Of vaccinations and base rates
A public-health advisory, as we start world's largest vaccination program
Over 10 million Indians die each year. If we tracked 300 million Indians, over a million will die within six months. That’s over 6000 deaths per day. This cruel statistic is an example of base rates, a fancy way of saying how frequently we can expect something to happen in a large population, just like that, for no specific reason.
Why get all gory now? We have started on a path of vaccinating 300 million Indians over six months. They’ll be closely tracked, by app, near-and-dear and media. Although we’ve had vaccination programs before, none was as high profile as this one. While we’ll never see a story about 9,99,999 people who are doing fine after yesterday’s vaccination, we’ll see many about the one who isn’t. While vaccines cause nasty side effects occasionally, moving anecdotes aren’t a reliable way to establish causality or assess cost-benefit. Deadlines and TRPs make for rushed sensationalism, not robust investigation. On day 3, I already see stories emerging. If we scale up to vaccinating 2 million per day, base rates imply that over 40 of them are likely to die that very day, for reasons unconnected to vaccination. (Numbers could be higher, if cohort is biased towards elderly). That’s a lot of stories to choose from.
My idea isn’t to trivialise individual tragedy or brush aside the (small) possibility of vaccines causing problems. It is a nudge to frame stories against the backdrop of a gigantic population getting vaccinated in a short period. Bad things will happen to many who take the vaccine, just as bad things will happen to many who don’t. As negativity sells, former category will disproportionately find its way into our screens. So, reader and viewer beware. Correlation ≠ causality. Don’t panic. We’re fortunate to not have anti-vaxx tendencies as a country. It’s good to keep it that way. Keep base rates in mind. Trust scientists and specialists. They’ve done well by us so far.
Plural of anecdote is not data :)
How do we know that vaccination is successful, as we are doing fine without any vaccination as well?