You’re trusted to run the code at 3 AM and call the time of death yourself. You’re not trusted to say out loud that it wrecked you, without first calculating who might write it down.
That’s the week’s pattern, and it isn’t a wellness problem. Roughly 400,000 (or so) people die in U.S. hospitals every year, and residents are in the room for an enormous share of them. Training programs have built elaborate machinery for what happens to the chart after a death — the review, the conference, the corrective action — and almost nothing for what happens to the person who was holding the bag valve mask. Emotional processing has no billing code and no ACGME checkbox.
Medicine does have a name for you in that moment: the “second victim.” The patient is first, you’re second. It’s a phrase that sounds like it was workshopped by a committee, which it was, and it describes something that feels nothing like a term — lying awake replaying a code you ran correctly, a pediatric trauma sitting in your chest at dinner, guilt that sticks even when every decision was defensible. We wrote the piece on what actually happens after a patient dies on your service because nobody hands you a framework for this. It won’t make the case leave. It will tell you what the experience is called and why it’s this heavy in training specifically: you haven’t seen enough outcomes to know whether this one was exceptional or ordinary, and everyone around you looks fine — because everyone around you is also performing fine.
Then residency adds the layer that makes grief nearly impossible to do honestly. One resident described missing the documentation of a negative FAST and spending the aftermath terrified of being seen as untrustworthy — not mourning the patient, just running the professional math. Another worried a stroke-alert decision would surface in their file. That isn’t paranoia. It’s an accurate read of a system where adverse outcomes trigger reviews, documentation gaps become liability, and the boundary between a learning moment and a career problem is never drawn for you. You cannot grieve openly while quietly estimating what a sentence might cost you in fellowship season. So you don’t.
M&M doesn’t fix this, and it was never built to. M&M exists to improve systems — attendings in the room, decisions on the screen, the evaluative hum in the background. Useful for the system. Structurally incapable of doing grief work. Meanwhile the thing research actually supports is unglamorous and cheap: talk to someone who was there, within 24 to 48 hours, about how it felt rather than what should have been done differently. A call-room conversation. A group text. Nobody teaches you how to ask for one, which means the most effective intervention available is the one you have to invent yourself, at 2 AM, while tired.
The part worth knowing before it matters: there’s a difference between grief that dissipates and grief that sticks. Carrying a case for weeks is normal. The case becoming proof that you’re inadequate, intrusive images you can’t shake past a month, avoiding the unit where it happened — that’s a different category, and the section separating normal grief from depression, PTSD, and moral injury exists so you can tell which one you’re in without diagnosing yourself at 4 AM. Moral injury especially: the death that followed an insurance delay, the bad outcome that followed a short-staffed night. That damage isn’t about your competence. It’s about being conscripted into a system that sometimes harms people despite you.
Here’s the career part nobody says at orientation. Unprocessed clinical grief is one of the strongest predictors of early attrition and first-job dissatisfaction. Residents who never get an outlet carry the weight into attending life, where the scaffolding is thinner and nobody’s pretending to check on you. Which makes debriefing culture a hiring criterion. When you’re evaluating that first contract, ask how the group handles a death, whether debriefs happen, whether peer support exists or you’re expected to absorb it privately. Ask whether the therapy benefit has appointments outside protected didactic time, because access you can’t physically reach is a brochure. The rest of the Residency Survival Guide archive is built around exactly this kind of question: the stuff that shapes whether you’ll still want the job in five years.
We should be honest about our lane. PhysEmp is a job board. We’re not therapists, we’re not your program’s counseling service, and we can’t give you a sanctioned place to fall apart. What we have is thirty-plus years of watching this system chew through good physicians, and a willingness to write it down plainly when nobody else will. The system won’t reform itself, and we can’t reform it either. We can hand you the vocabulary, the warning signs, and the questions to ask an employer before you sign — a starting point, not a rescue.
Programs expect death to build your resilience while maintaining evaluative and legal structures that punish showing any, and the only survivable response inside that design is silence — which is precisely the thing that follows people out of training and into the decision to leave. Grief has quietly become a workforce issue.
One thing an article can’t do is answer back at 3 AM. That’s why we’re building DocCommons — a place where residents can ask actual physicians, including retired ones who have carried their own version of this for forty years, the questions you can’t ask your program director. It isn’t live yet; right now it’s a waitlist. Free, quiet, and built because the gap was too obvious to keep ignoring.
P.S. PhysEmp is an AI-powered physician job board, which mostly means we do the tedious matching so you can spend your limited waking hours on something else. When you start looking at first attending jobs, bring the debriefing question with you. It tells you more than the signing bonus does.
Sources for this issue were gathered this week from social media sites like Reddit to find out what residents are actually talking about. AI helps us spot the patterns. A human decides what’s worth your already-overcommitted attention. [How we actually make this →]