This article was originally published on KevinMD.com, 8/1/2026.
In nearly every occupation in America, women live longer than men. This is one of the most stable findings in public health, true for teachers, lawyers, engineers, retail workers, and factory workers alike. Medicine is the exception.
A 2025 analysis in JAMA Internal Medicine, led by researchers at Harvard Medical School, examined more than 3.6 million deaths recorded in the National Vital Statistics System between 2020 and 2022. Physicians as a group had a lower mortality rate than the general population, but that protective effect belonged almost entirely to men. The mortality advantage that female physicians should have enjoyed, simply by virtue of being women, did not fully materialize. Black female physicians fared worst of all, with higher mortality than every other physician subgroup.
Suicide tells the same story in sharper relief. A 2025 study in JAMA Psychiatry, drawing on the National Violent Death Reporting System, found that male physicians die by suicide at a lower rate than men in the general population, the classic, decades-old finding. Female physicians are the inversion, with a 53 percent higher suicide risk than women who are not physicians. An independent meta-analysis spanning twenty countries and nearly a century of data confirms the pattern is not a fluke of one dataset: male physicians roughly track the general population, while female physicians carry a suicide rate ratio of 1.76, significantly elevated and essentially unchanged across a hundred years of medical progress.
Sit with that. We have brighter operating rooms, better drugs, safer anesthesia, and infinitely more sophisticated diagnostics than physicians had in 1935. And the woman treating you is still more likely to take her own life than the man is, relative to their respective peers outside medicine.
What we know, and what we don’t
We do not have a study that traces a straight line from any one of these factors to the extra deaths. Nobody has proven causation here, and anyone who tells you they know exactly why women physicians are dying sooner is overstating the science. What we have is a set of well-documented features of women’s experience in medicine, each independently studied, each plausible as a contributor, none of them confirmed as the mechanism. Naming them is not the same as explaining the mortality gap; it’s naming the most likely places to look.
The same caution applies to what they’re dying from. The sex gap shows up in cancer and chronic respiratory disease, not cardiovascular disease as commonly assumed, with suicide elevated but a smaller share of the total excess. None of it has been broken down by subtype, and none of it has been tied to a causal pathway. We know where the excess deaths concentrate. We don’t know why.
Delayed help-seeking: Physicians are trained to triage everyone else’s symptoms before their own, and women physicians, still disproportionately responsible for the mental load of home life, are triaging on two fronts simultaneously. Studies of physician health-seeking behavior document a real pattern here: later self-diagnosis, later self-treatment. Whether it accounts for any specific share of the mortality gap is not something the current research can tell us.
The emotional labor tax: Women physicians are asked to do more of the invisible work of medicine: the reassurance, the difficult family conversations, the smoothing over of a colleague’s bad day. This labor is real, it is unpaid, and it is disproportionately assigned by gender, a second shift layered on top of clinical duties that never appears on an RVU report.
Identity foreclosure: Medical training asks trainees to subordinate every other part of the self to the physician role for the better part of a decade, at precisely the developmental window when women are also expected to make decisions about partnership and childbearing. Many women physicians never get the chance to develop an identity independent of the white coat, which means when the white coat becomes a source of suffering, there is no other self to retreat to.
Moral injury: The now-familiar term for what happens when a clinician is repeatedly forced to act against their own conscience by a system optimized for throughput and liability rather than healing. Moral injury is distinct from burnout, and unlike burnout, it isn’t resolved by rest. Whether it shortens lives is not something anyone has measured directly. That it corrodes mental and emotional health is not in serious dispute.
Systemic gender bias: Women physicians are paid less for the same work, promoted more slowly, and disproportionately steered into lower-status, higher-empathy specialties. Chronic stress has a well-established physiological cost in the general research literature. Whether these specific structural facts translate into that cost for this specific population is a hypothesis worth testing, not yet a settled finding.
The culture of silent suffering: Medicine still treats the disclosure of struggle as a liability, sometimes literally, through state licensing boards that ask intrusive mental health questions no other profession requires. So physicians, and women physicians especially, learn to suffer where no one can see it. By the time anyone notices, it is often too late to intervene.
None of these factors is new information to anyone who has worked inside a hospital. What is new is the mortality data itself, and the honest answer to why it looks the way it does is: We don’t fully know yet. That is not a reason to look away from it. It’s the reason more research needs to happen, and it’s why the factors above deserve real investigation rather than a shrug.
A decade in, and a controlled experiment I didn’t mean to run
I have spent the last ten years building physician wellness infrastructure specifically for women in medicine: programs, coaching frameworks, and most recently OnCall for You, an anonymous, store-nothing platform built on a validated framework for identifying and addressing physician-specific stress patterns, paired with a coaching engine trained on two decades of clinical and coaching experience with physicians, and a full burnout curriculum spanning nutrition, exercise, and community.
The results are not theoretical. Hospital-based wellness apps typically see about 5 percent physician engagement, the number institutions themselves report, and our program has sustained roughly 50 percent engagement. That’s ten times the industry standard, in the population these institutions say they are most worried about losing.
Over the past year, I offered this program to fifty hospitals and health systems, a mix of academic medical centers and community institutions, some approached at no cost at all, specifically to remove the one variable everyone assumes is the obstacle. One institution said yes.
I want to be precise about what that is. It is not an anecdote. Forty-nine institutions, told about a proven intervention addressing the exact crisis their own wellness statements name, with the cost barrier deliberately removed, declining to act, is a real-world exposure study in what actually drives institutional behavior around physician wellness.
One conversation stays with me. A three-way call with the wellness office of a well-regarded institution. Near the end, the man on the call told me, kindly, that I seemed really nice, really great, but he didn’t think this would be better than what they already had in place. I asked what results their current program was producing. There was a pause. We’d really have to look into that, he said. Then silence.
That silence is the finding. An institution that could not tell me its own engagement numbers rejected a program that could tell them, on the spot, exactly what its numbers were, because the alternative would have required admitting that what they already had wasn’t working and doing something about it.
What forty-nine rejections actually measure
When an institution says it cares about physician wellness and then declines a free, evidenced intervention rather than sit with the discomfort of comparing it to a program with no measurable results, “caring” is not the operating variable. Liability avoidance, optics management, and the bureaucratic incentive to avoid admitting a prior investment failed are doing the real work here. A hospital that adopts an outside solution is implicitly conceding that its internal one didn’t work, and that concession is what most institutions are actually protecting against, more than the well-being of the physicians they employ.
This is not a claim that individual wellness officers don’t care. Many of them do, visibly, and are as trapped by this dynamic as the physicians they’re trying to help. The indictment is structural, not personal: an institutional immune system that treats “We don’t know our own results” as a safer position than “We tried something new and it might have failed publicly.”
What comes next isn’t a request to institutions
I am done waiting for institutions to act on data they already have access to. Forty-nine rejections is data too, and it tells me where the leverage actually sits.
To the women physicians reading this: The mortality gap is real, it is measured, and it says nothing about your resilience. The system was not built to protect you, and you shouldn’t have to wait for it to volunteer that protection before you act on your own behalf. Seek support before you’re in crisis, while you still have the bandwidth to choose it deliberately. Ask for outcomes when someone offers you a wellness benefit, the way you’d ask a drug rep for the trial data.
To everyone else reading this who has hiring authority, a wellness budget line, or a seat on a policy committee: You do not need another task force. You need to ask your own wellness office the same question that produced silence on my call: What are our current results, specifically? Notice whether you get a number or a pause. If it’s a pause, you already have your answer about what to fix first.
The data showing that women physicians are dying sooner has been published in JAMA Internal Medicine and JAMA Psychiatry, peer-reviewed and citable. Why it’s happening is still an open question, and it deserves better-funded research than it’s gotten. But the data on why nothing gets done about it isn’t a mystery at all. It’s been sitting in my inbox for a year, with forty-nine unanswered or declined proposals, and I’m publishing that dataset here. Someone with a budget line and the nerve to compare their own numbers to a competitor’s can act on it starting Monday morning.