In 1927, a Polish physician named Manfred Sakel was treating heroin addicts in Vienna. He noticed that overdoses of insulin — which crashed blood sugar — sometimes seemed to calm psychotic episodes. He developed insulin coma therapy: deliberately injecting enough insulin to put a patient into hypoglycemic shock, then reviving them with intravenous glucose 15 to 60 minutes later. Patients went through this six days a week for two months.

By the late 1930s, insulin coma therapy was the standard treatment for schizophrenia in the West. Hospitals in Britain, the US, and Germany ran dedicated insulin wards. Some treated 60 patients a day. The mortality rate was 1 to 5%. Many patients suffered permanent brain damage. None of it had been tested in a controlled trial.

Insulin coma was used alongside other practices we now consider barbaric. Lobotomy — severing the connections in the prefrontal cortex — won its developer the Nobel Prize in 1949. Malarial therapy infected syphilis patients with malaria, hoping fever would kill the bacteria. Electroshock therapy, performed without anesthesia, often broke patients' bones from the convulsions.

The first proper randomized trial of insulin coma therapy wasn't run until 1953. It compared insulin treatment with simply giving patients a barbiturate-induced sleep. The results: no difference. The treatment had been doing nothing useful for two decades.

Insulin coma was abandoned in the 1960s as the antipsychotic chlorpromazine became widely available. Lobotomy faded. Electroshock continued in modified, anesthetized form. The first generation of effective psychiatric drugs arrived.

The history of psychiatry is a reminder of how confident medicine can be while being completely wrong.