Take a look at this article in The Times. It tells the story of Oxford professors named Martin Landray and Peter Horby.
When they saw the pandemic unfolding in China and Lombardy, their first thought was to design a proper clinical trial to determine what works for COVID19.
Normally, it takes a year or more to design a medical trial, but they got it done in weeks. This process is super important. You have to write a design first and decide what you will measure. Here is the protocol.
There were two keys to success: one was that the trial had to be simple and the other was that it needed buy-in from hospitals and doctors. Buy-in comes easier in the UK because they have the NHS–a nationalized healthcare system.
The idea behind the RECOVERY trial is that COVID19 is a novel disease, and doctors don’t know what works. The fancy term for this is equipoise.
In the US and other countries, doctors often “throw the kitchen sink” at patients. The problem there is that you can’t figure out what works.
In the UK doctors enrolled patients in the RECOVERY trial. This meant the choice to use a drug like dexamethasone was not up to the doctor but was random.
Some patients got placebo, some got the steroid. Randomization is crucial as it usually balances the groups of patients and allows us to infer causation.
It turns out that the inexpensive and commonly-used drug was beneficial. In a NEJM paper, the authors reported a 17% lower rate of death in the group that got the drug.
In absolute terms the reduction was 2.8%. A rough way to think about that is for every 35 patients treated, one life is saved. Or this: for every one million patients with COVID19 treated with this simple drug, 28,000 lives are saved.
Readers of this blog and my podcast This Week in Cardiology know how much I believe in the randomized clinical trial. I’ve even said that the the RCT may be the most important medical development of our lifetime.
One hope I have is that COVID19 will help develop the infrastructure needed to do more clinical trials here in the US.
RCTs are not perfect, but they are the best way to know what works and what does not work.