The story I posted about Hilda Hudson recently () seemed to resonate, so here's another bit of epidemiological history, about some ideas that you've almost certainty seen, but might not know the origins of... 1/
First, some background. There are three really important methods in epidemiology: retrospective cohorts, case-control studies, and confounding... 2/
In a 'retrospective cohort', we look back at a group of people who previously have/haven't had some specific characteristic (e.g. certain age/exposure to infection) and examine what this tells us about risk. Here's a COVID-19 example: thelancet.com/journals/lanin… 3/
But what if we can't track specific group? In this situation, we can compare people who have a condition ('cases') with similar individuals who don't ('controls') and identify characteristics that explain the difference, e.g. this key early Zika study: thelancet.com/journals/lance… 4/
The problem with analysing risk, of course, is it may be subject to 'confounding'. E.g. we may think people in certain places are more at risk, but really it's just certain age groups at risk, and location reflects age. So we need to adjust for this confounding in analysis 5/
Now you're probably thinking, "these seem like really useful ideas, I wonder who came up with them?" Turns out they were pioneered the same person: Janet Lane-Claypon, in early 20th Century: mrc.ukri.org/news/blog/tale… 6/
But that's not all. In her cohort study, Lane-Claypon wanted to know if mean values were different. Sample size was small, so she used a method developed by W.S. Gosset (aka 'Student') – the first time 'Student's t-test' was applied in epidemiology. pubmed.ncbi.nlm.nih.gov/15234929/ 7/
I first heard about this full story when @LSHTM were proposing new names to put on their building. Think we might need to find some more room... 8/8
