The virus itself led to the decline in new cases and deaths
Manaus experienced exceptionally high infection rates (66%) by August, suggesting herd immunity significantly contained the epidemic's final size in the Amazon capital. São Paulo maintained much lower seroprevalence (22%) with flatter epidemic curve, likely due to sustained social distancing, mask usage, and business closures rather than herd immunity.
- Manaus reached 66% infection rate by August 2020; São Paulo reached 22%
- Study analyzed blood donations from February to August using serological testing
- Manaus deaths peaked in May-June then declined sharply despite rising infections
- If São Paulo had followed Manaus's curve, mortality would have been three times higher
- Research expanded to eight Brazilian cities to identify epidemic drivers
USP-led study suggests Manaus reached herd immunity with 66% population infection by August, contrasting sharply with São Paulo's 22% seroprevalence, indicating different epidemic trajectories.
By late August, two Brazilian cities had lived through the same pandemic but emerged from it in starkly different ways. Manaus, the capital of Amazonas state, had been ravaged. São Paulo, the nation's largest city, had been constrained. A study released in September 2020 by researchers at the University of São Paulo offered a stark explanation for why: Manaus had simply run out of people to infect.
The research, led by epidemiologist Ester Sabino and first-authored by Lewis Buss, a graduate student at USP's Institute of Tropical Medicine, analyzed blood donations collected between February and August from two blood banks—one in Manaus, one in São Paulo. The team selected roughly a thousand samples each month from each city and tested them for antibodies against the coronavirus. What they found, after applying mathematical corrections to account for the limitations of serological testing, painted a picture of two epidemics on opposite trajectories. In Manaus, the infection rate had climbed from less than one percent in March to nearly 66 percent by August. In São Paulo, it had reached only about 22 percent.
These numbers suggested something epidemiologists call herd immunity—the point at which enough of a population has been infected that the virus runs out of susceptible hosts and transmission slows dramatically. Sabino and her team argued that Manaus appeared to have crossed this threshold. The city's death toll had peaked in May and June, then declined sharply even as the infection rate continued climbing. The researchers proposed that the sheer prevalence of the virus in the population had, in effect, contained itself. "The exposure to the virus itself led to the decline in new cases and deaths in Manaus," Sabino explained, though she noted their findings showed infection rates substantially higher than previous estimates.
The methodology required careful calibration. Blood donors, the researchers acknowledged, tend to be younger and healthier than the general population, and in Manaus they skewed heavily male. The serological test itself had an accuracy rate of about 85 percent for asymptomatic or mildly ill people, meaning a 15 percent false-negative rate. Most critically, the team had to account for a phenomenon observed globally: antibodies against the coronavirus decline rapidly within months of infection. Without these adjustments, the raw numbers looked far lower—27.5 percent in Manaus by August rather than 66 percent.
São Paulo told a different story. The city's epidemic curve was flatter, its infection rate far lower, and its trajectory suggested that factors other than herd immunity were at work. By August, the city remained under restrictions: masks were widespread, schools and cultural venues remained closed, and mobility was still limited despite the reopening of commerce. Sabino pointed out that if São Paulo had followed Manaus's epidemic curve, the city would have experienced three times the mortality. Instead, the slower burn suggested that public health measures—not population immunity—were holding the line.
The contrast raised a troubling question: Why had two major Brazilian cities, adopting similar public health policies at roughly the same time and with comparable levels of social isolation, experienced such different epidemics? Buss acknowledged the puzzle. "The two cities have very different epidemiological curves and it's very difficult to explain why based solely on serological data," he said. The team's next step was to expand the analysis to eight other Brazilian cities—Rio de Janeiro, Salvador, Recife, Fortaleza, Curitiba, Belo Horizonte, and Campo Grande—to search for the factors that might explain these divergent paths.
The research, conducted under the Brazil-UK Centre for Discovery, Diagnosis, Genomics and Epidemiology of Arboviruses and funded by São Paulo's research foundation along with British institutions, represented a pragmatic approach to understanding a pandemic in real time. Using blood bank samples allowed the team to measure infection rates faster and more cheaply than door-to-door surveys, though the method carried inherent biases that demanded rigorous correction. As Sabino noted, the work was ongoing, and the answers it might yield could reshape how public health officials understood the forces shaping an epidemic's course.
Citações Notáveis
The exposure to the virus itself led to the decline in new cases and deaths in Manaus, though our results show infection rates substantially higher than previous estimates.— Ester Sabino, epidemiologist and study coordinator, USP
The two cities have very different epidemiological curves and it's very difficult to explain why based solely on serological data.— Lewis Buss, lead author and graduate researcher, USP
A Conversa do Hearth Outra perspectiva sobre a história
Why does it matter that Manaus reached 66 percent infection while São Paulo stayed at 22 percent? Aren't they just different cities?
The difference is the mechanism. Manaus suggests the virus burned through the population so completely that there was no one left to infect—herd immunity. São Paulo's lower rate, despite similar policies, suggests people there were protected by masks, closures, and restricted movement instead. One city let the virus run; the other held it back.
But if Manaus had herd immunity, shouldn't deaths have stopped entirely?
They declined sharply, yes, but not to zero. Herd immunity doesn't mean the virus vanishes. It means transmission slows because there aren't enough susceptible people. But the damage was already done—Manaus paid a much higher price to reach that point.
The study mentions antibodies fade quickly. Does that mean people who recovered in Manaus could get infected again?
That's the unsettling implication. The antibodies decline within months, which is why the researchers had to make such careful mathematical adjustments. It raises questions about how durable immunity actually is, and whether Manaus's apparent protection would hold.
Why use blood bank samples instead of testing the general population directly?
Speed and cost. You can test thousands of samples monthly without knocking on doors. But blood donors are younger, healthier, and in Manaus predominantly male—so you're not seeing the full picture. The researchers had to correct for that bias.
What happens next with the other cities?
They're looking for the pattern. If Rio or Salvador or the others show different curves, different infection rates, the team might finally understand what drives an epidemic's shape—whether it's the virus itself, the population, the policies, or something else entirely.