How close are we to herd immunity as we enter the new year?
And, how do scientists know?
Herd immunity is a form of indirect protection from an infectious disease. It occurs when a sufficient percentage of a population becomes immune, whether through previous infection or vaccination, thereby reducing the likelihood of infection for others lacking immunity. For COVID-19, scientists believe herd immunity will occur when about 70% of the population is immune to the SARS-COV-2 virus.
According to the website covid19-projections.com, there were somewhere between 47 and 106 million people infected with the SARS-COV-2 virus in the United States during 2020. This was largely before mass vaccinations. A best guess is that about 78 million people were infected by the end of December. Half or more of those people never knew they were infected because their symptoms were mild, like the common cold, or even nonexistent. But untold millions suffered from COVID-19 in the US and at least 337,000 died in 2020, likely many more. Time will tell how COVID-19, and its economic fallout, will affect American culture in the long-term.
The website covid19-projections.com was created by an independent data scientist, Youyang Gu to get a better handle on the pandemic. His earlier work, which proved quite accurate, as recognized by many (and according to Gu himself), projected the number of deaths from COVID-19 that would occur in the near future. Gu has now switched his focus to (1) rapid real-time reporting of infections and (2) estimating the path to herd immunity. Gu uses COVID-19 parameters now widely accepted by most epidemiologists (such as the average incubation period, infectious period, time until death, etc.). He updates his model daily using The COVID Tracking Project for the number of confirmed cases of COVID-19 and for testing positivity rates (percent of tests that came back positive out of all tests conducted). For the number of vaccinations, he uses CDC’s Vaccination Data Tracker). Case information is entered down to the county level using Johns Hopkins CSSE case ascertainment.
About infections, Gu says:
“We are nowcasting (what has happened/is happening) rather than forecasting (what will happen). See our methodology writeup, Estimating True Infections Revisited.”
Given 330 million people in the US, Gu’s nowcasting model estimates 23% (~78/330) of the population was infected with the SARS-COV-2 virus during 2020, or nearly a quarter of our nation. Infection rates varied by state, with Hawaii and Vermont on the low end, with an estimated 4–5% seroprevalence, and South Dakota at the high end, with an estimated 40% seroprevalence.
Evidence of prior infection (seropositivity) can be measured more directly by testing people’s blood for antibodies to the virus. In spring and summer of 2020, researchers did just that. They conducted studies in various locations enlisting representative groups of people to give samples of their blood. The researchers found seroprevalence of antibodies to SARS-COV-2 to be many times higher than the incidence of medically confirmed cases of COVID-19 in the area studied. This was expected due to the often mild or asymptomatic presentation of COVID-19. But, importantly, testing capacity in the early days of the pandemic was sorely lacking which led to a further undercount of infections.
Researchers then, as now, found that infection rates varied considerably by location. Meta studies like this and this suggested an average of 6 times the number of infections to the number of confirmed cases in spring and summer of 2020. But, these early antibody studies were criticized due to inherent problems of ascertainment bias (non-representativeness of study subjects) and inaccuracy of the tests (too many false positive and false negative results).
Data scientists, like Gu, now concur that the number of infections in a population can be estimated reasonably accurately—and a lot faster—without the need to draw blood. Here are Gu’s latest estimates of US infections through December 31 (he does not report the most recent two weeks due to an inherent lag in ascertainment):
Last Updated: Friday, January 15, 2021:
Total Infected (as of Dec 31): 77.6 million people (23.4%, or 1 in 4 people)
Adjusted Test Positivity Rate = 11.4% (as of Jan 14, out of all tests conducted, 11.4% were positive)
Infection-to-Case Ratio (as of Jan 14): 3.0 (33% detection rate on Jan 14)Rt (as of Dec 31): 1.02
Rt is the current transmissibility of the disease. With an Rt=1, each infected person, on average, infects one other person. Rt is based on the initial R0 (underlying transmissibility in a standard population without any mitigation) and further shaped by what we do and other factors. With Rt=1 the disease curve is flat and the incidence of new infections stays about the same in the population. This appeared to be the case for much of December.
Gu’s infection estimates averaged about 630,000 per day in December, considerably larger than the reported number of medically confirmed cases, which hovered around 200,000 per day through December. (To see daily age-adjusted rates for selected states go here.)
Despite fluctuations up and down, on average, both infection and case curves appeared somewhat flat in December. This gave a small bit of hope that the disease might not spike in the new year while we wait for a vaccine. That is, our mitigation efforts made at the height of the holidays, Thanksgiving through Christmas, might continue or even improve. But, unfortunately, incidence of new disease appears to be climbing in January, as shown in the above graph of new cases.
Whether flat or climbing, it shouldn’t take long now to get to herd immunity through infection alone, that is, to get to the threshold that scientists say non-immune people can be reasonably protected by the people who are immune—or 70% immunity in the population. In the meantime, at the current rate of infection, tens of thousands, even hundreds of thousands, could die. The apparent uptick in January suggests that the situation is becoming increasingly dire.
Of course everything changes with vaccinations. Gu is now tracking vaccinations, which began its first mass rollout on December 15, 2020. You can see the gradual beginnings in the first graph in this article (note the turquoise and indigo dotted lines).
The graph below is Gu’s projection of our path to herd immunity, as of January 15, 2021.
“We estimate COVID-19 herd immunity (>70% of population immune) will be reached in the US during summer 2021 (Jun-Aug 2021) … roughly half of the immunity will be achieved through natural infection, and the other half will be achieved through vaccination.”
Gu’s model assumes that states will continue to impose varying degrees of restrictions and interventions until herd immunity is reached. His model also considers only the rollout of two vaccines, Pfizer and Moderna. Other vaccines will be added to the model as they begin their rollout. A list of other assumptions in Gu’s model can be found here.
Of course, Gu’s path to herd immunity is a projection based on what is currently known and according to a model created by a single source. Other groups are also creating models and should be explored as well. How much or how quickly we get to herd immunity through vaccination will be determined by the efficiency of the rollout (the supply) and the acceptance by the public (the demand), and, of course, by what we do. How much longer natural infection, and thus continued illness and death, will outweigh vaccination protection, at least in the short and mid-term, will be determined by how much we abandon mitigation efforts, such as mask wearing, hand washing, social distancing, etc.
As long as the disease transmission rate stays above Rt=1 we will continue to see infection, illness and death at the levels we are now experiencing, or greater. Stay tuned to Youyang Gu’s website for easy-to-understand estimates and predictions as we enter 2021.
