Estimating the impact of implementation and timing of the COVID-19 vaccination programme in Brazil: a counterfactual analysis

Abstract

Background: Vaccines developed between 2020 and 2021 against the SARS-CoV-2 virus were designed to diminish the severity and prevent deaths due to COVID-19. However, estimates of the effectiveness of vaccination campaigns in achieving these goals remain a methodological challenge. In this work, we developed a Bayesian statistical model to estimate the number of deaths and hospitalisations averted by vaccination of older adults (above 60 years old) in Brazil.

Methods: We fit a linear model to predict the number of deaths and hospitalisations of older adults as a function of vaccination coverage in this group and casualties in younger adults. We used this model in a counterfactual analysis, simulating alternative scenarios without vaccination or with faster vaccination roll-out. We estimated the direct effects of COVID-19 vaccination by computing the difference between hypothetical and realised scenarios.

Findings: We estimated that more than 165,000 individuals above 60 years of age were not hospitalised due to COVID-19 in the first seven months of the vaccination campaign. An additional contingent of 104,000 hospitalisations could have been averted if vaccination had started earlier. We also estimated that more than 58 thousand lives were saved by vaccinations in the period analysed for the same age group and that an additional 47 thousand lives could have been saved had the Brazilian government started the vaccination programme earlier.

Interpretation: Our estimates provided a lower bound for vaccination impacts in Brazil, demonstrating the importance of preventing the suffering and loss of older Brazilian adults. Once vaccines were approved, an early vaccination roll-out could have saved many more lives, especially when facing a pandemic.

Funding: The Coordenação de Aperfeiçoamento de Pessoal de Nível Superior-Brazil (Finance Code 001 to F.M.D.M. and L.S.F.), Conselho Nacional de Desenvolvimento Científico e Tecnológico - Brazil (grant number: 315854/2020-0 to M.E.B., 141698/2018-7 to R.L.P.d.S., 313055/2020-3 to P.I.P., 311832/2017-2 to R.A.K.), Fundação de Amparo à Pesquisa do Estado de São Paulo - Brazil (contract number: 2016/01343-7 to R.A.K.), Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro - Brazil (grant number: E-26/201.277/2021 to L.S.B.) and Inova Fiocruz/Fundação Oswaldo Cruz - Brazil (grant number: 48401485034116) to L.S.B., O.G.C. and M.G.d.F.C. The funding agencies had no role in the conceptualization of the study.

Keywords: Bayesian model; COVID-19; Deaths; Hospitalisation; Pandemic; SARS-CoV-2.

Fig. 1

Estimated number of hospitalisations (top) and deaths (bottom) by epidemiological week with (dark orange) or without (light orange) vaccination roll-out, by age group (panels). The observed number of hospitalisations and deaths are given by the black dots. Line: median, shadow: 95% CI.

Fig. 2

Estimated number of hospitalisations (top) and deaths (bottom) due to COVID-19 by epidemiological week with the realised (dark orange), 4 (light purple) and 8 (dark purple) weeks faster vaccination roll-out, by age group (panels). The observed number of hospitalisations and deaths are given by the black dots. Check legend for grey scale. Line: median, shadow: 95% CI.

Fig. 3

Posterior distribution of hospitalisations (left) and deaths (right) due to COVID-19 potentially averted by vaccination between January 1, 2021, and August 29, 2021, by age group, with the realised (dark orange), 4 (light purple) and 8 weeks faster (dark purple) vaccination roll-out. Check legend for grey scale.