Paper:PSI-2026-0006

Abstract: My take on the perils of trying to do good and how to minimize harm in the process.

“Primum non nocere” — first do no harm — gets a lot of press in lay circles. Per the dictum sometimes attributed to Hippocrates, physicians are supposed to prioritize minimizing the harmful impact we have on our patients above anything else. It’s notable that this dictum doesn’t actually make it into most of the oaths that US physicians recite. The one that I recited doesn’t even mention the word “harm.”

While I don’t put much weight in oaths, it is striking that this canonical part of the Hippocratic oath has been removed in an era where medical errors are purportedly killing tens and hundreds of thousands of hapless Americans every year.

More striking to me, though, is what I have seen in observing US physicians up close for many years. I believe that there is a systematic, unconscious under-appreciation and under-estimation of the harm that we as physicians inflict upon our patients. For reasons too many to innumerate here, doctors are partially blind to the suffering endured by our patients as a result of our influence upon them.

I believe that the idea that we might even strive to not harm our patients is a fool’s errand. We begin harming our patients before we even meet them. I would appreciate it if you would entertain with me a definition of harm that is broad, such that it includes taking time off work, travel, financial cost, worry, and feeling demeaned. With such a definition, it becomes clear to anyone who’s had much of any interaction with any healthcare system that the harm starts early and is usually ongoing throughout the interaction. Most visits are during work hours, cost money, are anxiety-inducing, and the forms… They also frequently start late and are unpredictable in duration, making scheduling even more challenging and burdensome.

Most of that happens before the physician even meets the patient. Once they do, we find a minefield of potential harms. The patient may be quite uncomfortable with the provider, feel judged by them, feel demeaned by them, feel that their needs are not being addressed. They may have prior traumatic experience recapitulated and feel the ripples from that trigger for days, weeks, and months. Even if the physician is closely attuned to these possibilities, they may well be totally unaware of some impacts.

And this is all before the physician “does anything.” This is where things get especially messy. You see, we as physicians love doing stuff. The vast majority of us got into this business in one way or another to help people, and when someone comes to us with a problem for which we have a solution, there are few things that make us more satisfied. And since most physicians have a relatively limited arsenal of tools at their immediate, convenient disposal, we tend to rely heavily on the ones we know and have access to, which will vary widely by specialty and individual.

The problem with doing stuff, though, is that most anything that does much of anything does both good and bad stuff. Literally everything has side effects. If we again entertain the notion that harm should be conceptualized broadly, we see that all interventions incur some sort of harm. Even the best sugar pill costs money and requires swallowing a pill on a regular basis — it might even make you nauseated.

The point is that with every intervention we must weigh the harms and the benefits — the pros and the cons. We must carry in our mental balance the potential harms of an intervention on one dish and the potentially therapeutic benefit on the other. We must then decide which prevails, harm or benefit, and proceed from there.

The problem with this exercise is that this is infinitely easier said than done. If we’re lucky, we may have a halfway decent idea of how likely a medicine is to work for a person and maybe by about how much. We can figure out NNTs and effect sizes and get a rough idea of our ‘benefit’ dish. The ‘harm’ dish, though, is quite a bit trickier.

First, it’s extremely difficult to know how to compare harms and benefits. Like, is a NNH for new-onset diabetes of 53–67 worth a NNT of 217 for a nonfatal heart attack? How do we even compare these things? Not to mention that for any intervention, there tend to be almost countless potential sources of harm, many either rare or largely benign. How, then, are we to sum these harms and prepare our harm dish?

Beyond these difficulties, there are more fundamental reasons that our harm dish is so uncertain. Harm is fundamentally more difficult to measure than benefit. In general, when we are testing an intervention for a certain problem/condition, we have relatively clear, measurable targets that tend to improve rapidly after initiation of an intervention, even if doesn’t work.

Harm, on the other hand, is quite a bit more nebulous. Harms can frequently imperceptible, rare, delayed, ambiguous, unexpected, present in all study arms, or any combination thereof. This is not to mention the perverse incentive that many researchers have to exaggerate benefit and minimize harm. It turns out that no matter how carefully we study it, there will remain unforeseen, unpredictable harms of any intervention.

This all leaves us with an irrevocably uncertain mass adorning our harm dish, which we know is almost certainly larger than we think. How, then, do we compare this amorphous mass to our slightly more certain benefit dish? Very carefully, I suppose.

Such is the nature of things that the mass of neither harm nor benefit will ever be known with any meaningful precision, and the weighing of each will be idiosyncratic to each individual and each therapeutic interaction. This is much of where the art of medicine resides, as these calculations are profoundly complicated and grounded largely in individual values and intuition. Thus, our conviction as healers and lifelong learners must be to hone that intuition and align it as best as we can with the reality of the world and the values of our patients.

So, what to do?

I think we have to try our damndest to make the benefits of meeting with us outweigh the harms, and to acknowledge and mitigate whatever harms we can.