Poop Detection! Early Warning for COVID in Fecal Samples?

Last updated June 15th 2020, 1:09:26am

As the world begins to re-open, detecting new outbreaks has rapidly become a major policy challenge. If the disease starts to take off again, policymakers will want to fall back and return to a less-open state of affairs. The problem is, most of the obvious ways we’d know things are going badly would alert us rather late. By the time you start to see an increase in deaths or hospitalizations, the disease has already spread dramatically. Detecting cases through testing is better but, again, by the time someone seeks testing they are already sick and have likely already infected many others.

The best approach would be to be testing everyone all the time, but at the moment we neither have the capacity nor the desire to do that.

So we (along with others) were excited to see this paper on using sewage as a leading indicator for a SARS-CoV-2 outbreak. Some of us were also excited at the opportunity to explain it to our five-year-olds.

What’s the Basic Idea?

People who have COVID-19 shed the SARS-CoV-2 virus in their feces. Since viral replication — and shedding — typically begins before people feel sick (and certainly before they end up in the hospital) it is possible that you could detect the virus in poop before you’d see it show up in cases and hospitalizations.

Doing this on an individual scale is probably not feasible — if we cannot get around to testing people with a nasal swab on a regular basis, it is hopeless to imagine asking them to reliably collect their poop. But at a population level, we might be able to use this more effectively. That is, if you could test a whole city’s poop at the same time, you might be able to detect that the virus is growing just a few days earlier.

There is some precedent for this type of fecal surveillance. It is used for polio screening, for example. In addition, the general idea of group or batch testing (where samples are pooled and then tested) is something that has been discussed in the context of non-fecal COVID-19 viral testing. Fecal screening is kind of just like really large scale group testing.

What Do They Do?

These researchers worked with a sewage treatment plant near New Haven, CT to collect sewage and test it. They compared the amount of SARS-CoV-2 virus in the sewage to hospital admission and case counts in the area.

What Do They Find?

The core fact: poop is a leading indicator of the virus. The first plot below shows virus counts (red) and hospitalizations (grey) by day. The lines are smoothed versions of the dots. In the second plot we see only the lines. The two series definitely seem to move together and in the second plot you can see that the fecal virus increases just a bit before the hospitalizations. In fact, it’s about 3 days. (For the eagle eye reader: the scales change between the graphs since you need less range to look only at the smoothed lines).

Interestingly, if you look at case counts rather than hospitalizations, the fecal DNA leads by about a week. This would give you a lot more time to react quickly.

The researchers also take on the question of how highly correlated these series are. Correlations are measured from 0 to 1, where a value of 1 indicates perfect correlation (meaning the increase in concentration of virus particles in feces scales perfectly with the increase in COVID-related hospital admissions) and a value of 0 signifies no correlation (meaning there is no relationship between the concentration of virus particles in feces and the number of COVID-related hospital admissions). If you look at the lines only — that is, if you correlate the smoothed data — the correlations are really, really high. The paper reports these smoothed lines have a correlation of 0.99.

Comments on the paper have pointed out that at the day to day level — that is, if you look at the correlation in the dots rather than the lines — it’s lower, more like 0.5. But this is still really high! Bottom line: virus in poop moves first.

How is this Useful?

Fecal sampling is unlikely to be our major viral detection tool. However, as an addition to the slate of possible detection methods, it’s possibly really valuable. Most cities have sewage treatment facilities, and could in principle do something like this. Relative to reaching the whole population and doing medical testing on them, collecting some poop from a vat is much, much easier.

This may also have implications for non-municipal settings. Some universities, for example, may imagine using this type of fecal testing for dorm-wide surveillance. Large apartment buildings could do so too. The possibilities are endless.