By Karina Toledo | Agência FAPESP – A study published on July 24 on the medRxiv platform (in a preprint article that has not yet been peer-reviewed) estimates the threshold for herd immunity to the novel coronavirus SARS-CoV-2 at between 10% and 20% of the population.
Simply put, the herd immunity threshold is the fraction of the population that must become immune for the virus to stop spreading when all preventive measures, such as social distancing, are lifted.
If the estimate computed for the published study is confirmed in practice, the implications will tend to be positive in two ways. The first is that there is a low risk of a devastating second wave of COVID-19 in the countries that implemented measures to prevent transmission and are now seeing a decline in the number of new cases. The second is that a city, region or country can apparently reach the herd immunity threshold even if it implements social distancing measures to prevent a collapse in health services and minimize the number of deaths from the disease.
“Our model shows that it isn’t necessary to sacrifice many lives by letting people circulate freely in order to achieve herd immunity. On the other hand, it also suggests there’s no need to make people stay at home for months on end until a vaccine is approved,” Gabriela Gomes, a Portuguese biomathematician currently at the University of Strathclyde in Scotland (UK) and the last author of the study, told Agência FAPESP.
The mathematical model mentioned by Gomes was developed in collaboration with scientists affiliated with institutions in Brazil, Portugal and elsewhere in the UK. The other authors of the study include Marcelo Urbano Ferreira, a professor at the University of São Paulo’s Biomedical Sciences Institute (ICB-USP) in Brazil, and Rodrigo Corder, a PhD candidate whose thesis advisor is Ferreira.
“We’ve worked with Gabriela Gomes for some years using this approach to describe the dynamics of malaria transmission in the Brazilian Amazon, with FAPESP’s support. She’s also conducted studies on tuberculosis,” Ferreira said. “The model we use differs from the rest by taking into account the fact that the risk of catching a disease varies from one person to another.”
As Gomes explained, the factors that influence the risk of catching COVID-19, for example, can be divided into two categories. One consists of biological factors, such as genetics, diet and immunity, and the other consists of behavioral factors that determine how many people we each come into contact with in our day-to-day lives.
“This has to do with people’s jobs, where they live, how they commute, and even their personality,” she said. “Someone who prefers to stay at home and read a book is less at risk of exposure to the virus than someone who goes out a lot and has a lot of relationships.”
According to Gomes, models that estimate the coronavirus herd immunity threshold at between 50% and 70% consider an identical risk of infection for all individuals. “We’ve seen that in the case of COVID-19, the more heterogeneous the population, the lower the group immunity threshold,” she said.
Calculation methods and public policy
It would not be feasible to measure the factors that influence susceptibility to the disease in every member of a population to calculate the “coefficient of variation in individual susceptibility”, a key parameter in the model described in the article. For this reason, the researchers opted for the opposite approach.
“We knew changing the coefficient of variation influences the epidemic curve projected by the model, so we decided to do the opposite, using the epidemic curve for countries where the epidemic had reached an advanced stage to calculate the coefficient of variation,” Gomes explained.
The most recent version of the study is based on incidence data (new cases per day) for Belgium, England, Portugal and Spain. “Soon we will analyze the data for Brazil and the United States, where the epidemic is still spreading fast,” Gomes said.
According to the authors, although the coefficient of variation is different in each country, generally speaking, the herd immunity threshold always tends to fall in the 10%-20% range, and this is highly relevant to public policy formulation.
“In places where the herd immunity threshold has already been crossed, the number of new cases tends to fall even if the economy is reopened,” Corder said. “However, if distancing measures are eased before the herd immunity threshold is reached, the number of new cases will probably turn up again, so the authorities have to stay on their toes. Conceptually speaking, transmission continues after herd immunity is reached if control measures are rapidly lifted.”
Two distinct situations can be observed in Portugal, Gomes said. The north, where the virus entered the country, was far more affected at the start of the pandemic, and the number of new cases is now falling there, even with the economy reopened, while the number of new cases continues to rise in the south, where the capital Lisbon is located.
“Right now, these are localized outbreaks in parts of Lisbon, and they’re being contained locally by testing and isolation of infected people,” Gomes said. “People have been allowed to go back to work in Portugal only if they test negative.”
The situation in Brazil is comparable. In the North, the epidemic curve appears to have peaked in May in Manaus and the environs, where the health system was overwhelmed. The number of new cases has since trended down despite the reopening of the economy and schools. Serological surveys show that in cities such as Manaus and Belém (capital of the state of Pará, also in the North), over 10% of the population has antibodies against SARS-CoV-2. In the South, where the number of reported cases was initially low and only approximately 1% had antibodies in May, transmission has risen in step with economic reopening. In contrast with Portugal, in Brazil, investment in testing, tracking and contact tracing remains well below the level considered ideal.
As the authors of the article stress, the fact that the herd immunity threshold may be lower than expected does not diminish the importance of public health measures to contain transmission of the virus and reduce the number of deaths.
“Any experts or authorities who advocate targeting herd immunity as a matter of public policy are wrong,” Corder said. “Control measures are important to avoid overburdening health services. In any event, the new understanding of the dynamics of COVID-19 transmission offered by our model points to a more optimistic scenario.”
For Gomes, adherence to isolation measures tends to increase if people know the sacrifice will be required for a shorter time. “When we say the epidemic will only be surmounted when there’s a vaccine, people start thinking about disrespecting the rules because they can’t bear to go on leading such unsociable lives with so many restrictions,” she said.
The best way to make the simulations and projections more realistic is to feed real-world data into the model. With this in mind, Ferreira plans to conduct a field survey in Acre, a state in the North region of Brazil, to test two assumptions used by the group in their calculations: the difference between the actual number of cases and the number of cases diagnosed (the “disease detection rate”) and the duration of infection-acquired immunity against SARS-CoV-2.
“In the study just published, we assumed that around 10% of actual cases are detected by health services and that immunity against the virus lasts at least 12 months. Now, we mean to see whether this is confirmed in a population we’ve been tracking for several years in a town called Mâncio Lima,” Ferreira said.
Mâncio Lima is a small town in Acre on Brazil’s border with Peru. Ferreira and his group at ICB-USP have conducted household surveys of a sample of the town’s population every six months. They call on 2,000 homes, apply questionnaires, and collect blood samples. Currently, the idea is to assess seroprevalence and seroconversion to estimate how many people have antibodies against coronavirus and how long they remain immune. The research is supported by FAPESP (read more at: agencia.fapesp.br/32956/).