Pediatrics is the specialty that learned the cautious choice is not always the safe one.

In October 2005, the New England Journal of Medicine published a trial in newborn babies whose brains had been starved of oxygen around birth. Cooling had protected the brains of animals after this kind of injury. But nobody knew whether cooling was safe for a newborn baby. The trial cooled 102 babies to 33.5 degrees Celsius for three days, then warmed them slowly. Another 106 babies had usual care. The cooled babies had no more side effects during the cooling than the others. The researchers followed the children until they were about 18 to 22 months old. Death or a moderate or severe disability affected 44% of the cooled babies. It affected 62% of the babies given usual care. Cooling had protected the brain instead of harming it.

Four and a half years later, in May 2010, the same journal published a trial in babies born very early. These babies often need extra oxygen, but too much oxygen can damage their eyes. So more and more doctors were aiming for lower oxygen levels, to protect the eyes. Nobody knew whether that was safe. The trial included 1,316 babies born at 24 to 27 weeks of pregnancy. It kept each baby’s blood oxygen at one of two levels: 85 to 89%, or 91 to 95%. The lower level did protect the eyes. Severe eye damage affected 8.6% of the surviving babies on the lower level. It affected 17.9% of the surviving babies on the higher level. But deaths rose on the lower level, to 19.9% from 16.2%. The careful choice for the eyes had cost lives.

Almost five years later, in February 2015, the same journal published a trial about peanuts. Peanut allergy in children had doubled in ten years in Western countries. Avoiding peanut seemed the careful choice for babies at high risk of the allergy. The trial tested the opposite. It enrolled 640 babies aged 4 to 11 months with severe eczema, an egg allergy, or both. One group ate peanut regularly until the age of five. The other group avoided it. The main analysis covered 530 babies with no sign of peanut sensitivity at the start. The trial checked these children for peanut allergy at the age of five. Peanut allergy affected 13.7% of the children who had avoided peanut. It affected only 1.9% of the children who had eaten peanut. Both groups had similar rates of serious side effects. So eating peanut early prevented the allergy, and avoiding peanut did not.

These are three of the thousands of studies that shaped modern pediatrics. They taught pediatrics that cooling a newborn’s injured brain could protect it, that aiming for less oxygen to spare babies’ eyes cost lives, and that eating peanut early could prevent a peanut allergy, and each time the answer was in a paper years before it was in a textbook.

It is still happening. Recently, in September 2026, the New England Journal of Medicine published a trial in children with a high-risk form of leukaemia. Their treatment includes several cycles of highly toxic chemotherapy. The trial asked whether a newer drug could replace two of those cycles. The new drug is called blinatumomab. It directs the child’s own immune cells against the leukaemia. The trial gave 709 children either two cycles of blinatumomab or the two standard cycles of chemotherapy. An early analysis estimated that blinatumomab kept 83.0% of the children alive with no relapse or new cancer after four years. Chemotherapy kept 70.3%. Infections fell from 69.4% to 23.9%. Side effects on the nervous system rose from 3.2% to 12.0%. Replacing part of the chemotherapy had kept more children free of leukaemia. Practice follows the literature as it goes. It is a good culture. It has exactly one cost.