Handwashing, Germs and Statistics
“I have felt there was something more congenial and profitable to do than to compile statistics.” Joseph Lister
According to the WHO, “Washing your hands with soap and water or using alcohol-based hand rub kills viruses that may be on your hands”. Today, the vast majority of people believe this to be true and practice handwashing to avoid infection by the Corona virus.
Of course, it has not always been like that. The idea that washing your hands reduces the risk of infection, was first (*) proposed by Joseph Lister, in his seminal paper in the Lancet in 1867. Lister suggested that surgeons should wash their hands as well as instruments prior to performing surgery on patients, but his proposal was contested and met with skepticism during a long period of time. His opponents claimed that the whole thing was non-scientific lacking theoretical ground and demanded scientific arguments and statistics. In reality, however, professional pride and resistance to new ideas, could have played a certain role. And to be fair, around 1860, the existence of invisible, disease causing germs was still an idea advocated by a handful scientists like the French chemist Louis Pasteur. Moreover, Lister himself did not believe that statistics could add anything to the matter and had the attitude that “anything can be proven by statistics”. His reluctance to publish any “data” is explained by himself in the following quote from an article by Thomas Chlich. “The truth is, that life is short, and when every day begins, one has to consider what is the occupation which is most likely to be valuable. … I have felt there was something more congenial and profitable to do than to compile statistics”.
Eventually, Lister did publish some data in a paper from 1870: out of 40 subjects who were treated under the new cleaner conditions, 34 patients survived (85%) surgery while only 19 out of 35 patients treated using the old (control) group survived (54%). Today this type of data can be presented as a contingency table and analyzed in a variety of simple ways. The most obvious, albeit not the best, way is perhaps to compare the death rates in the two “arms” of the study directly. The corresponding absolute risk reduction is thus 0.46 - 0.15 = 0.31 giving a number needed to treat NNT of 3.22. Another approach is testing for association between survival and hand washing by performing Pearson’s chi square test or Fisher’s exact test. Applied to Lister’s data, these tests yield the p-values 0.003561 and 0.005018 respectively. Logistic regression provides a model based third alternative method where interesting covariates could be accounted for in the analysis. Of course, none of these was available in 1867 since Karl Pearson’s paper on the chi-2 test first appeared in the year 1900 and Fisher’s exact test article is from the year 1922.
By the time of the eruption of the Spanish flu in the united states, the idea of handwashing as a way to fight against the pandemic was more accepted. In November 1918, ads appeared in newspapers explaining to people exactly why washing their hands was important. Today, almost 100 years later, the use of statistical methods in what is called evidence-based medicine is universally accepted and Biostatistics has developed into an important branch of statistics.
(*) More about Joseph Lister can be found in a paper by Töhler from 2014. Interestingly, Steven Dugan reports in a blog about flu pandemics and handwashing that in 1846, the Hungarian doctor Ignaz Semmelweis had suspected that germs played a part in causing diseases, but that he had no proof. He instructed his staff at the Vienna General Hospital to wash their hands and instruments with soap and chlorinated water before delivering babies. As a result, the death rate of the mothers dropped dramatically, from 20% to 2%. Still, the medical community could not believe that handwashing so dramatically slowed the spread of germs and lowered death rates. Most doctors were offended by his claim.