When Antidepressants Stop Working: An Honest Look at Ketamine for Treatment-Resistant Depression

The patients who reach me at Tovani Health usually arrive at the end of a long road. They've been on Lexapro, then Zoloft, then maybe Wellbutrin, then a combination, then a different combination. Each medication came with weeks of side effects in exchange for, at best, a partial response. Some of them have been hospitalized. Some of them are still working full-time and look fine on the outside, and the question they bring me is whether they're allowed to stop pretending that fine is what this is.

If that's where you are, I want to talk to you about what treatment-resistant depression actually is, what ketamine does differently, and how I think about whether someone is a good fit for this treatment. There's a lot of hype around ketamine right now and I'd rather walk you through the honest version.

What "treatment-resistant" really means

The clinical definition is simple: you've tried at least two antidepressants from different classes, each at a therapeutic dose for at least six to eight weeks, and neither one worked well enough. (For the full clinical overview of ketamine therapy as a treatment option, see our complete medical guide.) By that definition, somewhere around a third of people with major depression are treatment-resistant. That's not a small population. That's millions of people.

What the definition leaves out is what it actually feels like. The patients I see usually don't describe it as "my SSRI didn't work." They describe it as: I did everything they told me to do. I waited the eight weeks. I took it every morning. I exercised. I went to therapy. And I'm still here. The exhaustion of trying — and of being told repeatedly that the next medication might be the one — is its own layer on top of the depression itself.

There's also a quieter group: people who responded partially to medication and have been told that's good enough. They're functional. They're not in crisis. But they describe a flat, low-level grayness that the medication chipped at without removing. Ketamine has something to offer this group too, even though they don't fit the textbook TRD picture. Specific demographics worth flagging: women in perimenopause, patients over 60, healthcare workers, and first responders all have particular reasons standard antidepressants can underperform.

Why ketamine is genuinely different

Standard antidepressants — SSRIs, SNRIs, bupropion — work on serotonin, norepinephrine, or dopamine systems. They take weeks because they have to gradually push neurotransmitter levels into a new equilibrium and let downstream effects on neuroplasticity catch up. When they work, they work slowly. When they don't, you've spent six weeks finding out.

Ketamine works on a different system entirely. It blocks NMDA glutamate receptors, which triggers a chain reaction: more glutamate release, AMPA receptor activation, BDNF production, and a window of accelerated synaptic plasticity. The clinical effect of all that biochemistry is the part that matters: a window — usually starting within 24 hours, sometimes within hours of the first dose — during which the brain becomes briefly, unusually capable of forming new connections.

That's the actual mechanism. The "dissociative experience" that gets all the attention is a side effect of the dose required to open that window. The therapy isn't the dissociation. The therapy is what your brain does in the days that follow.

In the published literature, sublingual ketamine produces a meaningful response in roughly 60-70% of treatment-resistant depression patients, with measurable mood improvement often within the first 24 hours. That's not a guarantee for any individual person. But it's a fundamentally different speed and a fundamentally different mechanism than another SSRI trial.

Who I think is a good candidate

I tell prospective patients there are three honest yes-es:

The first is the textbook TRD case — two or more antidepressant trials at adequate doses with insufficient response, no contraindications, motivated to engage. For this patient, ketamine isn't a long shot. It's the next reasonable step.

The second is someone with acute suicidal ideation alongside depression. Ketamine's speed matters most here. SSRIs take weeks; ketamine can quiet suicidal thinking within a day. I treat this group with extra care — there's a higher floor of clinical attention required — but the case for trying is strong, often urgently so.

The third is the partial-responder I mentioned: medications got you 40% of the way there and you've been stuck there for years. Ketamine can sometimes complete what an SSRI started, opening the door to gains the medication alone couldn't get to.

Who I tell no, and why

The honest no-list is shorter than people expect, but it matters.

Active or untreated psychosis is a clear no — ketamine's mechanism can destabilize symptoms rather than help. Untreated bipolar mania, similar. Pregnancy and lactation, because we don't have the safety data to feel comfortable. Uncontrolled hypertension, because ketamine transiently raises blood pressure and that has to land on a baseline we trust. Active substance use disorder involving stimulants or dissociatives, because the risk-benefit math changes.

For most of these I'd rather treat the underlying issue first and then revisit ketamine. The blood pressure conversation is a good example: a patient with poorly controlled hypertension isn't disqualified forever. They're disqualified for now, until we — together with their primary care doctor or cardiologist — get the numbers where they need to be.

The thing I want patients to understand is that the no-s exist to protect them. If I say yes to someone who shouldn't be on this medication, the consequence lands in their living room with no clinic between it and them. I'd rather have an awkward "let's wait" conversation than a preventable bad outcome.

What treatment actually looks like with us

If we move forward, the structure is this: an initial loading phase of about six to eight sessions over four to six weeks, then a maintenance schedule that's individualized to your response. The medication arrives at your home from a compounding pharmacy. You dose at home, in a prepared room, with a sober adult sitter present. I'm reachable by phone before, during, and after.

The session itself runs 45-90 minutes of altered experience, with another hour of recovery before you're back to baseline. No driving for at least six hours. The next morning we check in — partly to monitor, partly because what you do in the days after a session is where the lasting change forms.

Between sessions you'll track mood with PHQ-9 (the standard depression rating scale) so we have an objective measure of whether this is working. If the numbers move, we keep going. If they don't move after the loading phase, we have an honest conversation about whether to adjust dose, frequency, or whether ketamine isn't the right tool for you specifically — which does happen, and is information rather than failure.

What this isn't

Ketamine is not a cure. The neuroplasticity window it opens is real, but what fills that window matters: therapy, sleep, sunlight, movement, the people you love, the things you write down at 11pm when something has shifted. Patients who do well on ketamine almost always do well partly because they meet the medicine halfway. The ones who treat it as a magic pill that should fix them while they keep doing what they were doing — they're the ones for whom the response fades quickly.

It's also not a forever treatment for most people. Many patients move from active treatment to maintenance to occasional booster sessions, and some get off it entirely. The goal isn't lifelong ketamine. The goal is enough plasticity to break out of a stuck pattern, and then a life that's worth living without it.

If you've made it this far

If you've read to here, the next step isn't a sales pitch. It's our eligibility questionnaire, which takes about five minutes and tells you honestly whether ketamine is something we should talk about further. If we're a fit, we schedule an evaluation. If we're not, I'll tell you why and, when I can, point you at what might help instead.

See also: what comes after failed antidepressants, what if ketamine doesn't work, and comparisons to TMS, Zoloft, Wellbutrin, and Spravato.

Treatment-resistant depression is exhausting in part because it makes you doubt that anything will work. I can't promise this will work for you. But I can promise you'll get a real conversation, an honest evaluation, and either a careful plan forward or a clear reason why this isn't your tool.

— Dr. Ben Soffer