The researchers leading the RECOVERY trial were told by many people they should not have included dexamethasone in the study when it launched at the start of the first wave of COVID-19, delegates at the British Thoracic Society Winter meeting heard during a symposium on the conduct of trials during the pandemic.

Yet less than three months after the first person was recruited to the study, which at the time was comparing four treatments, dexamethasone had been found to reduce death by a third in ventilated patients and a fifth in those needing oxygen.

Professor Martin Landray, co-chief investigator of the RECOVERY trial, said it had recently been estimated that the use of dexamethasone since those results were published had saved 600,000 lives globally.

“This was a beautiful result. It was announced at lunchtime, it was NHS policy by teatime and saving lives by the weekend,” he told the online conference.

But he added: “Many people told us we should not even be trialling it because suppressing the immune system was a dangerous thing to do. I have the letters.”

Prof Landray said there had been huge uncertainty about how to treat COVID-19 and there was lots of opinion but no reliable data. They designed the trial to make it as practical and simple for staff and patients who were working in very difficult circumstances, he added.

Part of that was not collecting data that was already available in clinical systems. It led to the largest ever data linkage exercise, and something that only the NHS is in a position to do. Prof Landray said this is a key aspect to be taken forward into other studies.

It had been shown clearly at the start of the pandemic that very small or observational studies could not answer key treatment questions, and results could often be the opposite to those which were expected. Notably, the first result from RECOVERY was that hydroxychloroquine showed no benefit.

“It was unlikely there was going to be a single big win,” Prof Landray said. “What was much more likely was a series of pretty moderate benefits.”

RECOVERY has now recruited 37,000 patients and most recently solved the puzzle of whether tocilizumab was a useful treatment after conflicting results from other smaller studies.

Despite six previous trials and a meta-analysis, questions remained over tocilizumab because the confidence intervals were so wide. But RECOVERY was able to confirm this month that tocilizumab significantly reduced deaths in patients with severe COVID-19 with an absolute difference of 4%.

Large randomised trials are vital “when you’re looking for these modest effects which turn out to be really important”, he concluded.