Trying to figure out CT’s coronavirus infection rate

For every one person who catches coronavirus in Connecticut, the virus will spread to three more.

To be more specific, the reproduction number (or “R naught” as epidemiologists say) in Connecticut is 2.91, according to a tool developed by Ted Cohen of the Yale School of Medicine.

“If we can drive that number below one it means we're driving the epidemic toward extinction,” Cohen said.

Cohen, an epidemiologist, developed a mathematical model that calculates the infection rate in all 50 states. He said what his tool shows is “potential trajectory” in any given location.

So the infection rate in California, using Cohen’s model, is 1.89 — significantly lower than that in Connecticut.

Even New York state’s infection rate is slightly lower than Connecticut’s, 2.967, though New York City’s is somewhat higher: 3.12.

The State Department of Health is “working with a range of assumptions for R naught, according to spokesman Av Harris, who said it “may be difficult to estimate for Connecticut.”

“We know positive tests underestimates the total number of people who are sick and we aren’t doing systematic contact investigations and testing their contacts,” he said.

Cohen said he has confidence in his numbers, but there are a few caveats. For starters, Cohen said he’s derived what he calls a “basic reproduction number,” which he said “is only meaningful at the beginning of an epidemic.”

And his formula depends on a specific metric remaining constant: The percentage of total tests that are positive. So, if there are 100 tests a day and 10 percent of those are positive, that percentage must remain constant over time for the infection rate to be accurate.

One of Cohen’s colleagues, Virginia Pitzer, has been looking at just that question. She’s determined — at least preliminarily — that the percentage of positive tests in Connecticut has been somewhat consistent.

“In Connecticut it seems as though the percent of tests that are positive is relatively stable over time,” she said. “In other states it may be going up and down in a more concerning fashion.”

There is a question, too, of data availability. One might assume that Yale epidemiologists have access to all the data they could use, Cohan and said that’s not the case. Some of the data has been hard to come by.

“We’re not set up for that sort of data sharing. That’s one of the things we’ll need to reckon with when this is done,” he said. “I don’t even know what the source of that problem is. It points to what the challenge is. I don’t even know who to talk to.”

All those caveats acknowledged, Cohen does have some confidence that his model is producing accurate results. Data from other places, and other epidemics, is lending credence to Cohen’s work.

What his formula doesn’t do is predict. The basic reproduction rate is a snapshot in time — to find out how long the epidemic may last, you need an “effective reproduction number,” which Cohen said he’s working on.

“The question of how long requires a transmission dynamic model,” he said. “We’re working on a version of this tool to make time estimates.”

As for what can make a difference, both Cohen and Pitzer said continued social distancing is key.

“Social distancing is so important right now,” Cohen said. “It is the thing that we know will work.”

Though Cohen said he hasn’t yet “seen signal” that social distancing strategies are effective, he said he expects to. Pitzer, though, is much more hopeful.

“We’ll only know how effective they are a week or two from now when we start to see slowing in the number of cases, because cases are always reflecting transmission that occurs a week or two ago,” she said. “It will depend on people’s willingness to comply and take part in social distancing measures.”

Pitzer said there’s “some indication already in just the data from the last few days that you’re starting to see a little bit of a slowdown in terms of the epidemic.”

She estimated — as have other epidemiologists that Connecticut may see a peak in infections within a week or so, but that it’s still a tough judgment call to make.

“In Connecticut, I think it’s still a little bit too hard to tell but hopefully there’s some signs of it just from the last two or three days of data,” she said.