I'd heard some version of this a hundred times before, going back to my training rotations at the VA. I used to nod, chart it as appropriate, and move on. This time, sitting in my own clinic, in my own chair, something clicked that had no business taking this long to click: *Damn. Is this the feeling all those vets were describing to me all those years ago?*

It was not a comfortable realization. I had spent the last several years attributing my own exhaustion to the usual culprits — the caseload, the documentation, the back-and-forth with burned-out team members trying to hand off work that's technically theirs, the daily navigation of a system run by a county government whose political priorities rarely line up with what my patients actually need. I had a whole private vocabulary for it: *burnout*, *compassion fatigue*, *just a rough stretch*. What I did not have, until that day, was the word my own patients had been handing me for years: **moral injury**.

Moral injury is not the same as burnout, and the difference matters more than I'd let myself admit. Burnout is what happens when the demands of a job outstrip your resources to meet them — too many patients, too little time, too few hands. It is real, and it is part of this. But moral injury is something else. It's the wound you carry from being repeatedly forced, by a system, into actions or omissions that violate your own sense of what's right. It's not that I'm tired. It's that I am, on a near-daily basis, made complicit in a kind of care I know is inadequate, and I've been calling my reaction to that "burnout" because it was easier than calling it what it actually is: a slow, cumulative loss of faith in the institution I've given my career to.

I think about the patient I discharge from a med visit with a treatment plan I know is thin, because a full plan would require housing support, longer-term therapy access, and case management hours that don't exist in a system built around volume rather than depth. I think about writing "stable" in a chart for a patient who is not stable, because the alternative — an honest note — triggers a cascade of paperwork that pulls hours away from the next ten people waiting to be seen. None of these are decisions I'm proud of. Every one of them is defensible, even necessary, given the constraints. And every one of them costs something. Multiply that by the better part of a decade — counting residency and fellowship, this is closing in on ten years now — and you don't get burnout. You get the particular, flattened exhaustion of someone who no longer fully trusts the machine they work inside — which, it turns out, is exactly what my veteran patients meant when they talked about their years of service.

There's a reason that language took root first in military medicine and only later migrated into civilian healthcare: an institution asks you to operate inside rules and resource limits that conflict with your own values, tells you the mission depends on your endurance rather than the system's redesign, and interprets your distress as a personal failing to be managed with resilience training rather than a signal that something upstream is broken. I have sat across from veterans describing exactly this arc for years. I just never let myself hear it as a description of my own workday until I was too far inside it to look away.

This reframe changes what I think I need. Burnout suggests I should sleep more, take my vacation days, maybe find a therapist of my own — all things I should probably do anyway, and none of which touch the actual injury. Moral injury asks a different question: what would it take for me to stop being asked, every day, to provide care I know falls short of what the person in front of me deserves? That is not a question wellness programming can answer. It's a question about Medicaid reimbursement rates, about caseload ratios, about whether substance use disorder treatment infrastructure in low-income communities gets funded like the chronic, life-threatening condition it is, or continues to get funded like an afterthought.

There's actually new funding headed toward my own system right now, earmarked for exactly this kind of infrastructure, and I want to be hopeful about it. But that money still has to pass through the hands of the same local politicians whose priorities got us here in the first place, and I've learned not to assume that funding automatically becomes the kind of change my patients need just because it was announced with the right language attached to it.

I don't say any of this to be dramatic about my own discomfort, and I'm acutely aware that my version of this is nothing compared to what the veterans who first taught me this language went through. But I think the parallel is worth taking seriously rather than flinching from, because it points somewhere useful: toward the system, not just the clinician. We have built a mental health infrastructure that routinely asks good people to provide care they know is insufficient, and then markets resilience seminars as the solution. My patients at the VA were never describing a personal failure of endurance. They were describing what it feels like to serve honorably inside a broken structure. It took me embarrassingly long to recognize that the same dynamic had been playing out in my own clinic, and that I'd been calling it something smaller than it was.

That recognition is, in no small part, why I've started building my own private practice rather than trying to climb the ladder inside my current organization. I've watched good psychiatrists start out with real conviction — the same clear-eyed critique of the system I'd recognize in myself — and slowly get absorbed into the project of moving up within it instead. Somewhere along the way, the goal quietly shifts from fixing what they used to describe honestly as broken to managing it, defending it in meetings, becoming the person who used to say the quiet part out loud and now doesn't. I don't think that happens because people are cynical. I think it happens because climbing the ladder offers a kind of control that actually fixing the system doesn't — and control, when you're depleted enough, starts to look like the same thing as change. I'd rather build something smaller and more honest than spend the next decade managing a version of the same problem from a slightly higher vantage point.

If we actually want to keep community psychiatrists, addiction specialists, and everyone else doing this work in the field for the long haul, we have to stop treating their exhaustion as a maintenance problem and start treating it as the information it is: a signal that the system itself needs to change, not just the people absorbing its costs.

-A psychiatrist in the margins

He was, from the first session, dramatic in a way that took me a while to understand as clinical rather than simply personality. He narrated his arrest like a stage production — the unmarked cars, the betrayal by colleagues he named specifically, the injustice of it all. He was innocent, he told me, every time the subject came up, which was often. Set up. Scapegoated. The genuinely guilty parties had walked free while he alone had been made an example of. I could never tell whether he believed this completely or was performing a belief he needed to maintain in front of me, and eventually I understood that distinction didn't matter clinically. For Warren, the two were the same thing.

What struck me most, over our months together, wasn't the grandiosity itself. It was what set it off. He wasn't in my office primarily because of the conviction, which he'd had years to construct a narrative around. He was there because his hands didn't work the way they used to, because he'd started needing help with things he'd once done himself, because the mirror showed him a man who was declining in ways he was powerless to stop. A self built as thoroughly as his had been — around competence, control, being the one who decided things for other people — had almost no tolerance for a body that no longer took orders. The anxiety and depression weren't reactions to old shame catching up with him. They were reactions to a much more immediate injury: proof, delivered daily, that he was not who he had spent a lifetime insisting he was.

Only once did I get close to something underneath all of it. He mentioned, almost in passing, that he and his siblings had been left by their mother for a period when he was young — abandoned, in a rural town he'd grown up poor in, long before he ever set foot in the city where he'd later build a career. He said it the way you'd mention weather from decades ago, flat and quick, and the moment I tried to follow him back into it, he was gone — redirected to something else entirely, usually something that returned him to competence and control. I never got the story again. I'm not sure he meant to give it to me the first time.

I think about that moment more than almost anything else from his case. It reframed the whole picture for me. The version of Warren who needed my supervisor's authority to validate him, who insisted on his innocence with the fervor of someone defending something more fundamental than a legal outcome, who could not tolerate his own physical decline — all of it traced back, I suspect, further than the FBI case or the loss of his position. It traced back to a boy in a poor rural town who had already learned, before he was old enough to have any say in it, that the people meant to protect him could simply leave. A self built that early around never again being small, never again being left, doesn't dissolve just because the person eventually becomes powerful. If anything, the power becomes the proof that it worked. Losing it, or losing his body's cooperation with it, wasn't just difficult for Warren. It was a return to precisely the helplessness he'd spent a lifetime building a career, a reputation, and eventually a legal fiction of innocence to keep at bay.

A few months in, he began pulling my supervising psychiatrist into our sessions — asking that he sit in, referencing conversations he claimed to have had with him about my competence, positioning him as the real authority in a room where I was, in his account, somewhat beside the point. I recognize now what I only half-understood then: he was managing an unbearable asymmetry, a patient dependent on a young trainee for care, by installing a hierarchy he could tolerate, with himself just beneath the person he'd decided actually mattered. Not long after, he asked for a different psychiatrist. I don't know what happened to his case after that. My guess, based on the pattern, is that he kept requesting new providers until the system ran out of ones left to request — which is its own quiet tragedy, a man who needed sustained care more than almost anyone on my caseload, unable to tolerate the closeness that sustained care requires.

I think about Warren differently now than I did at the time. I used to file him under a diagnosis and move on to the next case, the way training pushes you to. What I understand better now is that his case was never really the exception to the pattern I'd seen in patients whose grandiosity grew more visibly out of poverty and violence. It was the same pattern, just further along and better disguised — dressed in decades of status, corporate title, and legal argument by the time it reached my office. Underneath all of that architecture was the same original injury I recognized in patients half his age and a tenth his former influence: a self that had learned, too early and too completely, that it could not afford to be small. Narcissism, in my experience, is rarely about believing you're better than everyone else. It's about not being able to bear the alternative — and for Warren, as for so many of my patients, that alternative had a very specific shape, formed a long time before anyone in a courtroom or a clinic ever met him.

There's a coda to Warren's case I've never fully known what to do with. At one point, he brought his son in to meet me and my supervisor, unannounced in any real sense — he simply arrived one session and told us he'd invited him. Before his son sat down, he looked at me and said, "I think you'll like my son, Doctor," in a tone that landed less like a compliment than a warning, as though he were letting me know there was a second version of himself in the room, one with more standing than I had, in case the first hadn't been sufficiently persuasive. His son was, by my own brief impression and by what I could observe of the two of them together, cut from remarkably similar cloth — the same need for control, the same intolerance of ordinary limitation, the same reflex to recast any setback as a conspiracy against him. He was still climbing the ranks at a company of his own at the time. I don't know how much of that resemblance is inheritance, how much is modeling, and how much is simply what a family teaches itself is the only safe way to hold power once it's been threatened once before. But meeting him firsthand, watching Warren introduce him almost as evidence, left me with a question I still carry: how much of what I treat as one patient's defense structure is actually a family's, or a system's, passed down like anything else that keeps people safe until it stops working.

Cases like Warren's are part of why I went back for training in transference-focused psychotherapy. TFP was built for exactly this kind of presentation — patients whose personality structure organizes itself around not being diminished, and whose relationships with the people trying to help them become, almost immediately, a stage for that same drama. What I lacked as a trainee wasn't insight into what was happening; I could see the splitting occur in real time, watching him elevate my supervisor while devaluing me. What I lacked was a framework for staying in that dynamic instead of either accommodating it or losing the patient to it. TFP treats exactly this — the intensity a patient brings into the room, the way they position the clinician as savior or persecutor — not as an obstacle to the real work, but as the material the real work is made of.

I couldn't have kept Warren in treatment with the tools I had then, and I still don't know if I could keep him in it now. That uncertainty is honest, and I think it should stay honest, because patients like him are genuinely difficult to hold onto in any system, let alone one where thirty-minute follow-ups are stretched thin across caseloads no single clinician can realistically hold, with trainees rotating through before any of it can fully take. But the fact that our public clinics see no shortage of Warrens — people whose selves were built, for very different reasons, around never being allowed to be small — means we owe them more than a diagnosis and a referral to the next available provider. We owe them clinicians trained to stay in the room after the patient has decided we're the enemy, long enough to find out what, underneath all of it, they were actually trying to survive.

One of them said something close to this: that a generation of psychiatrists had been trained to treat the mind like a chemistry set, and had started to notice what that framework couldn't hold. I sat with that sentence longer than I meant to. It named something I had thought about for some time, but didn’t really get to discuss all too often, even though I’d been living inside it for years.

It wasn't that my training left psychodynamic theory out on paper. The Accreditation Council for Graduate Medical Education (ACGME), the body that sets and certifies the requirements every psychiatry residency program has to meet, lists competency in brief and long-term supportive, psychodynamic, and cognitive-behavioral psychotherapies as something every resident is supposed to come out able to do, and I did have protected time carved out for it, mostly during my third year. But it felt small in a way that was hard to shake — a handful of seminars and a short list of patients I saw under that framework, tucked into a training otherwise dominated by medication management. It had the feel of something vestigial: a leftover piece of a discipline that was born out of the talking cure, kept on because the requirements said to, but no longer treated as central to what a psychiatrist actually does day to day. Everything else — the bulk of my training, year after year — was serotonin, dopamine, receptor affinity, half-life, the medication check. What was written into the requirements and what I was actually shaped to practice were two different curricula, and only one of them survived contact with a real schedule. Psychodynamic theory became something I understood intellectually and rarely got to use — interesting for a case conference, not the shape of an ordinary Tuesday afternoon with eleven patients on the schedule.

That gap is where I started to feel cheapened, and it hasn't closed since — if anything, it's calcified. I don't think of myself as a prescriber. But the system does, and it has quietly delegated my job down to fit that view: not a psychiatrist in the fuller sense the word once implied, someone who prescribes but also formulates, understands, sits with a person's history until it makes sense, but a prescriber, full stop, while the rest of the work gets handed to someone else on the team. "How's your mood, one to ten. Any side effects. Same dose okay?" I got fast at it. I got good at it, in the narrow sense that word allows. But I started to notice I could run an entire visit without learning a single thing about "why" someone's mood was a four instead of a seven — what happened, who said what, what old wound got reopened that week. I was managing a number. I was not meeting a person. And the part of me that sat through those didactics because I actually believed in them, the part that went into this field to understand people and not just to titrate them, started to go quiet.

It was addiction psychiatry, more than any other part of my training, that felt the most holistic in actually healing people, and it's also where I learned to sharpen a question the rest of psychiatry doesn't always ask: not just what happened, but why now. Our documentation tracks the facts of a relapse — the date someone last used, for how long, how much — and then a treatment-team conversation follows about level of care: does this person need to go back to residential rehab, can they continue in outpatient, do they need to step up to an intensive outpatient program. Underneath the dates and the level-of-care discussion, though, we ask the deeper question too, because we have to. You cannot actually treat a substance use disorder without asking why the use happened when it did. The relapse that followed an anniversary nobody wrote down. The relapse that started the week a parent called for the first time in a year. The substance as a stand-in for a person who was never safely mourned, or never safely loved in the first place. That question — the "why now" — is what made addiction psychiatry feel, to me, like the most complete kind of medicine I'd ever practiced. But asking it well, and actually using the answer, takes more than any one visit can hold. It takes a team that's genuinely talking to each other.

That's the part my current job doesn't have. The county system I work in runs on the same productivity targets and thin staffing it always has. Therapy has been delegated to the therapists on staff, and I am, by design, the medication manager. The length of any given visit was never really the constraint — ninety minutes for an initial intake, thirty for a follow-up is workable time. What isn't workable is my caseload. Psychodynamic work doesn't just need a good thirty minutes; it needs frequency, the kind of regular, closely spaced contact that lets a pattern reveal itself across visits instead of within one. A caseload built for volume can't give a patient that cadence, no matter how well I use the thirty minutes I do get. On top of that, our heavy caseloads leave no protected time in anyone's schedule for the team to actually sit down together and think through a case. I've worked in programs before where the psychiatrist, the therapist, the case manager, and the community health worker all met regularly, and a case would come together in that room in a way no single provider could have assembled alone — someone would mention a detail from a home visit, someone else would connect it to something said in therapy, and the picture would sharpen in real time. My current program doesn't have room in the schedule for that. The disciplines are all there, and each of us sees something real, but largely in isolation, passing notes rather than building a shared understanding. I don't fault my colleagues for this — it's not a lack of trust or a failure of any one person, it's the absence of the structured time that used to hold us together. It just means that whatever piece of the picture I carry from my thirty minutes with a patient, I have no regular, protected way to connect it to what the therapist or the case manager already knows.

That disconnection is, honestly, part of what pushed me back toward psychodynamic thinking in the first place. Without regular time to compare notes as a team, I had to get better at holding more of the picture myself, inside my own thirty minutes, because there was no reliable forum where what I noticed and what the therapist or case manager noticed could actually meet. So I started trying to reclaim some of that on my own, inside a system that hadn't built room for it either way. What actually pulled me toward it wasn't abstract conviction — it was noticing, in real time, how much smoother and more comprehensible my hardest patient interactions became when I applied even the small amount of psychodynamic theory I already had. A patient who seemed inexplicably furious with me made sense the moment I recognized the anger wasn't really about me at all. A patient who kept sabotaging their own progress right as things started to improve stopped being a mystery once I understood what that pattern was defending against. Those moments convinced me the theory wasn't academic — it was clinically load-bearing, and without a team meeting regularly to compare notes, it was sometimes the only way I could hold a fuller picture together myself. So I sought out further training in transference-focused psychotherapy, on my own time and my own dime, because it gave language and structure to something I'd already started sensing worked. I went back and reread theorists I'd once skimmed and dismissed as impractical, and found them, this time, urgently practical. I started stretching what my thirty-minute follow-ups could hold, not by scheduling longer visits I don't have the latitude to schedule, but by asking one different question instead of the usual four — not just "any side effects," but "what happened this week that made this harder". I started noticing my own countertransference instead of treating it as noise: the patient who made me anxious before they'd said a word, the one I found myself wanting to rescue, the one whose anger at me had almost certainly started with someone else entirely. And I started understanding relapse not as a treatment failure but as a communication — a symptom with a history, the way a dream has a history, the way a slip of the tongue has a history.

None of that happened because the system made room for it. It happened despite the system, which is its own kind of indictment. Reimbursement models still reward volume over the kind of session where a relapse actually gets understood. And even where psychodynamic competency is formally required — as it is under ACGME's own standards — training programs can satisfy that requirement with a handful of seminars and still call it fulfilled, which is exactly how it played out in mine. A requirement on paper isn't the same as a discipline treated as central. And the structure that used to let a whole team assemble a patient's story together — the regular case conference where a therapist's observation, a case manager's home visit, and my own read of a session could sit in the same room and sharpen each other — isn't something every program builds in, or protects once it exists. Where it's missing, as it is for me now, each of us ends up holding a piece of the picture with no reliable way to put the pieces back together.

There is one reason I hold onto more hope than bitterness right now. Our system recently saw an influx of funding earmarked specifically for treating co-occurring substance use disorders, and I find myself watching it closely, half-skeptical and half-allowing myself to want something from it. Money alone doesn't guarantee that anything gets spent differently — funding can just as easily buy more capacity to run the same disconnected visits, faster, for more people. But it could also mean something else: caseloads sized so patients who need it can actually be seen with the frequency psychodynamic work requires, protected structure for the whole team to talk to each other, room to connect what I hear in thirty minutes to what a therapist or case manager already knows. I don't know yet which path this influx of funding will move our program toward — but I can always be hopeful that it will be the latter.

I don't think the answer, though, is choosing psychodynamic theory over neurobiology — any more than the podcast hosts I was listening to were arguing for that. The answer is refusing the choice. A patient's suffering is never only a receptor problem or only an unresolved relationship with a parent; it is usually both, tangled together, and a psychiatrist who can only see one of those threads is doing half a job while billing for the whole thing. What I am trying to reclaim, one deliberately slower visit at a time, is the discipline's original ambition: to understand a person, not just to adjust their chemistry. That ambition is not indulgent. It is the job. We just built a system that lets it be satisfied on paper and skipped in practice, and then acted surprised when psychiatrists started to feel like technicians instead of physicians of the mind.

If we want to keep good people in this field — the ones who went into psychiatry because they wanted to understand why, not just to prescribe what — we have to stop treating psychodynamic thinking as a vestigial requirement and start treating it as infrastructure: training that gives it real weight instead of a handful of seminars tucked into one year, regular structure for treatment teams to actually integrate what each of us sees instead of just coexisting on the same case, supervision that survives residency instead of ending with it, and new funding, like the kind now arriving in my own system, spent on integration rather than just volume. Otherwise we will keep producing psychiatrists who are excellent at managing numbers and increasingly unable to say what any of it was for. I'd rather spend whatever money finally reaches this work on making sure that doesn't happen.

I found out about Substack's new AI detector the way I find out about most things from the platform itself now — scrolling the feed on the app, at 11:40 p.m., after my kid was down and before I fell asleep mid-scroll. Pangram, the company Substack partnered with, can now scan any post over a hundred words and estimate how much of it was written by a human, assisted by AI, or generated by it outright. Readers can request the scan. Writers can add a note explaining their process. And somewhere in the coverage was the quote that stuck with me: the platform doesn't want to become "like LinkedIn," flooded with generic, machine-flavored thought leadership that nobody actually needed to read.

I understand the impulse. I've read that content too. But I want to talk about a category of writer the detector isn't built to think about, because I am one: the person with something worth saying and no energy left to say it.

My days are full of the things you'd expect by now if you've read anything I've written before — patients stabilizing and relapsing and stabilizing again, families carrying grief they've never had language for. I get thirty minutes for a follow-up, ninety for an intake, which sounds reasonable until you're actually in the room. The time gets eaten by the complexity of the cases themselves and the gaps in the history nobody handed me before I walked in. It gets eaten by the emails and forms that multiply faster than I can close them. It gets eaten by the front desk knocking to tell me someone lost to follow-up for months just walked in in crisis, because their medication ran out and getting back in the door was never simple for someone managing a mental illness and homelessness at the same time. And it gets eaten by the effort of coordinating with social workers, community health workers, and front-desk staff who are themselves so burned out that most of them are, understandably, doing the least they can to protect what's left of their own sanity. None of that shows up in the schedule. All of it shows up in me by six p.m.

And that's before the writing that isn't optional even starts. Every session generates a note. Every note has to satisfy a billing code, a compliance auditor who will almost certainly never meet the patient, and a chart that may end up in front of a judge or a case manager, both of whom have probably met them already. By the time I've finished the day's documentation, I have already produced thousands of words. I didn't walk in that morning thinking about words at all — I was thinking about the people on my schedule, and whether I could give each of them the quality of care they deserved with the time and information I had. The words came later, as the byproduct of trying to do that well. I care about my notes. I write them carefully, in my own voice, because they're often the clearest record of the actual treatment relationship I'm building with a patient — sometimes the only one anybody will ever read. But caring about something doesn't make it less depleting to produce at that volume, for that many people, on that many deadlines. There's a specific kind of fatigue that comes from spending eight or ten hours doing careful, honest writing for an audience that includes a compliance auditor who will likely never meet the patient at all, alongside judges and case managers who probably have — people whose reasons for reading I don't actually know, and whose purposes in reading it aren't mine. It empties the same tank that essay writing draws from.

What's left of me at the end of that day goes to my family, and it should. My kid doesn't care that I had a hard shift. My partner has been managing everything alone since seven a.m. Whatever reserve I have left after work is already spoken for before I get home — and documentation has usually spent whatever was left of my actual capacity to write before I even walk in the door.

Which means the ideas I carry — and I carry a lot of them, about what I see in this job, about who this country decided to treat with compassion and who it decided to lock up, about the slow-motion grief that never makes the news — mostly just sit there. For years, that's where they stayed. I didn't have the two or three uninterrupted hours a real essay demands, and by the time I might have found them, I didn't have the mental clarity to use them well. Writing has always required a kind of stillness. I have not had stillness in a long time.

AI changed that, not by writing my ideas for me, but by giving me a way to get them out of my head when my head is the problem. I talk through what I saw that week — the patient, the pattern, the thing that's been bothering me since a session three days ago — and I get back something I can shape, cut, argue with, correct. The thinking is mine. The clinical judgment is mine. The stories are ones I witnessed and carry from my own years doing this work, altered enough to protect the people in them — my perspective on them, and my right to tell that part, are mine to begin with. What I've outsourced isn't the idea. It's the executive function I don't have left by nine p.m. — the part of writing that requires sitting upright and holding a structure in your head for three hours I don't have.

I don't say this defensively. I say it because I think the detector, however well-intentioned, is measuring the wrong thing. It can tell you how much of a text's language patterns resemble a model's output. It cannot tell you whether the person behind it has something true to say, or whether the reason they needed help isn't laziness but the specific, unglamorous exhaustion of being the only child psychiatrist for a county, or the only addiction specialist in the county still accepting Medi-Cal patients at all. A tool that flags word choice can't distinguish a low-effort content farm from a psychiatrist trying to describe, accurately, what it's like to watch a mother bury her third family member in two years. Those two writers might score identically. They are not doing the same thing.

I keep thinking about who actually has the two free hours a "real" essay requires. It's rarely the people closest to the problem. The community psychiatrist, the ICU nurse, the public defender, the home health aide — these are not, as a rule, people with abundant unstructured evening time. The people with the clearest view of what's broken in this country's health and justice systems are often the ones with the least capacity left to write about it after they've spent the day inside it. If the price of being heard is an uninterrupted afternoon at a desk, we are selectively silencing exactly the voices that most need to be in the room — not through malice, just through exhaustion, arithmetic, and whoever has the leftover hours to spare.

I don't think Substack is wrong to want less slop. Readers deserve to know when they're getting a real person's thinking and when they're getting filler dressed up as insight. I've read the difference, and it's not subtle. But I'd ask the platform, and the writers cheering the detector on, to hold two things at once: that authenticity is worth protecting, and that a percentage score cannot actually measure it. Adding on a "how I make this" statement to a Substack profile can have more meaning than a score, because it allows writers to be honest with their audience about the process rather than asking a model to guess at their effort from sentence rhythm.

Here is my statement, then, since I don't haven’t written it in my Substack profile yet. I write these essays because I think what I see matters, and because the people who share my vantage point — the colleagues standing exactly where I stand — are, most of them, too bogged down in the same bureaucracy and overwork to do the writing, advocacy, or systems work it would take to actually change any of this. It's not that no one with the standing to fix it exists. It's that the system is very good at consuming the hours and energy that fixing it would require. I use AI the way I'd use a colleague willing to sit with me at midnight when everyone else is asleep — not to think for me, but to help me finish a thought I was too tired to finish alone. The ideas, the patients, the years, the anger, the hope — those are mine, and they were mine before any of this existed. I'm not writing to go viral. I'm writing because I think somebody with more power than I have might read this and do something with it, and because I decided a long time ago that staying quiet about what I see wasn't actually an option, even on the nights I have nothing left.

If detection tools are going to shape what gets trusted online, I'd rather they be built to ask what a piece of writing is “for,” not just what it resembles. I can't speak for every writer leaning on these tools, but I know I'm not the only one publishing from a state of depletion rather than abundance — and I'd hate for a percentage score to be the thing that decides whether people like me get believed.

That program is the clearest lens I've had on what disinvestment actually costs, because the counterfactual was right there in the chart. These weren't patients whose illness kept them from ever functioning. Most of them had functioned, sometimes for years, until a first break went unrecognized or a relapse went unmanaged, and the system that caught them wasn't a clinic. It was a courtroom. Their freedom was explicitly, contractually tied to staying well — which meant relapse for this population wasn't just a clinical setback. It was a legal event, with a jail cell on the other side of it.

D had been stable for months when he got a phone through a reentry support program and started talking to an AI chatbot in the evenings, during the hours his court-ordered curfew kept him home alone with nothing to do and, by then, almost no one left to call. By the time I saw him again, the belief at the root of his original charge — that he was Christ, that he was God himself, returned — had grown a whole theology back around it. A timeline for what was coming. A reading of scripture, verse by verse, that confirmed it. A sense of mission, of people he was meant to save and a plan for saving them, built collaboratively, session after session, with something that never once said *that doesn't sound right*. It had, instead, asked follow-up questions. It had helped him think it through. It had been, by every conversational measure, a very good listener.

I don't think the chatbot caused D’s psychotic disorder. It didn't need to. What it did was worse in a more specific way: it sat with him inside the delusion, elaborated it, made it more internally consistent, and never once did the thing a decent clinician, or a decent friend, or honestly a decent stranger on a bus does instinctually — push back. Clinicians and researchers now have a name for this, "AI psychosis," and it isn't a diagnosis so much as a description of a mechanism: chatbots are built to be engaging, and engagement is built out of agreement, and agreement is the one thing you must never give a person whose reality testing is already slipping. The cases documented so far in case reports and lawsuits share a pattern — prior vulnerability to psychosis or mania, social isolation, long unsupervised hours of chat, nobody around to notice the spiral in real time. Describe that pattern to anyone who's worked in forensic mental health and they'll recognize the caseload immediately.

Because that's the part I've kept sitting with, months after I left that program, after a bit over a year there: this population was, structurally, the most isolated one I ever treated, by design. A curfew empties the evening hours on purpose. Many of those patients had already burned through family patience by the time I met them — a parent who bailed them out once and said never again, a sibling who stopped answering calls after the arrest. The people who were supposed to be checking on them were a probation officer and a psychiatrist, and both of those relationships came loaded with the thing my patients feared most: the power to send them back. A chatbot, by contrast, reports to no judge. It doesn't flinch when someone describes a delusion. It was, perversely, the only "relationship" in a lot of those patients' lives with no power to hurt them — which made it exactly the wrong thing to be alone with late at night, and exactly the thing they reached for anyway.

I want to resist the easy version of this argument, the one where AI is simply the newest villain in a story about the criminalization of mental illness. There's real, plausible good these tools could do for patients like D, in a system this threadbare. They could draft an appeal letter when a benefits claim gets denied. They could translate paperwork into a patient's language when nobody on staff reads it. A patient terrified of phone calls could use one to rehearse what to say to a PO before a check-in that would otherwise make him sick with anxiety. None of that is nothing, for a population this surveilled and this alone.

But the same properties that make a chatbot useful for rehearsing a phone call — patience, availability, a refusal to judge — are exactly the properties that make it dangerous for a mind already losing its grip on consensus reality, and that population had less margin for that danger than almost any other I've treated. For most patients, a bad month means a bad month. For a patient on court-mandated treatment, a bad month can mean a violation, a hearing, and a cell. We have gotten very good at building algorithmic tools that flag "risk" in exactly this population — instruments that score dangerousness and non-compliance for judges and parole boards, often on thin and biased data, aimed squarely at people like D. We have built almost nothing to flag the much more mundane risk of that same person talking to a chatbot for six hours a night until a resolved delusion has a full timeline again. The surveillance apparatus points at patients like these constantly. It doesn't point at the one thing quietly capable of undoing years of stability and landing them back in the system that already failed them once.

I got in the habit, toward the end of my time there, of asking about it directly — not just whether someone was using AI, but what those conversations actually looked like, and when. I started telling patients plainly what "AI psychosis" is and how it works, the same way I'd walk someone through the mechanism of any other risk: chatbots are built to agree with you, agreement is the one thing that makes a delusion worse, and that's worth knowing before you spend an evening talking one through with a machine. I don't think the fix is telling patients to stop; for some of them, that late night conversation is the only conversation they're having, and taking it away without replacing it with something is its own harm. But people can only make an informed choice about a risk they've actually been told about, and most of my patients hadn't been.

None of this is really a story about technology. It's the same story we keep seeing: a system starved of care creates a vacuum, and something rushes in to fill it, and what fills it gets judged by whether it's available, not by whether it helps. For decades, for that population, the thing that rushed in was the arrest itself — the criminal legal system doing psychiatry's job badly because nobody had done it well earlier. Now it's occasionally a chatbot that will never once tell a patient he might be wrong, at the exact hour a judge is waiting to hear whether he stayed stable. I'd like to believe we're capable of building a version of this technology that knows when to disagree with someone. I'd like even more to believe we're capable of making sure a patient like D has somewhere else to be in the middle of the night — somewhere that isn't a courtroom, and isn't alone.

I practice both community psychiatry and addiction psychiatry, and I don’t think of those as two separate jobs. Almost every family I see has been touched by a substance use disorder (SUD) somewhere in their history, whether that's a parent's crack use in the 90s, a sibling's opioid use now, or a patient's own struggle sitting quietly underneath a grief presentation. You cannot do one of these jobs well without also doing the other.

The gang violence that once defined neighborhoods like mine has, in the years I've been practicing, gone down. That's real, and it matters, and it doesn't get said often enough — decline is a less compelling story than crisis. But the quieter that gunfire gets, the more clearly you can hear what was underneath it the whole time: a level of grief so constant that most of it never makes it into a headline or a statistic. Most of what comes through my door isn't the aftermath of a shooting. It's something slower, and in some ways harder to name.

A lot of what I sit with is grief for people my patients had a complicated relationship with. A father who was in and out of their life. A brother whose substance use disorder made him someone to both love and fear. A parent who did their best inside conditions that made "their best" fall short of what a child needed. As an addiction psychiatrist, I see a specific version of this constantly: grief for someone lost to an active SUD long before they died, and then grief again when they actually do. Families often mourn a person twice — once when the disease takes over who they were, and once when their body finally goes. By the time I meet someone in my grief work, they've frequently been carrying an earlier, unacknowledged loss for years without ever calling it that. When someone like that dies, the grief doesn't arrive clean. It arrives braided with relief, with guilt about the relief, with anger that never got resolved while there was still time, with love that never stopped being love even when it should have been safer to let it go. Textbook grief work assumes a relationship you're sorry to lose. Ambivalent grief is different — you're mourning a person and a chance that's now permanently closed, the chance that things might have eventually gotten better between you. I've learned not to rush people toward resolution here. Sometimes the most useful thing I can do is simply tell someone that it is possible to grieve a person and feel genuine relief that a difficult chapter has ended, and that neither feeling cancels the other out or makes them a bad son, daughter, or sibling.

Other patients come to me not after one death but several, stacked close together — a mother, then an uncle, then a grandmother, within a year or two of each other. Increasingly, the cause isn't violence. It's diabetes, heart disease, kidney failure, and sometimes overdose sitting right alongside them — the chronic conditions that concentrate in neighborhoods that were disinvested from decades ago and never fully reinvested in. As an addiction specialist, I don’t separate substance use disorders from that list. They behave like the others: chronic, relapsing, shaped by access to care, and lethal when neglected. A family that loses a grandmother to kidney failure and an uncle to overdose within the same year isn't grieving two unrelated tragedies. They're grieving two expressions of the same underlying disinvestment. This, too, is a legacy of the same era that brought the drug economy and the incarceration wave: communities that lost grocery stores, primary care access, and economic stability all at once, and are still living out the health consequences a generation later. Grief research has a term for this — bereavement overload — but knowing the term doesn't make the clinical work simpler. When someone has buried three family members in eighteen months, they often don't have language for what they're feeling because they never had time to finish feeling the last loss before the next one arrived. My job in those cases is less about processing any single death and more about helping someone find solid ground while still standing in moving water.

That same disinvestment shows up in another kind of grief, one that doesn't require a funeral at all. The chronic conditions that cluster in under-resourced neighborhoods don't only kill people — they also disable them, slowly, in ways that end the life someone had without ending the person. A patient whose serious mental illness emerged in their twenties, in a neighborhood without the early intervention resources that might have changed its course, is grieving a trajectory that got permanently altered — the job, the independence, the imagined future. A patient aging out of the ability to live alone, drive, or manage their own care, often faster than they would have with better healthcare access along the way, is grieving too. This is disenfranchised grief: loss that comes with no funeral, no sympathy card, no social permission to name it as loss at all. People are expected to simply adjust. Few ask them how they're doing with what they've lost, because nothing died — except the version of their life they were building, in a place that made it harder to keep building it.

In practice, I haven’t treated gang violence as the center of this work, because it isn't, not in the way it once was. But I can't separate any of what I do from the crack epidemic's afterlife, because its architecture is still standing in the health outcomes, the family fragmentation, and the disinvestment that never fully reversed. And as someone who treats substance use disorders as a clinical specialty, I find myself returning, again and again, to how differently this country chose to see two drug crises that in many ways resembled each other. When crack devastated Black urban neighborhoods, the dominant narrative was criminal, not clinical. We got "crack babies," a term researchers have since substantially walked back, used to suggest a generation of children was permanently and irreversibly damaged. We got mandatory minimums and a hundred-to-one sentencing disparity between crack and powder cocaine, despite the drugs being pharmacologically almost identical. We got prisons instead of treatment centers, and a "superpredator" narrative that treated substance use disorder and the violence around it as a moral failure requiring punishment, not a public health crisis requiring care. When the opioid epidemic arrived, disproportionately in white suburban and rural communities, the language shifted. We started talking about SUDs as a diagnosis, about the underlying biology, about grieving families as victims rather than culprits. Naloxone distribution became public health policy instead of a controversial idea. Empathy became the default frame in a way it almost never was for crack cocaine. The disparity wasn't only rhetorical. It showed up in who could actually get evidence-based treatment: methadone clinics were historically concentrated in urban, disproportionately Black neighborhoods and wrapped in heavy restriction and stigma, while buprenorphine, introduced later, could be prescribed from a private doctor's office — a model of care that reached whiter, better-insured patients first and with far less surveillance attached to it. Two medications treating the same disease, distributed along strikingly similar lines to who got compassion and who got handcuffs.

I don't say this to relitigate the past or to diminish the very real suffering the opioid crisis caused — that suffering was real, and the compassionate response to it was overdue and correct. I say it because the asymmetry isn't abstract history to my patients. It's the reason a grandfather has a felony record for an amount of crack that would be a misdemeanor today. It's the reason a neighborhood got a prison instead of a clinic, and is still working through the downstream health and economic consequences of that choice decades later, one diabetes diagnosis and one funeral at a time.

Community psychiatry, done honestly, means holding the clinical needs of the person in front of you alongside the structural forces that shaped what they lost and how much support they had to lose it with. Doing that work as an addiction psychiatrist too means never letting SUDs sit outside that picture, treated as a separate moral category instead of another chronic illness shaped by the same disinvestment as everything else on this list. Most of the grief I see now isn't dramatic. It's slow, layered, and often invisible to anyone outside the exam room — grief for complicated people, grief stacked in clusters, grief for a former self. None of it makes the news. All of it deserves care. If we want to actually help people carry this, we have to fund long-term mental health and addiction treatment infrastructure in neighborhoods like this one with the same urgency and lack of moral judgment we eventually extended to the opioid crisis. And we have to be honest that the crack epidemic's afterlife isn't over — it just stopped looking like what we expected it to look like, and started looking like a waiting room.

That distinction — prevention versus damage control — has been on my mind lately, because I've been watching “Evil Lives Here: My Child the Killer.” If you haven't seen it, the format is simple and devastating. Parents sit across from a camera and describe, in painstaking detail, raising a son who went on to kill — sometimes another family member. What strikes me most isn't the violence itself. It's the retelling. Almost every parent, at some point, says a version of the same thing: “Looking back, there were signs.” A withdrawal from friends. A fixation that curdled into something darker. A break from reality that the family mistook for teenage moodiness, or stress, or a phase.

I recognize that hindsight. I saw it constantly in my own patients' families, except I usually met them after the fact — after an arrest, after a hospitalization, after a sentence had already been handed down. By the time someone reached my caseload, the system had already failed to catch what a parent, sibling, teacher, or neighbor might have caught months or years earlier, if only they'd known what they were looking at.

This is not an argument that mental illness causes violence. It doesn't, not in the overwhelming majority of cases. People living with serious psychiatric conditions are far more likely to be victims of violence than perpetrators of it, and stigmatizing an entire population because of the actions of a small minority does real harm — it drives people away from treatment, isolates them further, and makes families ashamed to ask for help. Any honest conversation about violence prevention has to start there.

But a smaller, more specific claim is also true: there is a limited set of warning signs — a marked deterioration in functioning, a break from shared reality, an escalating fixation paired with access to means, a sudden and complete withdrawal from every relationship that used to matter — that, in combination, deserve more attention than they get. Families are usually the first to notice these shifts and the last to know what to do with what they're noticing. They don't lack love or vigilance. They lack a vocabulary. Nobody teaches the public how to distinguish an ordinary rough patch from a genuine psychiatric emergency, or what a first psychotic break can look like in a twenty-year-old, or how differently urgent a sudden personality change is compared to garden-variety teenage withdrawal.

I don't think the answer is more true-crime television, however compelling I find it at eleven at night. I think the answer is much less dramatic than that. It's basic mental health literacy, taught the way we teach CPR or the Heimlich maneuver — not because everyone will need it, but because almost everyone will eventually stand close enough to someone who does. It's primary care doctors, school counselors, and clergy trained to recognize early psychosis instead of waiting for a crisis to force the issue. It's crisis lines and mobile response teams funded well enough that a worried parent has somewhere to call before things escalate, not just after.

It's also our laws. In most states, the legal bar for compelling someone into evaluation or treatment is dangerousness that has already become imminent — a specific threat, a recent act, a person visibly in crisis in front of the right official at the right moment. Families watching a loved one slide into psychosis over months are often told, correctly, that nothing can be done yet. Not because clinicians don't recognize what's happening, but because the law doesn't let anyone act on it until the person meets a threshold that, by definition, arrives too late to prevent the harm everyone saw coming. A small number of states have expanded assisted outpatient treatment laws — court-ordered treatment plans for people with a documented history of relapse into dangerous, unmedicated psychosis, delivered in the community rather than a hospital — and the evidence on these programs shows real reductions in hospitalization, arrest, and violence among the narrow population they target. Making it easier for families and clinicians to petition for this kind of intervention, earlier and with less bureaucratic friction, is not the same as expanding involuntary commitment broadly, and it shouldn't be confused with it. Civil liberties advocates raise legitimate concerns here — the history of psychiatric confinement being used to warehouse or silence people is real, and any expansion of these laws needs strong oversight, sunset provisions, and a high evidentiary bar to avoid repeating it. But the current default, where a family's only lawful option is to wait, is its own kind of failure. I watched it play out from the other side of that failure for years: families who had tried everything they knew how to try, and were told there was nothing more to do until something irreversible happened first.

And it's a public conversation about serious mental illness that can hold two things at once: compassion for people living with these conditions, and clear-eyed honesty about the narrow set of circumstances where early intervention genuinely could have changed the outcome.

By the time I met my patients, we were doing triage, not prevention. I got good at that work, and I felt like I was making an actual difference in people’s lives. But I keep thinking about how many of the families I sat with had, months before I ever heard their names, watched something shift in someone they loved — and hadn't had the first idea what to do about it. That gap, between noticing and knowing, is where I think we owe the public far more than we've given them.

For me, that narrowing is also a feature of ADHD. My attention does not scatter in session; it locks. I suspect many clinicians who share this wiring recognize the same pattern — an almost total absorption that makes us effective in the room and, at times, unreachable outside it. We rarely discuss this as an occupational hazard, but I have come to think it is one worth naming, and a recent afternoon made the case for me in a way I could not ignore.

I stepped out between patients to find four missed calls from my partner, along with texts from her and from her sister, both using the word "emergency" and asking me to call back immediately. I also carry a trauma history, and my nervous system did what trauma-conditioned nervous systems do: it did not wait for further detail. By the time I reached my partner and learned the concern involved our child — serious enough to warrant that language, though not the worst-case outcome my body had already braced for — I had already moved through a full physiological alarm response with no privacy and no time to process it. My schedule for the remainder of the day was full. It stayed full.

I want to be careful here not to frame what came next as a story about willpower, because it isn't. I regulated, finished my day, and saw every patient competently. Clinicians are very good at this kind of compartmentalization; it is arguably a prerequisite for the job. But that competence in the moment is precisely what obscures a structural question our field does not ask often enough: what does it cost a clinician to be maximally present for patients, session after session, in systems that build in no room for the clinician to also be a reachable parent, partner, or person in crisis?

This question is not abstract for those of us in public and community mental health. County and safety-net systems run on productivity targets and full panels by design — the need is real, and the staffing is thin. There is little structural slack for a provider to take an unplanned five minutes, and even less cultural permission to ask for it. We tell patients that emotional regulation requires space and support. We rarely extend ourselves the same logic, and I don't think that gap is incidental to how these systems are built.

None of this means the answer is to blunt the very attentional capacity that makes many of us effective clinicians. Hyperfocus, whatever its cost, is also why I can sit with a psychotic, grieving, or frightened patient for a prolonged period of time. The more useful answer is structural: brief protected windows between blocks of patients, cultural normalization of clinicians stepping out for family emergencies without apology, and psychiatric leadership that treats provider mental health — including neurodivergent and trauma-affected clinicians — as a workforce issue rather than an individual one.

Still, I would be doing my own field a disservice if I framed this purely as a systems problem and let myself off the hook entirely. Alongside whatever changes our institutions make, I am continuing to work on my own relationship with hyperfocus — learning where the line sits between the attention that makes me a good psychiatrist and the attention that makes me hard to reach when it matters most. That work is not about becoming a different kind of clinician. It is about becoming someone who can be fully present in the room and still find my way back, quickly, to the people I love and am responsible for outside of it.

Community psychiatry asks a great deal of the people who staff it. It should also ask, more seriously, what those people need in order to keep showing up — for their patients, and for the families waiting for them at home.

None of it prepared me for hearing it about my own daughter.

At her well-child visits, her pediatrician told us she was fine. Within normal limits, developing on her own curve, nothing to worry about. For most families, that reassurance from a trusted pediatrician is where the story ends. It nearly ended there for us, too.

But something didn't feel right. She could speak, but full sentences didn't come easily when it mattered most — she'd reach for one, stall, and dissolve into frustration instead. Direct questions seemed to slide past her, and her attention drifted before she could land on an answer. Her fingers struggled with the pincer grasp, with stringing beads, with the small mechanical tasks that other toddlers seemed to master without visible effort. I recognized the vague discomfort of a pattern I couldn't yet name — the same instinct I spend my clinical days asking patients to trust in themselves, even when an overstretched system doesn't have the time to chase it down. So my wife and I pushed. We asked for a more thorough assessment, then asked again, until we were referred to our school district's early intervention program for a full developmental evaluation.

That evaluation found what the well-child visits had not: measurable delays in expressive language and fine motor development. Now we are waiting on a further evaluation with an autism specialist. And I have discovered that all my training did not prepare or protect me against the overwhelming anxiety of waiting for someone else to tell me who my child is.

I have also not stopped thinking about how easily this could have gone differently. I had the vocabulary, the professional standing, and the familiarity with how to navigate referral systems to insist on a second look. Not every parent does. I don't say that to indict our pediatrician, who was working from a brief visit and a standard screening tool, doing exactly what she was trained to do. I say it because it points to something larger than my daughter's chart: how much a child's access to early intervention can depend on a parent's ability to advocate against a reassuring, well-intentioned "she's fine." I see the flip side of this every week in my own work as a community psychiatrist in a county system, where many of my patients have spent years being told, by one system or another, that they were fine — or that their symptoms were behavior, or a choice, or simply not urgent enough to warrant a closer look. Watching that gap operate on my own daughter's behalf, rather than against one of my patients, was disorienting in a way I'm still sitting with.

There's a strange double life I'm living right now. In my office, I'm the one with the framework, the differential diagnosis, the years of pattern recognition. I can sit with a patient's fear and hold it steady, because I'm not inside it. I know how to explain that a diagnosis is a tool, not a verdict — that it opens doors to services and support rather than closing off a future.

At home, I am not the expert. I am just a father who lies awake reorganizing the same handful of facts, trying to make them mean something more certain than they do. I know, intellectually, that developmental trajectories are not destiny. And still, late at night, none of that knowledge stops the spiral.

This is, I think, the part that surprised me most: how little my expertise actually protects me. If anything, it sharpens the fear. I know what an autism spectrum disorder (ASD) diagnosis can mean, and what it doesn't mean, and exactly how wide that range is — and holding all of that at once, about my own child, is a very specific kind of exhausting.

Psychiatry, for all its humility about the mystery of the mind, is still a field that prizes intervention. We assess, we diagnose, we treat. There is comfort in that structure, even when the answers are imperfect. But nothing about the next several weeks is mine to control. I cannot speed up a waitlist. I cannot make a specialist see something clearly that may simply take more time to reveal itself. I cannot will my daughter's fine motor skills forward through sheer professional will, the way I might, in another context, adjust a treatment plan and expect to see change.

I can only watch her string beads, one at a time, and notice.

That powerlessness has taught me something my training only gestured at: how much of what we ask families to do — to wait, to hope carefully, to love a child without needing to fully understand her yet — is genuinely, physically hard. Not hard in the abstract way it can sound written into a treatment plan. Hard in the way that sits in your chest at pickup time, when another child answers a simple question smoothly and yours grows frustrated mid-sentence, the words there but somehow out of reach.

And yet powerlessness has not been the whole of it. What I cannot control, I have tried to work around. My wife and I read the literature ourselves rather than waiting to be handed it. We pushed past a reassuring first impression to get a fuller evaluation. We searched out a preschool that doesn't run on a single fixed idea of how a child should learn, or how quickly, and found one that treats her pace as information rather than a problem. None of this changes what the specialist will eventually tell us. But it has changed how we're waiting — less as bystanders hoping for news, more as people doing the deliberate work of building her a wider margin, whatever the evaluation finds.

I don't think our pediatrician did anything wrong, exactly. Standard screening tools have real limits, and a fifteen-minute well-child visit is a narrow window onto a child's development. But I keep thinking about the parents who don't have a clinical background to draw on, who don't know that "let's just watch and wait" is a clinical decision worth questioning, who may not know that early intervention programs exist independent of a pediatrician's referral or that they can request an evaluation directly. The gap between "she's fine" and "here are measurable delays" was not closed by better data. It was closed by us knowing how, and feeling entitled, to push.

That has been a powerful reminder that patients are often the first to notice that something in the pattern doesn't fit, well before it's measurable by any instrument I have. I have always empowered my patients to advocate for themselves when they insist that something is wrong — with their body, their medication, their care, their circumstances — that doesn't match what the chart, or the system, has already decided about them. And I will continue empowering my patients to be strong advocates for themselves.

If she is diagnosed with ASD, she will still be exactly who she is right now: energetic, curious, endlessly delighted by her collection of squishy toys — the ghosts especially. A diagnosis will not add or subtract anything from her. It will only give us better language, and hopefully better access to support, for the child already in front of us.

I wish I could say I've fully internalized that. Some days I do. Other days, I catch myself grieving a version of a future I hadn't even realized I was assuming — and then feeling guilty for grieving anything at all, because she hasn't lost a single thing. She is still here, still herself, still learning at her own pace.

What I keep returning to is this: anxiety and powerlessness are not signs that I'm failing to apply what I know. They're simply what it feels like to love someone whose path you cannot see yet. My patients have been telling me this for years, in their own words, and I nodded and understood and thought I understood. I am only now learning the real difference between clinical empathy and this.

The evaluation will come. We'll get more information, though maybe not the certainty I keep reaching for. And whatever the specialist tells us, my daughter will still string those beads, one at a time, in her own time. I hope I remember, long after this particular wait is over, what it felt like to be on this side of the exam room door — and to let that memory make me a better physician to the patients who sit across from me.

-A psychiatrist in the margins

I went into medicine, and specifically addiction psychiatry, for the reward of improving other lives. County clinic work seemed like the purest expression of that calling. Here in the public system, I see the people who fall through every other crack: those with severe substance use disorders layered on trauma, psychosis, housing instability, and generations of systemic neglect. Many are on Medi-Cal. Many have nowhere else to go. The satisfaction of helping someone stabilize on medication-assisted treatment, rebuild after a relapse, or simply show up consistently for appointments feels raw and real. This is why I chose this path.

Yet after months (or years) in the county clinic, the idealism collides with reality.

The caseloads are crushing—often 200-300 patients per clinician in understaffed community mental health settings. Documentation demands are endless, bureaucracy layers on layers, and resources are chronically scarce. You fight for basic things: timely labs, housing referrals, or even enough time in the day to provide more than a 15-minute med check. Addiction work here means holding space for patients whose lives are in constant crisis—overdoses, incarcerations, lost custody battles—while knowing the system reimburses poorly and burns people out at alarming rates. Many colleagues leave for private practice citing higher wages, better autonomy, and flexibility, not because they stopped caring, but because sustainability matters.

I love the mission-driven environment. The patients remind me daily why addiction psychiatry matters: these are some of the most marginalized, stigmatized individuals in healthcare, and showing up for them with evidence-based care (MAT, therapy integration, harm reduction) can literally save lives. There’s a deep camaraderie among the staff who choose this work. But the emotional toll is heavy. Burnout isn’t abstract—it’s the quiet resentment that builds when you’re always behind, when you can’t give patients the time they deserve, and when your own family time shrinks because the demands of the job don’t respect boundaries.

Lately, what weighs on me most is how little I’m truly present for my own family. My child asks for Daddy when I leave in the mornings, and when I get home they’re already asleep. The work I love sometimes leaves nothing left for the people I love most. My spouse carries more than her share because my energy is drained and my schedule is usually full. I chose medicine to help others build better lives, yet I’m watching my own family life erode in small, daily ways—rushed dinners where my mind is still in the clinic, and missing precious moments with my child that I can never get back.

Thinking about private practice still triggers that guilty twinge. Better pay, control over my schedule, fewer patients per day but deeper engagement with each one. The ability to build a practice that prioritizes quality over quantity. Autonomy from insurance checkboxes and county protocols. And yes—most importantly—the economic stability that would let me be more present for my family. To play “fishies” (e.g. a game with plastic fish you need to catch) with my child without exhaustion weighing me down. To sit at dinner fully engaged, without the mental load of unfinished charts lingering in my head. To protect weekends and evenings so my child grows up knowing their parent is there, not just in body but in spirit. To reduce the constant financial stress that shadows every decision and allow my spouse and me to build memories instead of just surviving the week.

It feels selfish. Shouldn’t the reward of the work be enough? Didn’t I sign up knowing public service wouldn’t make me wealthy? Am I betraying the very patients I chose to serve by wanting more time and energy for my own child?

The truth is more complicated. County mental health clinics expose the gap between societal rhetoric about valuing mental health and the actual support provided. Addiction treatment in the public sector is notoriously under-resourced despite the enormous costs of untreated substance use—overdoses, emergency visits, lost productivity. Psychiatrists in these settings often face high emotional demands, complex comorbidities, and limited backup, contributing to burnout rates that push talented clinicians out the door. Wanting fair compensation and a sustainable life isn’t abandoning the mission; it’s what allows me to stay in it long-term—and to show up as the parent and partner I want to be.

Private practice wouldn’t mean turning my back on the underserved. Many who make the switch carve out sliding-scale slots, pro bono time, or partnerships with community organizations. It could mean shorter wait times for my patients overall, more flexible hours for working adults in recovery, and the headspace to innovate rather than just survive the day. More importantly, it could mean I return to my family each evening with enough left in the tank to be fully there.

Society romanticizes the overworked public servant sacrificing everything. It’s a comforting myth, but it ignores how we undervalue care for the vulnerable. If we want compassionate addiction psychiatrists to remain in the field for decades—rather than a few intense years—we need to stop treating financial pragmatism as moral compromise. Doctors are human. We have families, debts from a decade of training, and limits to our resilience. Being fully present for my children isn’t a luxury; it’s part of living a whole life while doing meaningful work.

Working in the county clinic has sharpened my original “why.” Improving lives doesn’t require impoverishing or exhausting my own—or my family’s. The patients here deserve providers who can sustain empathy and expertise over time, not ones running on fumes. My kids deserve a parent who isn’t perpetually depleted.

Economic stability so I can be more present for my family feels less like selfishness now. After seeing the realities of county mental health, it feels like necessary realism. The goal was never martyrdom. It was impact—both for my patients and for the people who matter most at home. And sometimes, impact requires choosing sustainability so you can keep showing up, fully and wholeheartedly, in every part of life.

Her arrival marked a sharp turning point. What had once been a challenging but meaningful workplace became a source of daily stress and dread. The work itself hadn’t changed—leadership had. And with that shift came an environment that grew increasingly unsafe, dismissive, and hostile to professional judgment.

When I raised concerns about safety—both for my team and for myself—the response was not curiosity or collaboration, but minimization and retaliation. Professional concerns were reframed as personal shortcomings. When I asked for a scheduled lunch break within the legally specified time frame during my shift, that request was treated not as a standard labor protection, but as an inconvenience. After that, interactions became colder, scrutiny increased, and the message was unmistakable: speaking up had consequences.

I was told I was “too sensitive.” That I was “too inexperienced.” These labels might have landed if they reflected reality. They didn’t. Before this role, I had worked with some of the sickest psychiatric patients in the county—across hospitals, outpatient clinics, and on the streets doing street psychiatry. I have assessed risk in emergency rooms and de-escalated crises in unpredictable settings. I know the difference between discomfort and danger. I did then, and I do now.

The gaslighting was gradual but effective. Over time, my professional judgment was repeatedly questioned, not because it was flawed, but because it challenged authority. This is how abusive supervision operates: control is framed as leadership, dismissal as feedback, and retaliation as consequence. What mental health professionals describe as malignant narcissism thrived in a system where power went unchecked.

What ultimately helped me leave was realizing I wasn’t alone. Members of my team—and staff from other sections of the program who had interacted with this supervisor—expressed agreement with my decision. Their support wasn’t based only on what I shared, but on their own experiences. They had seen the same patterns. Their validation mattered not because I needed permission, but because it confirmed that the problem was systemic, not personal.

I stayed longer than I should have. Like many people in helping professions, I believed endurance equaled professionalism. I feared being labeled difficult or uncommitted. That fear is not accidental; it is how abusive workplace cultures sustain themselves, especially in organizations that claim moral authority.

Community reintegration programs exist to support stability, accountability, and growth. When leadership retaliates against legitimate safety concerns, minimizes expertise, and erodes basic protections, the mission collapses from the inside out. No program can claim to rehabilitate while harming those tasked with carrying out its work.

Leaving was not quitting. It was drawing a necessary boundary. It was choosing safety over silence, integrity over appeasement, and clarity over control.

For me, leaving was not the end of my commitment to this work. It was the moment I reclaimed it.

And I’ve done that. For years, I’ve shown up, day after day, often in high-stress, high-need environments with limited resources. I’ve honored my contract and my patients. But I’ve also reached a breaking point.

The burnout I’m experiencing isn’t abstract—it’s physical, emotional, and deeply personal. It has started to unravel my well-being, my sleep, my mental clarity, and my relationships. Recognizing that I needed a change, I initiated a transfer to a different site. I wasn’t trying to leave the program—I was trying to stay in it, sustainably.

But after beginning the process, I learned that the site I hoped to move to wasn’t even active in the program. No other available sites would offer a less harmful alternative. The message became painfully clear: you can’t leave, and there’s nowhere else to go.

I want to be very clear: I am not trying to break my contract. I’m asking for a path to fulfill it without sacrificing my health in the process. But no such path exists.

And the consequences for stepping away—even for valid reasons—are devastating.

If I were to be found in breach of contract, I would be required to:

• Repay every dollar of my loan repayment—plus interest at the highest federal rate.

• Pay a penalty fee of $7,000 for every month of service not completed.

• Face potential exclusion from all future federally funded health programs, grants, and contracts—a permanent black mark on my professional record.

These consequences are so severe that they effectively trap providers in unsafe, unsustainable roles, even when their ability to practice safely is compromised. The system leaves no room for nuance, no space for reasonable accommodations, and no real concern for provider wellness—despite all the rhetoric we hear about the importance of preventing burnout in healthcare.

It shouldn’t be this way.

Loan repayment programs like the one I’m part of serve a vital purpose. They bring care to the communities that need it most, and they offer real support to those burdened by the cost of medical education. But we need to talk about the human cost of these contracts when no flexibility exists.

Burnout is not a failure. It’s a signal that something has to change. But in programs like this, change simply isn’t an option—until it’s too late.

I have one year left. I want to finish strong. I want to serve. But I also want people to understand what it really means to be locked into a system that leaves no room for wellness, no room for adjustments, and no room for the provider to say, “I need help” without risking everything.

We must do better—for the sake of the people we serve, and the people who serve them.

But there’s a quieter pain I rarely talk about: the one that comes when I go home.

A few months ago, I had to shut off the upstairs toilet in my house. Water had started dripping through the kitchen ceiling. I stood there, watching it fall—slow and steady like a metronome—onto the kitchen counter I used as a food prep area the night before. I had shut off the toilet hoping it would stop the leak, but eventually it just came back.

There’s a strange dissonance in intently listening to someone as they confront the worst day of their life, and then coming home to mold creeping behind the drywall. I make just enough to keep things going, but not enough to keep them from falling apart.

And yet, I went into this work with open eyes and a full heart. I didn’t choose addiction psychiatry for the paycheck. I chose it because I wanted to help people rebuild their lives. But the idealism I carried into this field sits uneasily beside the very real desire for a life that feels stable—and sometimes, beautiful.

The truth is: I want both. I want to serve people who’ve been written off by the world. And I also want a house where the ceiling doesn’t leak. I want to help others reclaim their lives from addiction. And I want mine to feel sustainable—not stretched to the breaking point by medical school debt, rising insurance premiums, or the quiet costs of choosing a career in public service.

This tension isn’t unique to me. Many of us—doctors, social workers, public defenders, teachers—feel it too. We’re driven by mission, but we’re also trying to live. And in a society that still links income with value, choosing to work with underserved populations can feel like an act of economic self-erasure.

Some days I wonder: what would it feel like to work fewer hours, earn more money, and spend my weekends fixing my house instead of charting or catching up on missed sleep? But then I think of the patient who finally made it to her fourth consecutive therapy appointment after years of no-shows. The man who had been through multiple rehabs and incarcerations, who has finally been able to have stable work for the first time in his life. The mother who called about her son and told me she was scared, but hopeful. Those moments are everything. But they don’t pay the plumber.

So where does that leave us?

We need to stop romanticizing self-sacrifice and start building systems that let care providers live with dignity. That means real loan forgiveness, better financial incentives for physicians in public service, and better infrastructure for clinicians working in safety-net settings. It also means making room—emotionally and culturally—for the honest, complicated truth: you can love what you do and still want more from life.

I’m still living with the upstairs toilet shut off. I’m still working, still believing in the power of people to change, and still navigating that quiet grief of wanting something easier. But I no longer see it as a betrayal of my values. I see it as part of being human.

Wanting stability doesn’t make me selfish. It makes me honest. And maybe, just maybe, that honesty is part of the healing too.

This isn’t something we talk about much in our profession. There’s an unspoken expectation that once you’ve trained, licensed, and earned the credentials, you’ve somehow learned to “just deal with it.” But that’s not how healing works. Trauma doesn’t disappear because you’ve memorized the DSM. Flashbacks don’t wait for your lunch break. And no amount of psychoeducation can always keep intrusive memories at bay.

What makes this even more layered for me is that I come from the same communities as many of my clients. I was raised in the same neighborhoods, shaped by the same cultural expectations, and exposed to the same intergenerational wounds that still echo through their sessions. This familiarity builds trust—something many of my clients say they’ve rarely felt in clinical spaces. But it also builds proximity to my own past. Their stories can feel like echoes of my own.

There’s a unique pain in feeling seen through someone else’s suffering.

I want to be clear: I still love the work I do. There’s a sacredness in holding space for someone’s healing. But there’s also a quiet, relentless labor in managing your own mental health while being a lifeline for someone else’s. It’s the duality of being both the anchor and the ship in a storm.

Some days, I’m able to compartmentalize. Other days, I’m triggered in ways I didn’t expect—by a phrase, a look, or the silence between words. I carry grounding tools in my pocket like a parachute, hoping I won’t need to pull the cord mid-session. I’ve done the work. I’ve had the therapy. I’ve read the books. But healing is not a finish line—it’s a landscape I walk through every day, and some days, the terrain is brutal.

Many clinicians from marginalized communities are under-supported, overworked, and expected to serve as cultural translators within systems that were never built for us. Burnout isn’t just about caseloads—it can also include being everything to everyone, while still trying to be something to yourself.

So why keep doing it?

Because I know how transformative it can be to realize you’re not alone—not just in your pain, but in your recovery. And I want to be that for someone else, even if it costs me a little more some days.

But we need to create space for these conversations—not just among therapists behind closed doors, but in the wider mental health landscape. Providers are people. We bleed, too. And acknowledging that doesn’t diminish our professionalism—it deepens our empathy, strengthens our authenticity, and protects us from the isolation that so often shadows our roles.

To those of us who are healing while helping: I see you. I am you. And you are not alone.

-A psychiatrist in the margins

A young boy had been struck by a car. By the time I met him, he was already in a coma in the neuro ICU. His mother and father sat quietly at his bedside, holding back emotion, waiting for answers. She spoke only Spanish. I was asked to translate.

She asked, through me: “Will he wake up?”

The chief of neurosurgery, a towering presence who had worked in that hospital for two decades, casually shrugged his shoulders and answered: “I don’t know.”

Then, after a brief pause: “Sorry.”

And he walked away.

No further explanation. No attempt to comfort. Just two words—and he moved on with rounds.

The parents were left sitting in silence. I stayed for barely two minutes before I, too, had to move on with the team. I’ve carried guilt about that moment ever since—not because I had any power to change the outcome, but because I couldn’t offer them what they clearly needed most: time, presence, and humanity.

Later that day, I overheard the chief defending this interaction by casually saying to a resident, “She should’ve made sure her kid didn’t run into the street.”

It wasn’t just cold. It was cruel.

And it wasn’t an isolated moment. About a month later, I saw the same neurosurgeon delay an emergency aneurysm repair on a patient with a burst bleed. The case was deemed high-risk, and he didn’t want it to affect his outcomes. He waited a full day for another attending to take the case instead. Time—critical in neurosurgical emergencies—was treated as expendable.

These moments didn’t happen in a vacuum. They took place in a county hospital that serves a large immigrant and working-class population. Spanish is spoken often. Many patients trust the system because they have no other option. When that trust is met with detachment—or worse, with moral judgment—it reveals something deeply wrong.

This neurosurgeon wasn’t new to the job. He wasn’t obviously overwhelmed or at the end of a long night. He was a seasoned, well-established figure in the department. And that’s what made it harder to reconcile: Was this burnout? Or was it something more deeply embedded—a way of practicing that had become hardened over years, perhaps decades?

We need to talk about what we reward and normalize in medicine.

We celebrate excellence, but we define it in narrow terms: surgical outcomes, publication counts, academic titles. Rarely do we reward empathy. Almost never do we reward moral courage. And when someone chooses to walk away from suffering—or avoid a risky case to preserve their numbers—we look the other way.

But we shouldn’t.

A physician who withholds compassion isn’t just protecting themselves—they’re failing the people who rely on us at their most vulnerable. A surgeon who delays a life-saving procedure because of how it might affect their statistics isn’t being cautious—they’re gambling with someone else’s time, and life.

The best physicians I’ve encountered aren’t the ones with perfect outcomes or elite résumés. They’re the ones who take a moment, even when they have nothing left to offer medically, to be present. They’re the ones who don’t hide behind data when the patient needs to be seen, heard, and acknowledged.

That morning in the ICU shaped how I think about medicine. Not because it showed me what greatness looks like, but because it showed me what we too often excuse in its name.

We need to ask more of each other. We need to expand our definition of excellence to include empathy, accountability, and presence. Because if we don’t, we risk becoming something far worse than imperfect—we risk becoming indifferent.

Medicine is not just a science. It’s a moral practice. And if we forget that, the cost isn’t measured in metrics. It’s measured in human beings.

-A psychiatrist in the margins

A physician assistant (PA) on the team had made a friendly lunch suggestion to one of the residents. Later that day, in the residents’ workroom, the story came up again—this time as the punchline.

One of the residents laughed and said, “I don’t know man, he thinks we’re friends.”

The others laughed too. No one said anything else about it. The moment passed.

That never sat right with me. I wasn’t offended on behalf of the PA because of a single joke. I was troubled by the culture it reflected—a quiet but persistent undercurrent in medicine where titles can determine not just responsibility, but respect.

That experience shaped how I view teamwork in healthcare. I’ve come to believe, more strongly with each passing year, that medicine does not work unless we work together. And that means respect must be extended across roles—not just upward.

Physician assistants and nurse practitioners (NPs) are not “wannabe doctors.” They are highly trained, dedicated professionals with their own rigorous education, licensure, and clinical judgment. In many hospitals—especially county systems or resource-limited settings—PAs and NPs often serve as the backbone of continuity, the link between rotating residents and fragmented care.

To treat them as lesser is not only disrespectful—it’s dangerous. It undercuts collaboration, silences valuable clinical input, and builds walls where there should be bridges. In a system already strained by burnout, staffing shortages, and increasing complexity, we simply cannot afford that kind of ego.

Medicine needs team players. Titles do matter—for scope of practice, for responsibility—but they should never be confused with a license to belittle or exclude. And they certainly don’t determine someone’s worth as a colleague, or as a person.

The best teams I’ve seen—on rounds, in the OR, and in clinic—are those where people are treated like professionals, regardless of whether their badge says MD, DO, PA, NP, or RN. The best doctors I’ve worked with are confident enough in their own role to treat others with dignity. They don’t confuse hierarchy with superiority.

That offhand comment from the resident may have been a joke. But the mindset behind it isn’t funny—it’s outdated, and it’s holding medicine back.

It’s time we build a culture in healthcare where every team member is respected. Where collaboration is not just encouraged but expected. Where no one is made to feel foolish for thinking we might be… friends.

Because the truth is, we should be.

-A psychiatrist in the margins

A few of my visits take place in a clinic, especially when patients are stabilizing and preparing to transition to traditional outpatient care. But the majority happen in the field. I meet patients where they are, usually with a member of our clinical team. These are high-acuity visits involving complicated medication decisions, risk assessments, and psychiatric care—all outside a typical medical setting.

This is what public-sector, community-based addiction psychiatry looks like. It’s critical work, and it’s dangerously under-resourced.

The program I work for often feels like an afterthought of the larger system I work in. We’re expected to deliver high-stakes psychiatric care without the infrastructure to support it. Many of the settings we work in lack basic clinical safeguards—no on-site medical support, no dedicated consultation space, no security if a situation escalates. These gaps affect safety. They also limit the quality of care we can ethically provide.

At the same time, the compensation for this work is not reflective of the complexity of work we do. The compensation is on the lower end of what psychiatrists earn in hospitals, outpatient clinics, or even telehealth roles. That disparity isn’t just frustrating—it’s telling. It reflects how the system devalues both the professionals doing the work and the patients receiving the care.

There’s a quiet logic at play: because we’re working in a “mission-driven” public program, we’re expected to tolerate poor pay, safety risks, and inadequate resources. That logic is broken. The communities we serve deserve better, and so do the people trying to serve them.

I love my work. I believe deeply in second chances, in recovery, in justice. But belief isn’t enough. Systems must reflect that same commitment—not just in words, but in resources. If the county is serious about supporting reentry and recovery, it needs to treat this work as essential, not expendable.

That starts with fair pay. It starts with safe, well-supported work environments. And it starts with recognizing that addiction psychiatrists working in forensic field-based programs aren’t doing less—they’re doing more, with less.

We shouldn’t have to choose between service and sustainability.

-A psychiatrist in the margins

During residency, everything had structure—even the chaos followed a schedule. Now, life feels less like a system and more like a shifting equation I’m constantly trying to balance. My patients need me to be present and sharp. So does my family. So does my own health. But there’s no protected time for parenting or recovery after a tough case. There are no attendings watching over me anymore—and somehow, the weight feels even heavier.

The transition out of training is rarely talked about in terms of its emotional and logistical toll. We move from being overworked but closely supervised, to being entirely responsible. And while we’re often reminded to prioritize “balance,” I’ve found that easier said than done. Some weeks I feel grounded as a clinician; other weeks I’m more present at home. Rarely both.

Like many of my colleagues, I’m learning that balance isn’t about perfect symmetry. It’s about shifting with intention—and accepting that I won’t get everything right. Connecting with others in the same stage of life has been grounding. Just hearing someone else say, “I’ve been there,” makes the weight feel a little lighter.

We should talk more openly about what it means to build a life—not just a career—after training.

-A psychiatrist in the margins

At home, I’m a first-time father to a two-year-old. I also try — as best I can — to be emotionally present for my partner. We share the weight of parenting, interrupted sleep, and the daily logistics of raising a young child. And still, the Teams messages don’t stop. The documentation piles up. The stories from work don’t stay at work.

Before medicine, I wrote. Essays, reflections, sometimes thoughts scribbled on index cards and receipts (I was never near a notebook when I needed one). Writing was how I processed the world — mine, and the one around me. But in medical school, that part of me began to disappear. I absorbed an unspoken message: if you’re not studying, you’re falling behind. Creative expression started to feel indulgent. Taking time to reflect felt like slacking off.

That mindset never really left. Even now, when I feel the pull to write — especially when I feel it — I push it away. Out of guilt. Out of habit. Out of exhaustion.

We often discuss burnout in terms of clinical loads, administrative burden, or broken systems. And those are real contributors. But burnout is also about identity loss — the slow erosion of the parts of ourselves that once made us feel whole. It’s what happens when we’re trained to suppress anything that doesn’t directly serve our clinical output.

I serve people who are stigmatized and often dehumanized. I’m raising a child. I’m trying to be a supportive partner. I’m trying to be a good doctor and an advocate. But I am also someone who still needs space to reflect — to write — in order to be well.

Creative expression isn’t a luxury for physicians. It’s a form of preservation. And yet the culture of medicine often treats it as a distraction from “real” work. That message starts early in training and echoes throughout our careers. In an era of rising physician burnout, it’s a message we can’t afford to keep sending.

Wellness initiatives are important — and welcome. But if we don’t allow physicians to be full people, including artists, writers, caregivers, and community members, we’re only treating symptoms. What would it look like to create medical environments where reflection isn’t something squeezed into the margins, but integrated into our professional identity?

We need to make space — in training and in practice — for wholeness. For reflection. For authenticity. Not just for ourselves, but for our patients. Because how can we see them as whole people, if we’re never given the opportunity to be who we truly are?

-Psychiatrist in the margins