The call comes to a program director's phone at 6:14 in the morning, before the sun is fully up over the hospital parking structure.
A second-year resident has died by suicide overnight. That is the whole of what needs to be said about what happened, and it is the last time in this article that a specific death will be described at all. What matters now is not the how. What matters is what the program director does in the next six minutes, then the next six hours, then the next twelve months, and that she is, in this moment, entirely alone with the question.
She has run this program for four years. She has handled a resident's cancer diagnosis, a resident's divorce, two residents' parents dying, a resident who relapsed on the ward. None of it prepared her for this specific thing, because this specific thing has almost no institutional memory anywhere near her. She pulls up the hospital's employee assistance program number, which she has never called, and which she suspects the eleven other residents in the program have never called either. She drafts a message to send to the team. She cancels the morning didactic. She calls her chair, who is as unprepared as she is.
Somewhere in the country, right now, is another program director who did exactly this eighteen months ago. Who knows what worked and what made things worse. Who knows what the third week looks like, and the third month, and what to say to the co-resident who was the one who found out first and now cannot stop apologizing for something that was never hers to prevent. That person exists. The program director on the phone at 6:14 has no way to reach them, no way to know their name, and no fifteen minutes to spend looking before the residents start arriving for morning report.
She is about to lead a department through the hardest week of her career using judgment alone, because nothing in American medicine connects her to the only people who could actually help.
How common this actually is
Start with the scale, because the instinct is to treat this as rare.
Fifty-five percent of physicians report knowing a physician who has considered, attempted or died by suicide, according to a 2021 Physicians Foundation survey; a follow-up in 2023 put the figure at 51 percent. That is not a small subgroup with an unusual history. That is a majority of the profession.
Every residency program of meaningful size will, at some point, experience a trainee death. Research on suicide's ripple effect estimates that each suicide affects roughly 80 to 135 people, with about one in 30 of those people affected deeply enough to be considered directly exposed. A training program is exactly the kind of dense, interdependent unit where that exposure concentrates: co-residents who worked every third night together, an attending who supervised the resident's first codes, a program coordinator who knew the resident's family.
The risk is not evenly distributed. Female physicians have a suicide rate 53 to 76 percent higher than women in the general population, according to research published in JAMA Psychiatry in 2025 and the British Medical Journal in 2024. Anesthesiology, psychiatry, emergency medicine and surgery carry documented elevated risk among specialties. None of this is abstract risk modeling. It is a description of an event that has already happened, multiple times, in departments across the country this year, and will happen again.
What actually happens afterward, and what does not
Here is the honest accounting of institutional response, drawn from the closest thing available to a systematic look at it.
A 2022 critical integrative review concluded flatly that "no models exist to guide the delivery of postventions after a colleague's suicide." Not weak models. Not underused models. No models.
The same review reported results from a survey of anesthetists after a colleague's death: 22 felt supported, versus 179 who did not. That is not a program failing to reach a minority of an affected staff. That is a program reaching roughly one in nine.
A 2024 multi-method study conducted for the NHS, drawing on 51 staff interviews, documented "numerous gaps and barriers that prevented affected staff from accessing the postvention support they needed." It named two specific mechanisms behind those gaps: cultures "emphasizing worker invulnerability," and, separately, "skill deficiencies among support providers." The people whose job it nominally was to help did not, in many cases, know how.
Separate research on how supervisors respond to a related situation, a patient's suicide, found residents perceived their supervisors "emotionally distancing themselves from discussing the loss." If that is the pattern when the person who died was a patient, it should not surprise anyone that a colleague's death produces a comparable retreat, amplified by grief the supervisor is also carrying.
The actual workflow, reconstructed from what these studies describe, looks like this: death, an institutional communication, a memorial, a referral to the employee assistance program, and then silence. The EAP is generic mental health infrastructure not built for this specific loss. The memorial happens once. The referral is a phone number, not a relationship. And the leader improvises, because that is what is left to do.
Two different failures wearing one name
"Postvention" gets used as if it names a single thing. It names at least two, and conflating them is part of why nothing has been built.
The leader's failure is a peer-capacity failure. The program director, the managing partner, the nurse manager who has to hold a unit together has no verified colleague who has actually led a department through this specific kind of loss and can tell her, plainly, what to expect and what to do differently than the guidebook suggests.
The unit's failure is a support-capacity failure. The residents, the partners, the nurses who worked alongside the person who died are simultaneously the most affected and the least equipped to support each other, because they are all grieving the same loss at the same time, inside the same hierarchy, often with the person one rung up on that hierarchy also grieving.
A generic EAP addresses neither well. It was not built with the specificity either failure requires. It was built to handle an individual employee's distress, not a department's shared trauma with a leader who is simultaneously managing that trauma and her own.
The structural failure: why nobody owns this
Run through who could plausibly build cross-institution postvention capacity, and notice that each candidate has a reason not to.
Hospitals handle each death privately. There is no mechanism, and often no appetite, for one institution's postvention experience to travel to another institution's crisis. Legal caution around a death by suicide compounds the instinct toward silence.
ACGME sets wellbeing requirements without supplying capacity. Accreditation standards can require that a program have a wellness plan. They cannot manufacture the specific, lived experience of having already done this, and no accreditation body maintains a registry of program directors who have led a unit through a trainee's death.
AFSP and other foundations publish guides. The American Foundation for Suicide Prevention's postvention materials, and comparable guides from groups like Doctors' Health NSW, are genuinely useful and represent real institutional knowledge. They are documents. A document cannot sit with a program director on the phone at 6:20 in the morning and tell her whether to cancel the whole week of clinic or just the morning.
EAPs are generic by design, contracted to serve every kind of workplace distress, not specifically built around the peculiar intensity of a training program or medical group losing one of its own to suicide.
Nobody runs a roster. Program directors who have been through this exist, plausibly in the hundreds across the country given how many programs there are and how long postvention research has been documenting this problem. The newly bereaved director has no way to find them. That absence, more than any single missing guideline, is the actual structural gap.
Why this is not the same problem as crisis response
It is worth being precise about what is missing, because confusing it with suicide prevention infrastructure produces the wrong fix.
Crisis hotlines and immediate mental health resources matter enormously and are addressed later in this article. But postvention is a different clock. It runs for a year, not a night. The most acute early failure documented in the research, the 22-versus-179 gap and the "no models exist" finding, is not primarily about whether a phone number for crisis support exists. It is about whether the people responsible for leading a grieving department, and the people inside that department, have any structured, sustained, outside support as the anniversary approaches, as the next resident cohort arrives not knowing what happened, and as the co-resident who found out first becomes chief and has to lead new interns through a program with an unspoken history.
That is the 43-month problem hiding inside a single terrible week. The news cycle around a physician death ends. The department's grief does not, and nothing currently scheduled checks back in at month three, month six, or month twelve.
What would actually work
A matched leader peer within 24 hours. Not a hotline. A specific, verified colleague, ideally another program director or group leader who has led a department through a comparable loss, reachable directly, not through a general referral queue.
Outside peer supporters for the exposed unit within a week. People from outside the affected institution, so that support does not depend entirely on colleagues who are themselves grieving and exhausted, and so residents can speak without worrying it will reach their own program's leadership.
Scheduled follow-up at 1, 3, 6 and 12 months. The research is explicit that support tends to concentrate in the first days and evaporate afterward. A structured check-in schedule, built in advance rather than left to whoever remembers, directly answers the finding that institutional support fades exactly when delayed grief and anniversary effects intensify.
A survivor-written playbook, not a generic wellness pamphlet. Written by people who have actually led postvention, revised as new leaders go through it, so the next program director is not starting from nothing.
Explicit coordination with the institution, never replacement of it. Any cross-institution support must work alongside a hospital's own crisis response and communications, not around it. Families' wishes, legal considerations and the institution's own processes take priority, and outside supporters need to be built to defer to them.
No diagnosis, and an immediate, clear escalation pathway to crisis services. Peer supporters offer lived experience and presence, not clinical care, and the structure needs to make that boundary unmistakable rather than implied.
Confidentiality that survives contact with the institution. People need to be able to say what they actually feel, including anger at the institution or at colleagues, without it becoming part of a record anyone else can see.
What you can do now
If you lead a residency program, department or group
Find out, now, before you need it, whether your institution has ever done this before, and who handled it. Ask your chair, your GME office, your predecessor. If the honest answer is "we don't know," write that down as the actionable finding it is.
Identify one outside colleague you could call at 6 a.m. if this happened tomorrow. Not a hotline number. A person, by name, who leads a comparable program elsewhere and would take your call. If you cannot name one, that gap is fixable this month through your specialty society, your program-director listserv, or simply reaching out to a peer director you already know at another institution and asking the question directly.
Build the twelve-month calendar before you need it, not during. A one-page plan naming who checks in at each interval, and with whom, removes the single most consistent failure the research documents: support that is real in week one and absent by month three.
If you are a colleague, co-resident or team member
Say something rather than managing your own discomfort by staying silent. The research on supervisors "emotionally distancing" from these conversations describes a widespread, understandable, and harmful instinct. Naming that instinct in yourself is the first step to not acting on it.
Do not assume someone else is checking on the most exposed person. The co-resident or team member closest to the loss is frequently the person everyone else assumes is "being taken care of" by someone else, and is therefore checked on by no one.
Use the resources at the end of this article, and encourage colleagues to use them, without waiting for a formal referral.
If you fund or govern GME, wellbeing programs or physician mental health
Treat leader-to-leader peer capacity as infrastructure, not an afterthought to a wellness curriculum. The 22-versus-179 finding and the "no models exist" conclusion describe an infrastructure gap, not a knowledge gap; the fix is a functioning roster and a fast, reliable way to activate it, not another guide.
Fund the registry rather than another one-time toolkit. A roster of program directors and group leaders with lived postvention experience, willing to be called, maintained and kept current, is a durable asset a single guidance document cannot replace.
Frequently asked questions
What should a residency program do when a resident dies by suicide? There is no established, evidence-based model for this specific situation; a 2022 critical integrative review concluded plainly that "no models exist to guide the delivery of postventions after a colleague's suicide." In practice, programs typically move through institutional communication, a memorial, an employee assistance referral, and then have little sustained structure afterward, which is the gap this article addresses.
How common is physician suicide exposure? Very common. A 2021 Physicians Foundation survey found 55 percent of physicians know a physician who has considered, attempted or died by suicide, with a 2023 follow-up survey finding 51 percent. Research on suicide's broader impact estimates each death affects roughly 80 to 135 people.
Do physicians get adequate support after a colleague's suicide? The available evidence says no, for most. One survey cited in the 2022 integrative review found only 22 anaesthetists felt supported after a colleague's death, against 179 who did not. A 2024 NHS study of 51 staff documented barriers including workplace cultures that emphasize invulnerability and skill gaps among the people meant to provide support.
Are female physicians at higher risk of suicide? Yes. Research published in JAMA Psychiatry in 2025 and the British Medical Journal in 2024 found female physicians have a suicide rate 53 to 76 percent higher than women in the general population, a disparity not mirrored in male physicians compared with men generally.
What is postvention? Postvention refers to the structured support provided to people affected by a suicide after it occurs, aimed at reducing the risk of further harm among the exposed and helping a community or workplace process the loss. In medicine, the research base describing effective postvention specific to physicians and trainees is described by reviewers as essentially absent.
Where can a physician get immediate help after a colleague's death by suicide? The 988 Suicide and Crisis Lifeline is available at any time in the United States, and the Physician Support Line offers free, confidential telephone support staffed by volunteer psychiatrists specifically for physicians and medical students. Both are described in full at the end of this article.
The bottom line
A program director who has never done this before will, this year and every year, get a phone call before sunrise and have to lead a department through the worst week it has faced. The evidence says she is not unusual in facing this: 55 percent of physicians already know someone touched by suicide, and every training program of any size will eventually meet this moment directly.
The evidence also says she will likely do it with almost nothing built to help her. No models exist, a leading review concluded, in plain language, after looking for them. Fewer than one in nine anaesthetists in one survey felt supported after a colleague's death. Supervisors distance themselves from the conversation, not out of indifference, but because nobody trained them for it either.
Somewhere in the country is a person who has already been exactly where she is standing. Who knows what the third week looks like, what to say to the co-resident who found out first, what the twelfth month brings. That knowledge exists. It is scattered across hundreds of program directors and group leaders who have quietly carried their own department through this, with no roster, no registry, and no way for the next bereaved leader to find them.
This is not a problem that will be solved by a better pamphlet. It requires a person, reachable fast, and a structure that keeps checking back for a year. Building that is not a product decision so much as a trust decision, and it may be the deepest one this series describes.
At 6:14 in the morning, a program director is standing where hundreds of others have stood before her. She is about to lead a department through the hardest week of her career using judgment alone, because nothing in American medicine connects her to the only people who could actually help.
If you are in crisis, the 988 Suicide and Crisis Lifeline is available in the United States at any time. The Physician Support Line has offered free, confidential telephone support staffed by volunteer psychiatrists.
Part of a series on the missing professional infrastructure of healthcare. Previously: Compensation Opacity: Why Physicians Negotiate Blind
Evidence note: sources include the Physicians Foundation's 2021 and 2023 physician surveys on suicide exposure; Zimmermann et al., British Medical Journal, 2024, and a related JAMA Psychiatry 2025 analysis on female physician suicide risk; a 2022 critical integrative review on postvention for physicians (PMC9517643), including its citation of the anaesthetist support survey; a 2024 NIHR/NHS multi-method study of 51 staff (NBK618822); and 2025 BMC Medical Education research on supervisor response to patient suicide. No details of method or circumstance of any death are described in this article, in keeping with safe-messaging guidance on suicide reporting. The 80-to-135-people exposure estimate and figures on postvention support gaps come from the sources above and should be read as the best currently published estimates rather than precise counts; this article does not diagnose or offer clinical guidance, and nothing in it substitutes for immediate crisis support.
