Wednesday, October 22, 2014

Was John Snow more of an empiricist than the miasmatists?

If you know anything about epidemiology, you know that the iconic Broad Street pump in the Soho district of London is the site of what is considered to have been the first modern, epidemiological study.  This is where the man remembered as the first epidemiologist, John Snow, in the first empirical study of its kind, demonstrated that cholera is a waterborne disease.  Or at least that's the legend.

John Snow
The story is well-known in the field of epidemiology, but also beautifully told in Steven Johnson's 2007 book The Ghost Map. John Snow was a physician, and even the anesthesiologist to Queen Victoria at a time when anesthesiology was just being developed, given to testing innovative ideas that hadn't yet caught on.  He had proposed during a cholera outbreak in London in 1849 that the disease wasn't due to 'miasma', or bad air, as was widely thought, but instead was caused by a contagion in the water.  So the outbreak in 1854 in Soho, near where he himself lived, became the perfect natural experiment for him to test -- or confirm -- this idea.


He gathered evidence
The first death in the 1854 epidemic occurred on August 31, and by September 10, 500 people had died.  Snow took the opportunity to collect as much data as he could.  He did an exhaustive review of the deaths, interviewing surviving family members, and drew a map (the 'ghost map') which showed that all the deaths clustered around the Broad Street water pump.  From his interviews, he determined that all those who became ill had drunk water from the well.  He confirmed his earlier reasoning that the worst symptoms were intestinal rather than respiratory, which meant to him that the agent was ingested, not inhaled.  He even found the index case, the case that began the epidemic -- a mother had dumped waste from the diaper of her infected baby near the well and, he reasoned, this had contaminated the well water.

Adding to the credibility of his theory were such findings as that there were no deaths among the 70 workers in a Broad Street brewery, because the men were given free beer all day so never drank water.  And, of the 530 inmates in the workhouse around the corner from the pump, only five contracted cholera, because the building had its own well so few inmates used the Broad St pump. He tried looking at the water under the microscope, but since he didn't know what he was looking for, he was unable to find the offending agent.

Armed with all this evidence, he was able to convince the local authorities that the Broad St pump was the source of the contaminating agent, whatever it was, that had caused the outbreak.  He urged them to remove the handle from the pump to prevent further contamination, and they did so, though grudgingly.  In fact, they recanted not long afterwards, most of the council never believing that cholera was caused by something in the water.

The removal of the handle is not what stopped the epidemic, however, as even Snow recognized.  Incidence had already begun to decline by the time the council took action.  By the end of the epidemic, Snow had at least convinced himself that his theory was correct, and he would certainly be pleased that history vindicated him.

Monster Soup commonly called Thames Water

More convincing evidence
But here's what I think is most interesting about this story.  The miasmatists had evidence, too.  An editorial in the London Times in 1849, for example, cited by Johnson, proposed five possible causes of cholera:
  • “A … theory that supposes the poison to be an emanation from the earth,” 
  • An “electric theory” based on atmospheric conditions,
  • The ozonic theory -- a deficiency of ozone in the aid,
  • “Putrescent yeast, emanations of sewers, graveyards, etc.,”
Or, most unlikely, because it "failed to include all the observed phenomena,"   

  • Cholera was spread by microscopic animalcules or fungi.
It can't be argued that the editors at the Times, indeed, miasmatists everywhere, were not empiricists. They certainly were, and they were all weighing the evidence as well.  It was well-known that most cases were in cities, where the air smelled bad, and open sewers emanated putrid smells, and rivers were filthy and smelled atrocious.  Most cases were in low-lying areas, closer to swamps and so on than hilly areas.  So, the idea that disease was due to bad air made a certain amount of sense, based on available evidence.  Correlation equaled causation.

Wellcome images

But it has to be said that the same was true of Snow's idea.  He had a lot of circumstantial evidence, but he certainly didn't have the smoking gun.  The actual immediate cause of cholera was not to be identified for several more decades, when German virologist Robert Koch (re-)discovered the causal organism, Vibrio cholerae, in 1883.  (It had first been discovered in 1854, however, by Italian anatomist Fillip Pacini, though this wasn't well-known.)  So, Snow's evidence could only in fact be confirmed with hind sight.

Vibrio cholerae

Needed: formal method for determining evidence
Koch and Louis Pasteur in the decades after Snow's study of cholera contributed to the death of the miasma theory and the rise of the germ theory of disease by discovering first the organism that causes puerperal fever in 1860, and then in the 1870's and 80's, the causes of tuberculosis, anthrax and cholera, and many more followed.  They formalized the germ theory in 1878 and 9, and this soon led to the beginning of public health in Europe and the US.

How did they know that a microbe caused disease, though?  To standardize this, Koch proposed four postulates.
1. The microorganism must be found in abundance in all organisms suffering from the disease, but should not be found in healthy organisms.

2. The microorganism must be isolated from a diseased organism and grown in pure culture.

3. The cultured microorganism should cause disease when introduced into a healthy organism.

4. The microorganism must be re-isolated from the inoculated, diseased experimental host and identified as being identical to the original specific causative agent.
But, even he knew that many microbes didn't meet these criteria; they couldn't be grown in the lab, they might be found in healthy individuals, and so on.  Molecular Koch's postulates have been proposed in the modern era, but they, too, aren't always met.  So, it seems that demonstrating causation still often can't be done conclusively.

This all lead to, or at least coincided with the use of statistical criteria in epidemiology and genetics, and the rise of population-based evidence for causation.  In the 19th century, governments began to collect data for demographic and other statistics, and insurers and actuaries began to collect data and compute group statistics to calculate the probably of future events from known data.

Philosopher, logician, mathematician C.S. Peirce is credited with inventing randomized experiments in the late 1800’s, and they then began to be used in psychology and education.   Randomized experiments were popularized in other fields by R.A. Fisher in his 1925 book, Statistical Methods for Research Workers. This book also introduced additional elements of experimental design, and was adopted by epidemiology. Austin Bradford Hill published Principles of Medical Statistics for use in epidemiology in 1937.  And, population genetics, the Modern Synthesis (which showed that Mendelian genetics is consistent with gradual evolution), and discoveries in genetics laid the foundation for approaches to looking for the genetic basis of traits and diseases.

AB Hill suggested a set of nine criteria that could be useful for determining causation in epidemiology.  Called the "Hill Criteria" they are still considered useful today, even though Hill himself wrote that none except the requirement that cause precede effect are necessary.  

Where are we now?
Despite all the progress, we still have trouble determining causation.  Does the use of antibiotics in livestock cause antibiotic resistance in humans?  There's some evidence that it does, but not every study supports this.  Thus, many animal breeders protest that the rise in antibiotic resistance is not their doing, and they should continue to be allowed to use antibiotics to promote growth.  As Maryn McKenna, author of books, magazine pieces and blog posts on antibiotic resistance, noted during a recent excellent discussion of the association of antibiotic resistance and use in animals on the radio program "What Doesn't Kill You" (here), it's true that the evidence isn't unequivocal, but as with smoking and lung cancer, it's time to declare an association and make policy from there.  That's the best we can do.

Demonstrating causation can be difficult in genetics, as well.  There are more than 2000 variants in the CFTR gene that are assumed to cause cystic fibrosis, for example, but this can lead to circular reasoning such that newly identified variants in the gene associated with CF are considered to be causal without causality being demonstrated, largely because it is very difficult to demonstrate.

Indeed, we now insist, as if this is a new discovery, on what is not-so-modestly proclaimed as 'evidence based medicine,' as if our predecessors ignored the facts.  We can say we're empirical, but so were the miasmatists. It's a matter of what we consider 'evidence.'

No comments: