I always wanted to be an FBI agent. The polished badge, the brown trench coat, the small notebook flipped open as another observation was hurriedly scribbled on the page — I found it all irresistibly intriguing. But what fascinated me most was the art of deduction: the satisfaction of gathering scattered details until they all made sense. Perhaps it is less surprising that I became an internist instead.
As a child, I assumed every mystery ended with an answer. Then I grew up. Real investigations don't always end with closure, and neither does medicine. Sometimes the diagnosis is clear, yet the hardest decisions begin only afterward.
Internal medicine is sometimes perceived as the specialty without a defining procedure — the one where the work consists of prescribing and coordinating care. What that perception overlooks is that much of what we do is not in performing interventions, but in deciding whether one is warranted at all. Does the patient with intermittent abdominal pain and a history of heroin use need reassurance, laxatives, or an urgent CT scan? Does a mole without any screaming ABCDEs warrant careful observation, or is it time for a dermatology referral? Complexity is rarely defined by the sophistication of an intervention. More often, it is defined by the weight of the decision and by the consequences of acting, or choosing not to.
Some cases present almost exactly as the textbooks describe — cirrhosis or shingles. Others offer the satisfaction of watching seemingly unrelated clues converge into a single coherent explanation. I recently cared for a patient with years of unexplained fatigue and mild hyperbilirubinemia who was ultimately found to have pernicious anemia. Three months after starting cyanocobalamin injections, his energy had returned, and the elevated bilirubin that had been dismissed for years had normalized. Watching physiology reveal the logic it had been following all along felt like resolution.
But medicine does not always offer that kind of clarity. Many decisions are straightforward, and evidence provides welcome guidance. In the cases that stay with you, however, the challenging part is not arriving at the diagnosis or even choosing the treatment. It is living with the uncertainty that remains afterward.
Prescribing a medication can appear deceptively simple — a few clicks, and off the patient goes to the pharmacy. The thinking behind that decision rarely is. During residency, an attending once told us that there would come a day when even prescribing aspirin would give us pause. At the time, the statement seemed absurd. Aspirin is among the oldest and most familiar medications. And then I had my first patient for whom I needed to prescribe it. He required dual antiplatelet therapy after an acute coronary syndrome but had also been found to have cerebral microhemorrhages just three months earlier. As I stared at the computer, I could still hear my attending's voice. One "what if" led to another. I searched UpToDate and combed through the literature — long before AI could summarize it all in seconds — hoping to find an answer that would make the decision for me. None came.
That was the first time I understood that evidence can narrow the possibilities without eliminating uncertainty. No algorithm to follow, no one to call in that moment at night. I prescribed the aspirin. For days afterward, I found myself checking his chart, hoping I wouldn't discover an intracranial hemorrhage.
Medicine is often presented as though one correct answer is waiting to be found. More often, several reasonable ones exist. In detective stories, solving the mystery is the ending. In medicine, it is often where the challenge begins. The diagnosis may explain the illness, but it cannot tell us what is best for the person sitting in front of us. Guidelines act like a compass. They can orient us, but they cannot guarantee we will reach a specific destination. That is where clinical judgment comes in.
Deciding which harm to risk is never straightforward. Medicine even has a word for the unintended harm caused by our own interventions: iatrogenic. Too much diuresis may relieve congestion but leave a patient volume depleted and alkalotic. An SGLT2 inhibitor may slow the progression of kidney disease while increasing the risk of recurrent urinary tract infections in susceptible patients. Sometimes the most important choice is not which medication to prescribe, but whether a patient is truly safe to return home. Other times, it is recognizing that the best prescription is no prescription at all. Explaining why an antibiotic is unnecessary for a viral illness may disappoint a patient today, yet that restraint preserves its effectiveness when they truly need it tomorrow. Likewise, anyone who has tried to deprescribe a proton pump inhibitor knows it is often easier to start one than to help a patient let it go.
Not every consequential decision in medicine involves medication. Some of the most important begin with conversations about goals of care before a crisis removes the opportunity to have them. And while moments like this rarely look dramatic from the outside, they may shape the course of a patient's life — or death — in ways that are no less profound.
I think of a patient who, after I explained two equally imperfect options, looked at me and said simply, “Whatever you think, Doc.” Nothing in the literature prepared me for the weight of that sentence. We may never know whether another path would have led to a better outcome. The cases I remember most are not the ones where the clues led to a clear answer. They are the ones where the evidence ended, and we were left to navigate uncertainty together. Not the tidy mystery I imagined in my youth, but meaningful, messy work all the same.
How do you deal with uncertainty in your practice? Share in the comments.
