Psychiatry is the specialty that learned not to give up on the patients nothing else had helped.

In September 1988, the Archives of General Psychiatry published a trial in people with schizophrenia whom no drug had helped. Each patient had already tried at least three antipsychotic drugs without success. The trial tested a drug called clozapine. It had been held back for years, because it could cause a dangerous fall in white blood cells. First, every patient took haloperidol, another antipsychotic, for six weeks. The 268 who still had not improved then took either clozapine or an older drug, chlorpromazine, for six more weeks. Neither they nor their doctors knew which. Clozapine worked far better. Of the patients on clozapine, 30% improved, against 4% on chlorpromazine. No patient had the blood problem, but the trial lasted only six weeks. So the authors advised giving clozapine only to patients whom other drugs had failed. For almost a third of those patients, clozapine was the first drug that worked.

Eighteen years later, in August 2006, the same journal published a trial in depression that had resisted treatment. Antidepressants take several weeks to work, and patients suffer while they wait. The trial asked whether a drug could work within hours. It gave 18 patients a single infusion of ketamine, an anaesthetic, and on another day a placebo, a week apart. Neither the patients nor the raters knew which was which. Within 110 minutes, the patients were less depressed after ketamine than after placebo. The day after the infusion, 71% of them had responded, and 29% were in remission. For about a third, the response lasted at least a week. Depression could lift in hours, not weeks.

Almost twelve years later, in April 2018, The Lancet published a study that asked whether antidepressants work at all, and which work best. It combined every double-blind trial it could find, published or not: 522 trials of 21 antidepressants, with 116,477 adults with depression. Every one of the 21 drugs worked better than placebo. The strongest, amitriptyline, roughly doubled the odds that a patient responded. The drugs differed more in how well patients tolerated them. Most of the trials carried some risk of bias, so the certainty of the evidence was moderate to very low. Even so, the answer pointed one way. The medicines in daily use did work.

These are three of the thousands of studies that shaped modern psychiatry. They taught psychiatry that a drug held back for its dangers could help patients no other drug had helped, that depression could lift in hours rather than weeks, and that the antidepressants in daily use really do work, and each time the answer was in a paper years before it was in a textbook.

It is still happening. Recently, in May 2026, The Lancet published a trial of a weight-loss drug in people who drink too much. Alcohol use disorder causes 5% of deaths worldwide each year, and new treatments are badly needed. Earlier studies suggested that semaglutide might reduce drinking. The trial gave 108 people with alcohol use disorder and obesity a weekly injection of semaglutide or of salt water, for 26 weeks. Everyone also had cognitive behavioural therapy. Heavy drinking days fell by 41.1 percentage points on semaglutide, against 26.4 on placebo. The side effects were mostly mild stomach upsets, and they passed. A weight-loss injection had cut heavy drinking. Practice follows the literature as it goes. It is a good culture. It has exactly one cost.