Finally Some Epidemiological Science

New York State’s recent study regarding statewide infection rates provided more epidemiological science than all of the Federal government’s contribution since the start of the pandemic. On April 23rd Governor Cuomo announced preliminary results of an antibody study designed to estimate the true number of people who have been infected with the coronavirus in New York state.

New York tested 3000 people state-wide using an antibody test designed to detect if a person has ever been infected with the coronavirus. The 3000 people were not chosen based on their risk of disease, nor their frontline-worker status. They were deliberately chosen to be “representative” of the general population. When you want to test for something in a large population, without testing everyone in that population, you need a method that will help you find a small population that will be “representative” of the larger population. In Epidemiology the method by which you choose who to test, or survey, is referred to your sampling frame. The New York study choose a relatively simple sampling frame, they collected specimens from 40 locations in 19 counties from people going to shopping centers and grocery stores. Epidemiologists will debate the quality of this sampling frame. For example, people going shopping are probably slightly healthier than the average person, as sicker people might be more likely to stay home and eat canned beans, or ask someone else to shop for them. But the singular beauty of the New York anti-body effort is that we actually have a study that was intended to estimate something very important to understanding the dynamic of the pandemic. Experts will undoubtedly debate the strengths and weaknesses of the study’s methodology, but for me the exciting news is that we actually have some new epidemiological evidence to discuss.

Contrast this with a Federal leadership that gives medical advice based on “hunches”. The President touting untested medical treatments like hydroxychloroquine led to inappropriate focus on the drug, caused poisoning and even a couple of deaths as people erroneously tried to follow the President’s advice. Even more ominous, was the President’s recent suggestion that “injecting disinfectant” could clean the lungs of coronavirus patients. This suggestion was made during the US coronavirus daily update.

What does the NY Study Tell us?

Though there are plenty of caveats, the study is interesting in a number of ways. First it supports the commonly held belief by public health experts that the on-going testing being performed around the country is not capturing anything like the true number of infections. In the case of New York, the estimated number of people who have had the virus based on the this anti-body study is more than 10 times higher than the number of positive coronavirus tests conducted by the state. One of the more interesting inferences drawn from the study is that as much as 21% of people living in New York City may already have had the virus.

This higher level of infection would imply that the virus spreads faster than previously estimated and it also it lends evidence to the notion that a large percentage of the population may have either limited or no symptoms.

An important corollary of the increased number of infections is that the case fatality rate (the percentage of patients who die from the coronavirus infection) may be lower than was previously assumed. Estimates from Wuhan China had put the death rate perhaps as high as 3%, more recent estimates had trended down to be more in the 1% range. The New York data suggests the death rate may be as low as .5%.

So a short summary of what this data from New York is more people get infected, fewer have symptoms, and overall a smaller percentage of those who are infected die.

Like a breath of fresh air:

One of the reasons this new data feels like a breath for fresh air, is that it illustrates how testing can and should be used to advance scientific understanding of the pandemic. The daily Coronavirus task force updates in Washington DC talk about testing ad nausium. Each day the President and his political associates brag about the number of tests performed in the US compared to the other countries. But all of this comes across as if by shear force of testing volume, the virus will be suppressed. For testing to be helpful it needs to be available early on, and it needs to used strategically as part of a larger public health plan. Saying that “Every American who wants a test can have one” is not a public health plan.

Evidence that success is not measured by the number of tests executed can be found in Taiwan. Taiwan has only conducted about 59,000 tests (2478 per million population) while the United States has conducted 4.9 million tests (14,823 per million population). But Taiwan has only had 6 coronavirus deaths compared to 50,000 in the United States. Taiwan used its testing capacity early on to support a robust containment strategy and is now reaping the benefits of early strategic planning.

Coronavirus antibody tests use a small amount of blood to test if someone has already had the virus.

Testing for Science:

In our “whats in it for me” society, there is an assumption that testing is something the individual needs or wants. However, sometimes, effective use of testing does not tell the individual anything. Though I have yet to see many details about the New York State anti-body study protocol, I would bet money that the people who were actually tested in this study, were not informed of their individual test results. The primary reason being that most anti-body tests developed so far are not considered accurate for individual care. Many of the tests allowed on the market under FDA “Emergency Use” rules have not been rigorously tested. News reports have indicated that many of these tests have high false positive and/or high false negative results. Such flaws can make the tests very dangerous for informing individual behavior. If someone has been told they already had the disease based on one of these anti-body tests, that person might assume they were now immune and proceed to visit their elderly relatives. But if they had received a false positive result, their visit to an elderly relative could turn out fatal.

So though a test with unknown false positive or false negative rates might be dangerous to guide individuals, it can still be a powerful tool for research. I am sure New York officials made an effort to vet the the particular anti-body test they used. In addition, their results provide some support that the tests were reasonably sensitive. For example, up-state areas where the regular coronavirus tests have found fewer sick patients, the anti-body test found only 3.6% of the population tested positive. While in New York City, where traditional testing has shown the highest rate of positivity, the anti-body study found a prior infection rate of 21%. If the test had a severe false positive rate of say 10%, then you would not expect any region of have a test rate below 10 percent. So the low rate in up-state of 3.6%, indicates that any false positive rate associated with the tests used would likely be well below 3.6%. If the test were to have a high false negative rate, it would not undercut the main news for the study (the high infection rate in the community) because a test with many false negatives would just mean the true measure of past infection would be even higher than what the study found. So a test that may not be ready for use to inform individual about who might be safe to go back to work, can still be used in the service of Epidemiological science to learn about the pandemic.

Return of the Herd:

Herd immunity is the notion that if a high enough percentage of the population is immune to a disease, either by vaccination, or by having already had the disease, the rate of new infections will decline and even disappear. For more discussion of herd immunity see my post on the other side of the curve.

Early on in the pandemic, epidemiologists who spoke of the eventual need to develop herd immunity, through the eventual infection of a large percentage of the population, were seen as doomsayers and pessimists. But their logic was simple, with a possible vaccine 12 to 18 months away, lacking breakthroughs in therapeutics that could prevent or cure the infection, that left only the epidemiological strategies of containment and mitigation to control the outbreak. Only a handful of nations, who were well organized and well prepared for the outbreak, were able to succeed at containment (Taiwan is perhaps the most impressive example). Most of world has had to rely heavily on “mitigation” which for the most part boils down to social distancing.

For reasons becoming increasingly clear, social distancing has costs both psychological and economic that make them poor long term (12 to 18 month) solutions. So for much of world, it is increasingly likely, that the ultimate curb on the pandemic will be some amount of herd immunity. People who presented this as a likely outcome were criticized because acceptance of this proposition entails a high cost in lives. If herd immunity requires half the nation to become infected (say 150,000,000 people) and if we assume only a 0.5% case fatality rate, that still means 750,000 deaths in the United States. It is not surprising that government planning agencies were not eager to project that kind of scenario.

The New York Anti-Body study estimates that 21% of New York City residents have already been infected with the coronavirus. This is a far higher rate than most experts anticipated. In the State of Massachusetts, state officials estimated that after the the surge in that state, somewhere between 0.7% and 2.5% would have been infected by coronavirus. This would mean that between 99.3% and 97.5% of the residents in Massachusetts would still be susceptible to infection. Given the infectiousness of the coronavirus, this leaves more than enough susceptible people to fuel a second and even a third outbreak spike in infections.

Most of the models used to understand how the pandemic will go over the next few months speak in terms of a second wave of infections either when social distancing restrictions are lowered, or perhaps in the fall when the weather grows cooler, and kids go back to school. What fuels these second waves is the high percentage of susceptible people still in the community. But when we begin to see that 21 percent of the people in New York may have already had the infection, herd immunity may start to have an impact.

The rate at which a virus spreads is measured using a concept called R naught (R0). It represents the number of new cases that will result for each current case of a disease. Most R0 estimates for coronavirus range between between 1.3 and as high as 5.7, but most estimates fall between 2.0 and 2.5. So for example, if the coronavirus has a R0 of 2.0, then each person who has coronavirus would infect 2 other people (on average). The two people who were infected by the first person would then each go on to infect 2 other people etc etc. And as you can imagine, the infection spreads pretty fast at that rate. This theoretical R0 is just that, theoretical. Many factors impact actual R0. Density of population and social customs can influence the effective R0 experienced in any given community. Social distancing works to reduce R0. And lastly the theoretical R0 also assumes the entire population is susceptible to infection. Many of the models used to predict the course of the pandemic assume that various levels of social distancing can reduce R0 by 20 to 60%. The kind of social distancing enforced by the Chinese government in Wuhan have been estimated to reduce R0 by up to 60%. While milder sets of social distancing such as banning large gatherings, keeping 6 feet parts and not shaking hands might reduce R0 by 20%. Social Distancing is all about reducing the number of human to human interactions where the virus could be transferred from one person to another. It turns out that having some of the population immune to the virus also reduces human to human interactions where the virus could be transferred from one person to another. If, as the New York Antibody study indicates, that 21% of the people in New York have already had the disease, than 21% of the people an infected New Yorker is likely to interact with is no longer susceptible to infection. For New Yorkers, R0 may already be reduced by 21%. Once R0 goes below 1 (each infection on average infects less than one additional person) the outbreak will begin to decline on its own. New York City is a long way from a level of herd immunity to suppress the virus all on its own, but at 21%, it may mean that places like New York may be able to make significant reductions in social distancing without increasing the risk of a renewed spike in infections.

It may be somewhat ironic, but given the new data provided by the New York Antibody study, it may be places that have been hit hard, like New York, where steps toward opening up may be safer than in those locations that have seen very little infection.

During your next virtual cocktail hour, drink a toast to more Epidemiological Science to fight the coronavirus!

3 thoughts on “Finally Some Epidemiological Science

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Create your website with
Get started
%d bloggers like this: