Pulmonology is the specialty that learned what everyone does is not always what works.

In May 2000, the New England Journal of Medicine published a trial about ventilators. Ventilators then gave patients with badly injured lungs large breaths, 10 to 15 ml per kilogram of body weight. Those large breaths may have stretched and damaged the lungs further. The trial gave 861 patients either 12 ml per kilogram or 6 ml per kilogram. It stopped early. Death rates were 31.0% with the small breaths and 39.8% with the large ones. The small-breath patients also spent more days off the ventilator. A gentler machine setting had saved lives.

Eleven years later, in August 2011, the field turned to lung cancer. Earlier screening had not cut deaths from it. The National Lung Screening Trial tested a new scan, low-dose CT. It enrolled 53,454 people at high risk. Each person had three yearly scans, either CT or a chest X-ray. CT found more cancers than the X-ray, 1,060 against 941. It cut deaths from lung cancer by 20.0%, and deaths from any cause by 6.7%. The price was false alarms. Across the three rounds, 24.2% of CT scans came back positive, and 96.4% of those positives were not cancer. For the first time, a screening test had cut deaths from lung cancer.

Nine months after that, in May 2012, a trial tested a treatment doctors already widely used. For idiopathic pulmonary fibrosis, doctors gave three drugs together: prednisone, azathioprine and N-acetylcysteine. Nobody knew if the combination was safe or if it worked. The trial compared it with placebo and stopped the combination early. On the combination, 8 patients died and 23 went to hospital. On placebo, 1 patient died and 7 went to hospital. The combination showed no benefit. A routine treatment had been harming the patients it was meant to help.

These are three of the thousands of studies that shaped modern pulmonology. They taught pulmonology that smaller breaths on a ventilator save lives, that a CT scan can cut deaths from lung cancer, and that a common fibrosis treatment was doing harm, and each time the answer was in a paper years before it was in a textbook.

It is still happening. Recently, in June 2025, the New England Journal of Medicine published a trial in idiopathic pulmonary fibrosis, the scarring lung disease of the 2012 paper above. The trial gave 1,177 patients a new pill, nerandomilast, or placebo for a year. Most were already on one of the two existing drugs. Over the year, lung capacity fell by 114.7 ml on the higher dose, against 183.5 ml on placebo. The cost was diarrhoea, in 41.3% against 16.0%. Serious side effects were no more common on the pill. A disease with two drugs had a third, and it worked on top of them. Practice follows the literature as it goes. It is a good culture. It has exactly one cost.