Seroprevalence! Wassat?

Quantifying the reach of an epidemic

Today’s little discussion is brought to us by the number 8 and the word “seroprevalence”. I don’t think Sesame Street got that far, but it’s interesting.

We ALL by now know that one of the quirky things about COVID is asymptomatic transmission: people get it, but don’t know they have it, and pass it on anyway. And some people get it, but don’t suffer much, recover and get on with life. How many people have got it beyond the numbers which have tested positive?

Seroprevalence is a key word used by epidemiologists (scientists who study epidemics for a living). What it means is: how much of a population HAS caught the disease, how far the disease has propagated into the population - separate from the number that TESTED for it.

Quantifying that is difficult, without lots of testing and, as we all know, the USA at first dropped the ball on testing. Despite wailing from on high, the USA today is routinely performing more tests per 100k of population than any other country: in Marin County, about 19% of the population has been tested. California has carried out nearly 900,000 tests in the past week, and nationwide there are more than 2 tests per thousand people per day - nearly 400,000 per day in total.

116253769_10223222906581571_776983233649012758_n.jpg

But as many people as are being tested, there are some people who contract COVID and don't show up in the numbers. (Perhaps because they don't know they've got it - the symptoms are mild - or because they DO know they've got it, alongside a family member or neighbor who was tested.)

The huge increase in testing in the USA means scientists now have some idea of how far the disease has spread beyond confirmed cases. The lowest credible ratios I can see between confirmed and total cases are 2x to 4x. Some renowned scientists believe the ratio to be as high as 7x or 8x. Applied to Marin that'd mean perhaps 8% (one in 12) of the county has contracted it - perhaps as many as 20% in central San Rafael. Simple summary:

For California there may be more than eight times as many people in California who have contracted COVID as have tested positive for it. So, over 9% instead of 1.1%.

Is that good or bad news? How do we know about people we haven’t tested? Good questions:

Is it good or bad? Yes. Both.

Suppose that eight times as many people have contracted COVID as have been tested. That’d mean that 8% of the county’s population has - or has had it. If (and this is probably a bad assumption) that 8:1 ratio applied uniformly throughout Marin County, then nearly 22% - nearly one in four people - in San Rafael (94901) would have been affected, but just 4% - one in 25 - in Southern Marin. Bad news, absent 100% mask compliance and distancing and hygiene, it’s easy - and getting easier - to get this disease.

The good news - obviously - is that the actual fatality rate is lowered. But don’t be fooled: it’s still far worse than any other contagion to have shown up in the USA in ages. 150k deaths on 4.4 million confirmed cases and perhaps 32 million total cases across the USA is still ghastly, whether it’s the 3% of confirmed cases or the 0.5% of all cases. (Flu fatality is typically 0.1% on confirmed cases, far less again versus total cases, and its spread is limited by vaccines.)

Also good news is that the chance that a random person in southern Marin is COVID-positive is quite small.

However, the worst news is that we’ve a long, long way to go before there’s any form of herd immunity. Herd immunity only occurs when something like 70% of a population cannot transmit the disease. They can’t transmit it because they have immunity from having had it - or because they’ve been immunized by vaccine. When lots of people in a crowd are immune, the disease cannot propagate.

In contrast: southern San Rafael might be about one third of the way to immunity. In principle. In theory. And in the absence of a vaccine.

Note: the journey to full immunity without a vaccine is a journey in which at least seven times as many in the county would have to catch it. You can imagine what that means in terms of illness, death and grief.

How do we know?

I assume that everyone now knows someone who has tested positive for COVID. There was a cluster on one of the docks where four people were sick with identical symptoms and one tested positive: the others weren’t tested. The County has no (official) knowledge of this. But there are ways that allow researchers to get increasingly accurate estimates of who is not being counted otherwise:

  • Test lots of people: there was a hot spot in SF’s Mission District, and a big push to test LOTS of people, without regard to any known exposure. This suggested, back in late May, that at least 80% of the people in that area who had contracted COVID had not previously been tested. This is the best approach to getting an accurate read on seroprevalence, but it assumes widespread and repeated testing availability.

  • Following local test surges in hot spots around the Bay Area, UCSF’s Monica Gandhi said: If we did a mass testing campaign on 300 million Americans right now, I think the rate of asymptomatic infection would be somewhere between 50% and 80% of cases. Millions of people may be spreading the virus without knowing it.”

  • Test lots of blood: not as much as before, perhaps, but lots of people have blood drawn for lots of reasons. Some samples are additionally being tested (I assume there is strict compliance with privacy regulations) for COVID. A University of Minnesota study based on this approach early on found twice as many people had COVID as had tested positive. This methodology has a systemic flaw: it only tests people who are already having blood drawn for something and are in a healthcare system.

  • Mathematical modelling: as we learn more about the disease, particularly with data on how people get it and recover, scientists can make reasonable estimates of what fractions of a population will be asymptomatic compared to those who get tested and are found positive.

As of this week, somewhat over 450,000 Californians have tested positive: that’s 1.1% of the state’s population of 39.5 million. The estimate of seroprevalence in California that has won at least some highly credible support is that of YouYang Gu of MIT. The published estimate there is that 9.5% of Californians are infected, whether they know it or not, and whether they’ve been tested or not. So, about 8 times as many people are infected as have tested positive for it. Nationwide, the same source estimates that slightly over 10% of the entire country’s population is or has been infected. Thus, instead of 4.6 million (the known number of positive cases), there could be 32 or 36 million infected to date.
Remember: this is good and bad news.

116156854_10223222596453818_3047640589983844211_o.jpg