Dynamics of COVID-19 in Amazonia: A history of government denialism and the risk of a third wave

Abstract

The city of Manaus (the capital of Brazil's state of Amazonas) has become a key location for understanding the dynamics of the global pandemic of COVID-19. Different groups of scientists have foreseen different scenarios, such as the second wave or that Manaus could escape such a wave by having reached herd immunity. Here we test five hypotheses that explain the second wave of COVID-19 in Manaus: 1) The greater transmissibility of the Amazonian (gamma or P.1) variant is responsible for the second wave; 2) SARS-CoV-2 infection levels during the first wave were overestimated by those foreseeing herd immunity, and the population remained below this threshold when the second wave began at the beginning of December 2020; 3) Antibodies acquired from infection by one lineage do not confer immunity against other lineages; 4) Loss of immunity has generated a feedback phenomenon among infected people, which could generate future waves, and 5) A combination of the foregoing hypotheses. We also evaluated the possibility of a third wave in Manaus despite advances in vaccination, the new wave being due to the introduction of the delta variant in the region and the loss of immunity from natural contact with the virus. We developed a multi-strain SEIRS (Susceptible-Exposed-Infected-Removed-Susceptible) model and fed it with data for Manaus on mobility, COVID-19 hospitalizations, numbers of cases and deaths. Our model contemplated the current vaccination rates for all vaccines applied in Manaus and the individual protection rates already known for each vaccine. Our results indicate that the SARS-CoV-2 gamma (P.1) strain that originated in the Amazon region is not the cause of the second wave of COVID-19 in Manaus, but rather this strain originated during the second wave and became predominant in January 2021. Our multi-strain SEIRS model indicates that neither the doubled transmission rate of the gamma variant nor the loss of immunity alone is sufficient to explain the sudden rise of hospitalizations in late December 2020. Our results also indicate that the most plausible explanation for the current second wave is a SARS-CoV-2 infection level at around 50% of the population in early December 2020, together with loss of population immunity and early relaxation of restrictive measures. The most-plausible model indicates that contact with one strain does not provide protection against other strains and that the gamma variant has a transmissibility rate twice that of the original SARS-CoV-2 strain. Our model also shows that, despite the advance of vaccination, and even if future vaccination advances at a steady pace, the introduction of the delta variant or other new variants could cause a new wave of COVID-19.

Keywords: Brazil; Coronavirus; Delta variant; Epidemiology; Gamma variant; Immunity loss; Manaus; Multi-strain SEIRS model; Pandemic; Public health; Public health policy; Reinfection; SARS-CoV-2; Transmissibility.

Fig. 1

State-transition diagram for the compartments of the COVID-19 epidemiological model.

Fig. 2

a Daily counts of COVID-19-related hospitalizations and deaths of Manaus residents by date of occurrence (source: FVS Bulletin). b Without immunity loss, even a hypothetical new strain that is five times more transmissible cannot explain the sudden rise of COVID-19-related hospitalizations in December 2020 (the thin blue (green) line indicates the new (old) strain). c Assuming an average period of immunity loss of 240 days and a new strain with twice the usual transmission rate (thin blue line), the computed daily counts of COVID-19-related hospitalizations fit the observed data well. d-e Six non-intervention scenarios of the COVID-19 pandemic in Manaus from March 2020 to December 2021. Scenarios are shown with average periods for loss of immunity from 150 to 720 days, plus a non-immunity-loss scenario (none). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

a Observed and projected daily average COVID-19 hospitalizations. Yellow line: Observed hospitalization; Green line = Original lineage of SARS-CoV-2; Blue asterisk = appearance of the gamma variant; Blue line = gamma variant; blue dotted lines = percentages of the gamma variant for different dates. b Manaus urban mobility index compared to the baseline. Blue asterisk = appearance of the gamma variant; 1- call to return to face-to-face classes in elementary schools; 2 – Actual observed return of students in schools; 3- First round of elections for mayor and governor (according to the SEIRS model, an estimated 2000 individuals had been infected by the gamma variant by this date); 4- Second round of elections for mayor and governor; 5- Christmas. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

a Projected hospitalizations from the delta variant: Yellow line: Observed hospitalization; Blue line = Projection of gamma variant; Red line = Projection of delta variant. b Projection of deaths from the delta variant: Black line: Observed deaths; Blue line = Projection of the gamma variant; Red line = Projection of the delta variant. c Projection of cumulative deaths caused by the gamma and delta variants until the pool of susceptible individuals is exhausted: Blue line = Projection of the gamma variant; Red line = Projection of the delta variant. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)