Epidemiology, the statistical study of population health instead of a single person's health, has taken population-wide sampling to a new level—underground.
Instead of collecting data on individuals representing a subset of the population and then averaging it, the researchers let an existing piece of infrastructure do the averaging for them. This also made sure they were actually getting properly anonymized data from every single resident of the study area—because the area of study was "everybody connected to the sewer", and everybody who has one, uses the toilet.
What they were testing, specifically, was the percentage of people who took their medicine, by having the entire city collectively pee in a cup.
To test this, they had to know a number of things:
- how much medicine was prescribed per person
- how many people were prescribed the medicine (as a percentage of population)
- what the medicine looked like after it had been used by the body and excreted
- how much wastewater was produced per person per day (flush volume, that is)
- how much rainfall contributed to the combined stream on sampling day (dilution)
From this, they could figure out how much of the used medicine should have been in the sewers if everybody took their medicine exactly as prescribed, and compare that to the test results representing how much was actually in the sewers. They also compared their results to the results of other studies, which used surveys.
Provided their assumptions, calculations, and test protocols are accurate, this is easily the most unbiased way to find out on a population level whether a prescription is being used as directed. Surveys are biased to reflect the demographic of people who answer surveys, or (in the case of school populations) people who go to a particular school. Depending on the school, those populations could be heavily biased, for example by family income.
This is a rather different take on sewage than the first epidemiology study.
Of course, at the time of that first study they didn't have the analytical capacity to test the concentration of something in a liquid, but a certain nurse named Florence Nightingale, who I've talked about before, was both a nurse and a statistician and basically invented epidemiology. She kept statistics on deaths and showed a clear and dramatic drop in mortality once the injured soldiers were no longer exposed to raw sewage. (This is obvious to us now, but in the decade prior to Pasteur proving the germ theory of disease, most people had not yet made the connection between sewage and disease transmission.)