Where are we on the curve of this epidemic?

When the City of San Francisco effectively grounded its residents on March 16, there were just 23 confirmed local cases. It was hard not to wonder if the restrictions were really necessary.

The trouble is, those 23 confirmed cases – which had become 223 by the time of this writing – are just a fraction of the whole. (I avoided saying tip of the iceberg because we don’t really know how big or small a fraction they may be.)

To have a reasonable picture of where in the trajectory of the Covid-19 epidemic we are, we need data on how it is spreading.

That is the question epidemiologists Paul Wesson, PhD, and Travis Porco, PhD, MPH, and George Rutherford, MD, are trying to answer, working with local public health officials to help them strategize their response.

Wesson’s research focuses on finding the true size of a population that isn’t readily counted. He generally focuses on groups at high risk of HIV infection, including men who have sex with men, people who inject drugs and transgender women. He’s applying the needle-in-a-haystack skills he’s developed to estimate their population sizes to estimate the prevalence of diagnosed and undiagnosed COVID-19 infections in the Bay Area.

Why do we need to know how many undiagnosed cases there are? Well, anyone infected with the new coronavirus can infect others, and those who think they just have a cold are even more likely to spread it. And only the true total number of cases will reveal the risk of hospitalization or death associated with COVID-19 infection.

If the population that had been tested for COVID-19 were representative of the whole population, one could apply the percentage of positives across the entire population (leaving wiggle room for the fact that current tests only catch about 80 percent of actual infections). But there has been no randomization or strategic population sampling at work in testing.

There’s another way to approximate how many people will be infected by a disease, using a measure of infectiousness called R0 (pronounced R naught). Say each person who gets COVID-19 infects an average of 2.2 new people. That makes 2.2 the R0. If you can identify when the first person was infected, you can calculate that they infected 2.2 new people and each of those 2.2 people infected 2.2 more. Go on calculating each new generation of infections using the R0 of 2.2 and you can map out a predicted trajectory of new cases.

But there’s a problem. We don’t know for certain when the first person in the Bay Area was infected. If you map a clear trajectory of new cases for any length of time, you could work backwards to the first case. But even the R0 for COVID-19 remains tentative. The data from various countries puts the R0 anywhere from 2 to 3.86.

Wesson’s challenge is to find as many reliable numbers as he can and then extrapolate the others until he can complete one or more of the equations that reveal the trajectory of the disease.

Stay tuned for more on this work as it progresses.