation falls short, and for whatever reason, a complex and fully human combination of experience, instinct, and intuition must step in. Artificial intelligence can substitute for that, someone must be thinking. An AI “agent” could run through all the describable possibilities trillions of times, then advise me. That might be excellent augmented preparation, but the second I start doing what the agent told me to do, a new future is spooling out, certainty is gone, and my agent and I must react to that new future. And fast. Once I had a hole in the graft, I had seconds to minutes for decision-making. No time to send the script back to the writer’s room.
At the end of our path, what’s up with Wittgenstein? I think of the task in this context as the object of preparation, on which all available facts are brought to bear. The solution is sought only after that profound, provocative leap into the unknowable future. The facts might have had some predictive value, in preparing for the leap—but that was then, this is now. Does “this is a fact” have a future tense? Perhaps, for certain principles of math and physics. Perhaps not, in the sandy colosseum of life writ large. Zadie Smith rued “…the wicked lie, that the past is always tense and the future, perfect.” A lie because it’s the other way around—the past is perfectly complete, and the future tense.
Related:
- What is Clinical Excellence?
- A Noteworthy Connection
- An Interview with Dr. Craig Smith, Heart Surgeon, Chair of Surgery, and Author of Nobility in Small Things
- Lessons Learned: Surgeon Craig Smith Reflects on Career in the OR