“Don’t hate the playa, hate the maths”. It’s fair to say that medical statistics is not a particularly sexy subject. But if you’re being served something and asked to swallow it, it can pay to read the label…

And the quickest and easiest way to read the ‘label’ on scientific studies is to grab the figures and then to calculate the Actual Risk, Relative Risk and Numbers Needed to Treat. This is the first step on understanding what the study is saying. Don’t expect to hear Boris, Whitty or Hancock mention these terms.

The AstraZeneca paper that I talk about here can be accessed freely (side-effects in the July paper, at https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(20)31604-4/fulltext, diagnosis numbers in the November paper, https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(20)32661-1/fulltext). Note: the figures in the paper were actually 23,848 (not 20,000) but I’ve used the figures they gave in the press release for explanatory purposes. This is an explanation, not a scientific analysis.

Moderna and Pfizer are yet to release data (although, the last 24 hours, the FDA actually released their own analysis of the Pfizer figures, viewable at https://www.fda.gov/media/144245/download).