Right, With the Data, and Dismissed

The Vienna General Hospital in the 1840s had two maternity wards. Women were admitted to them on alternating days, which is close to random assignment and makes the comparison unusually clean.

The first ward was staffed by doctors and medical students. Maternal mortality from childbed fever ran around 10 to 18%. The second was staffed by midwives, and mortality ran around 2 to 4%.

The difference was well known. Women begged to be admitted to the second ward, and some gave birth in the street rather than enter the first, which turned out to be a rational choice; street births had better outcomes.

Ignaz Semmelweis, an assistant in the first ward, worked through the possible explanations methodically and rejected them: overcrowding, climate, the position women gave birth in, the priest walking through with a bell. Then a colleague died after cutting himself during an autopsy, with symptoms indistinguishable from childbed fever.

The doctors came to the ward from dissections. The midwives did not.

He required staff to wash their hands in chlorinated lime, and mortality in the first ward fell to around 2%, matching the midwives. He had a controlled comparison, a plausible mechanism, and a dramatic result.

He was dismissed from his post. His findings were rejected across Europe, he became increasingly angry and intemperate (writing open letters calling obstetricians murderers, which did not help) and in 1865 he was committed to an asylum, where he was beaten by guards and died of an infected wound at forty-seven.

It is tempting to read this as a story about stupidity, and the reasons for the rejection are more instructive than that.

Semmelweis could offer no mechanism. He spoke of cadaverous particles, which meant nothing within the medical theory of the time. That theory was miasma: disease arises from foul air produced by decay, which is why it clusters around swamps, sewers, crowding and filth.

Miasma theory was not stupid either. The correlation it was built on is real (disease genuinely does cluster in those places) and it motivated the great sanitation programmes of the nineteenth century, which saved enormous numbers of lives. Being right about what to do while wrong about why is a recurring pattern in medicine, and it is why a working intervention is not evidence that the theory behind it is correct.

The theory also made handwashing look pointless. The doctors' hands did not smell of decay, and the ward was ventilated. Smell is a proxy for bacterial contamination, and it is usually a decent one, and it failed precisely in the case that mattered, which is the general hazard with proxies.

And the implication was that respectable physicians had been killing their patients, which is not a proposition anybody adopts eagerly.

John Snow's cholera work a few years later shows what the era could achieve without a mechanism. In 1854 he plotted cholera deaths on a map of Soho and found them clustered around a single water pump in Broad Street. The anomalies confirmed rather than contradicted it: a nearby workhouse with its own well had almost no cases, and a brewery whose workers drank beer rather than water had none at all.

He persuaded the parish to remove the pump handle. He could not say what was in the water (Vibrio cholerae was not identified for another thirty years) and the epidemiology was decisive without it.

Germ theory itself took decades to be accepted, and the resistance was not confined to the ignorant. Pasteur's work on fermentation and spoilage in the 1860s was strong, and Lister's antiseptic surgery, published in 1867, was still being argued about in the 1880s, partly because carbolic acid was unpleasant and laborious to use, partly because results varied with how conscientiously it was applied, and partly because it implied that surgeons had been the vector.

What eventually settled it was Robert Koch's postulates. To establish that a specific organism causes a specific disease: find it in every case, isolate it in pure culture, produce the disease by introducing it to a healthy host, and recover it from that host. Applied to anthrax and then tuberculosis, this converted a plausible general theory into a demonstration about particular diseases, one at a time.

Koch knew his own criteria were too strict. Asymptomatic carriers break the first postulate, some organisms cannot be cultured, and ethics forbids the third in humans. The framework's logic (that causation requires more than association, and that the more must be specified in advance) is what survived, and it underlies causal inference in infectious disease to this day.