Rheumatology is the specialty that learned to test the drugs it already trusted.

In March 1985, the New England Journal of Medicine published a small trial in rheumatoid arthritis. Other treatments had failed to control the patients’ arthritis. The trial gave 28 of them low-dose methotrexate or placebo for 12 weeks, then swapped the groups. On methotrexate, patients had fewer tender joints, less morning stiffness and less disease activity. The drug had side effects: raised liver enzymes in 21%, nausea in 18% and diarrhoea in 12%. The authors admitted they did not know how it worked. It worked anyway, and it became the drug rheumatology built on.

Six years later, in January 1991, a trial in lupus asked the opposite question: what happens when you stop a drug? Many patients took hydroxychloroquine, an old malaria drug, but nobody had proved it helped. The trial took 47 patients with stable lupus. It kept 25 on the drug and switched 22 to placebo, without telling them which. Over six months, 16 of the 22 on placebo flared, against 9 of the 25 who stayed on the drug. Flares were 2.5 times more likely off it. A drug taken on habit had been shown to matter.

Almost ten years after that, in November 2000, the field gained a new kind of drug. TNF is a signal molecule that drives the inflammation of rheumatoid arthritis. Infliximab is an antibody that blocks it. The trial gave 428 patients, whose arthritis stayed active on methotrexate, either infliximab or placebo on top of it for 54 weeks. On infliximab, 51.8% responded, against 17.0% on methotrexate alone. The X-rays told the bigger story. On methotrexate alone, joint damage kept growing. On infliximab, it stopped, whether or not the patient felt better. For the first time, a drug had halted the destruction of the joints.

These are three of the thousands of studies that shaped modern rheumatology. They taught rheumatology that methotrexate works even before anyone knew why, that an old malaria drug keeps lupus quiet, and that blocking one signal molecule can stop joints being destroyed, and each time the answer was in a paper years before it was in a textbook.

It is still happening. Recently, in April 2025, the New England Journal of Medicine published a trial in lupus nephritis, the kidney inflammation of lupus. The trial gave 271 adults an antibody, obinutuzumab, or placebo, on top of standard treatment. After 76 weeks, the kidneys had fully recovered in 46.4% on the antibody, against 33.1% on standard treatment alone. More of them got there on a low dose of steroids, 42.7% against 30.9%. The cost was infections: serious adverse events were more common on the antibody. A kidney disease that steroids had held back was now being treated at its source. Practice follows the literature as it goes. It is a good culture. It has exactly one cost.