I
The idea of a ‘natural experiment’ has an odd conceptual history in public health. Simmelweis used Vienna’s arbitrary segregation of antenatal clinics in 1840 to uncover the role of handwashing in puerperal fever, and Snow’s evidence on water and cholera emerged from the haphazard mapping of water to London households and the fortuitous timing of a change in one supplier’s intake. But neither Simmelweis nor Snow used the term ‘natural experiment’ to describe their work,1 though Snow compellingly argued that his design was comparable to an experiment. In the 19th and early 20th centuries, the term natural experiment was used to describe rare situations that facilitated understanding the natural history of a disease, for example, head trauma accidents as ‘natural experiments’ to understand the ‘laws of mental growth and decay’.2 In the postwar years epidemiologists used ‘natural experiment’ to describe situations of a ‘naturally occurring…