Here is a contradiction most clinicians carry without noticing: the treatment we still call "shock therapy" and hold in reserve for the sickest patients is the one associated with lower all-cause and lower suicide mortality.
Not neutral. Not merely tolerable. Protective.
The Chan et al. meta-analysis of 268 cohort studies—10,842,094 subjects with depression against nearly 2.84 billion controls—found depression carried a doubled all-cause mortality risk and a tenfold suicide-mortality risk. And in that same body of work, ECT was associated with reduced all-cause mortality, reduced natural-cause mortality, and reduced suicide mortality. Antidepressants, by contrast, moved only all-cause mortality.
Meanwhile, the risks that justified holding ECT back are, on modern data, largely failing to appear. A recent British Journal of Psychiatry review of administrative health data concludes that after adjusting for confounding, ECT does not increase the risk of dementia or major adverse cardiovascular or cerebrovascular events—and is likely associated with substantial reductions in both suicide and all-cause mortality.
So the reserve position rests on harms the newer data don't find, while ignoring a mortality benefit the data do find. That is not caution. That is inertia wearing caution's clothes.
The reserve position rests on harms the newer data don't find, while ignoring a mortality benefit the data do find. That is not caution. That is inertia wearing caution's clothes.
Last week this column argued about how we define treatment resistance—the staging criteria that decide when a patient has "failed enough." This week the question moves downstream: once resistance is established, what dogma blocks access to the treatments that actually work?
Because the burden of proof in psychiatry is applied with a strange asymmetry. We demand mountains of evidence before escalating to an effective option. We tolerate flimsy tradition as if it were settled law.
Consider how ECT is framed even by its defenders. It "remains relevant today as, arguably, a gold standard treatment for severe depression"—and yet "its use is largely restricted to depressed patients who are very severely ill, suicidal, psychotic, catatonic, and/or treatment-refractory." The gold standard is quarantined.
Part of that quarantine is real: cognitive side effects exist, and in the head-to-head trials, ECT produced more cognitive impairment and musculoskeletal events than ketamine did. Nobody serious pretends the amnesia is imaginary.
But stigma does work the evidence cannot justify. As one systematic review put it plainly, ECT faces "a general reluctance to undergo so-called 'shock therapy,' a reductive colloquialism based on the public perception of the benefits of ECT." Perception, not data, sets the default.
And access compounds the perception. In some regions there is roughly one psychiatrist qualified to perform ECT per million patients. When a treatment is both feared and physically unavailable, "last resort" stops being a clinical judgment and becomes a description of who happened to get through.
Now watch the same asymmetry operate on a drug that costs almost nothing.
Oral racemic ketamine costs less than half a US dollar per session. It has three decades of practice behind it across more than half a dozen countries, and is currently in wide use in at least one. It carries no anterograde or retrograde amnesia—the very harm patients fear most about ECT.
Yet the field's default remains intravenous racemic ketamine or intranasal esketamine. In low- and middle-income countries, IV ketamine lives in tertiary care centers in cities, and intranasal esketamine is unavailable or unaffordable outside urban areas.
Translate that into geography. A depressed, suicidal patient in a rural district has no rapid-acting option at all—not because oral ketamine doesn't work, but because "we administer ketamine by IV" hardened into a rule while nobody was checking whether the rule served the patient.
The comparison with acute pain makes the absurdity vivid. When a patient has a visible injury and is in pain, we offer immediate relief—not a drug that might work in four to six weeks. When the injury is invisible and the patient is suicidal, we default to the slow path and reserve the fast one.
Here is the crucial move: the case for oral ketamine is not that it beats IV in a trial. It's that the IV-only rule was never a finding. It was a habit that acquired the authority of a finding by being repeated.
Given oral ketamine's simplicity, as its prescriber's guide argues, "there may not be a need to reserve its use for TRD" at all. The reserve, again, justifies itself only by tradition.
Which brings us to the smallest and sharpest example: bupropion in OCD.
The textbook line is clean. Bupropion "is not a standard treatment for OCD and is generally avoided due to concerns regarding potential symptom exacerbation." Read that phrasing carefully. The stated reason to avoid it is fear that it will make OCD worse.
Now the case. A 23-year-old woman with OCD, major depression, and anxiety. Her obsessions included aversions to the smell and textures of fruit; her compulsions included repeatedly wiping surfaces with detergent, mentally replaying songs, and repeating a bedtime routine in precise order. She avoided any job that involved handling fruit.
SSRIs—sertraline, fluoxetine—were tried and stopped for adverse effects. Adjunctive N-acetylcysteine hadn't produced remission. Then, largely because her depression featured prominent amotivation and her mother had responded well to it, she was started on bupropion.
At one month: marked improvement in mood, energy, and sleep—and reduced frequency and intensity of obsessions and compulsions, no longer distressing or impairing. She tolerated fruit in occupational settings without distress. Over three years of follow-up, minimal to no OCD symptoms, no functional impairment, stable on the drug.
The exact thing the avoidance rule predicts—exacerbation—is the opposite of what happened. Her tolerance of the previously avoided stimulus improved.
Be honest about the limits: this is a single case, with patient-reported outcomes and concurrent NAC, and it "does not support routine deviation from first-line treatments." One case cannot overturn a default.
But it doesn't have to. The point is that the default was never built on much. The evidence on bupropion in OCD is, by the authors' own account, "limited and mixed." The avoidance rule is a null-evidence default—an absence of study converted into a prohibition.
That is the pattern across all three. ECT's reserve rests on harms the modern data don't confirm. Ketamine's route rests on a habit mistaken for a requirement. Bupropion's avoidance rests on a fear that limited, mixed evidence doesn't establish.
In each case, "we don't do that" is presented as if it were a finding. In each case, it is a vacancy.
We would never tolerate this asymmetry in the other direction. Propose oral ketamine in a rural clinic and you'll be asked for a mountain of evidence. Propose that the IV-only default itself be justified, and the room goes quiet—because nobody thinks defaults owe explanations.
They do. A default is a clinical decision made at scale, applied to every patient by omission. If a deviation must justify itself, so must the rule it deviates from.
And the cost of leaving the rule unexamined is not hypothetical. Depression carries a tenfold suicide-mortality risk. ECT is associated with reduced suicide mortality. The BJPsych review reportedly calls it "a life-saving treatment" whose stigma drives "hesitancy regarding its use."
Stigma, in other words, is not a soft variable. It is a clinical input with a body count—it determines who reaches the mortality-reducing treatment, who reaches the sub-half-dollar rapid option, and who reaches neither.
Which points to the outcome we keep failing to measure. We measure remission rates and effect sizes down to the standardized mean difference per day. We do not measure access—how many patients who would benefit are simply never offered the option.
So the editorial question should be inverted. The burden of proof does not attach only to innovation. It attaches to inaction, to the reserve list, to the habit that calls itself prudence.
Make the defaults defend themselves the way we make the deviations. Most of them can't.