Last week, we discussed the link between exposure and outcome.  In today’s post, we will discuss how Snow tested his communication theory. On Thursday, we will learn about the importance of accurate outbreak investigations and keeping an open, dogma free-mind when conducting them.

The Snow series is an educational course. We hope you will recognise our efforts by donating to TTE or becoming a paying subscriber, as writing the series took a lot of time and effort.

So the water intake site for the Lambeth Water Company had changed between 1848 and 1852 from Hungerford to Thames Ditton, but not that of the Southwark and Vauxhall Company, which still drew its water from Battersea Fields, one of the most polluted stretches of the river Thames (in fact the water inlet was near the sewage outlet). Much later, it was discovered that the latter company had no filtration or sedimentation tanks. We wonder whether Snow suspected so.

Now, the question was: Did the change in exposure lead to a change in cholera incidence compared to 1832 or 1848, when both companies got their water from downriver? To answer this question, Snow and his assistant, Mr Whiting, went door to door in the area in dark red on the map reproduced below. 

Pay attention because this is one of the rarest cases of natural experimentation and cause-and-effect sequences known to us.

Snow already knew through his hard work that the 1854 cholera mortality for areas supplied only by the Lambeth Water Company (in red on the map below) was one-tenth of that of the areas supplied by the Southwark and Vauxhall Company (in green - labelled as “blue”). He also knew such a difference was not visible in the 1832 and 1848 epidemic data. However, if you thought miasmata were the cause of the problem, you could not discount the geographic difference between the two areas. However, now the dark red area in “Boundaries of the Registrar-General’s Districts on the South side of the Thames and also the water supply of these districts” represented an area where both companies operated, so miasmata, if they existed, would not be very different in the dark area and could be discounted as a cause of differential mortality.

Snow described mixing water suppliers in the same street as “the mix of supplies of the most intimate kind”.

Going door to door and asking people who their water supplier was, he discovered that most respondents had no idea because agents or landlords paid water bills. He also noticed that the area was mixed with all sorts of different people living there: wealthy folk in nice houses, “tradesmen”, and “vagrants” in worse conditions so that he could discount social class and quality of housing as causes. The denominator was huge, too: 300,000 people living in the area.

In effect, the distribution of the water pipes in the streets was random, and there was no ascertainment bias because the supplier was ignorant of the water's origins.

You cannot be biased, even subconsciously, against Company X if you do not know who brings the water to your well or pump.

It appeared that only one variable could account for the difference in mortality: water supply. 

In 1854, Snow could only calculate the mortality by sub-district, not by house, as the denominator was unavailable. It was eventually made public in 1856 by Sir John Simon, the then medical officer to the Privy Council, who came to the same conclusions without citing or crediting Snow’s work, although this must have been the stimulus.
CONTEMPORARY THEMES

Taking, for example, the 2020-21 scatter-gun approach to distancing policies (see also Lecture 7), we think no one will ever be able to assemble robust evidence as to whether they impacted transmission.

Snow’s map data are partly retrospective and partly prospective. From his perch, he could reconstruct what had happened and what was unfolding before his eyes.

However, the district boundaries for the two companies were stable, and so were the numerators and denominators of the people who lived there, unlike today’s offerings, where people's travel and movement are more transitory.

Snow’s analysis had the great merit of recognising the significance of the nearly random differential exposure but similar ambient conditions.

This completely contrasts with the poor quality evidence retrospectively used by governments to justify interventions such as mask mandates.

What is even more surprising is that government ministers were unaware of how the advice given to them had been assembled: Rishi Sunak, then UK’s Chancellor of the Exchequer, told the Spectator that “whoever wrote the minutes for the Sage meetings – condensing its discussions into guidance for government – would set the policy of the nation. No one, not even cabinet members, would know how these decisions were reached.”

Further Readings