Inference of COVID-19 epidemiological distributions from Brazilian hospital data.

Iwona Hawryluk,Thomas A Mellan,Ricardo P Schnekenberg,Axel Gandy,Christl A Donnelly,Henrique Hoeltgebaum,Charles Whittaker,Samir Bhatt,Swapnil Mishra,Harrison Zhu,Seth Flaxman

doi:10.1098/rsif.2020.0596

Abstract

Knowing COVID-19 epidemiological distributions, such as the time from patient admission to death, is directly relevant to effective primary and secondary care planning, and moreover, the mathematical modelling of the pandemic generally. We determine epidemiological distributions for patients hospitalized with COVID-19 using a large dataset (N = 21 000 − 157 000) from the Brazilian Sistema de Informação de Vigilância Epidemiológica da Gripe database. A joint Bayesian subnational model with partial pooling is used to simultaneously describe the 26 states and one federal district of Brazil, and shows significant variation in the mean of the symptom-onset-to-death time, with ranges between 11.2 and 17.8 days across the different states, and a mean of 15.2 days for Brazil. We find strong evidence in favour of specific probability density function choices: for example, the gamma distribution gives the best fit for onset-to-death and the generalized lognormal for onset-to-hospital-admission. Our results show that epidemiological distributions have considerable geographical variation, and provide the first estimates of these distributions in a low and middle-income setting. At the subnational level, variation in COVID-19 outcome timings are found to be correlated with poverty, deprivation and segregation levels, and weaker correlation is observed for mean age, wealth and urbanicity.

Highlights

Surveillance of COVID-19 has progressed from initial reports on 31 December 2019 of pneumonia with unknown aetiology in Wuhan, China [1], to the confirmation of 9826 cases of SARS-CoV-2 across 20 countries one month later [2], to the current pandemic of greater than 28 million confirmed cases and 900 000 deaths globally to date at the time of writing [3]
The onset-to-death distribution [4,5], characterizing the range of times observed between the onset of first symptoms in a patient and their death, proved crucial in early estimates of the infection fatality ratio (IFR) where it was used to estimate the cumulative number of deaths in the beginning of the epidemic in Wuhan [6]
The Bayes Factors (BFs) used for model selection are shown in appendix B, table 6

Summary

Introduction

Surveillance of COVID-19 has progressed from initial reports on 31 December 2019 of pneumonia with unknown aetiology in Wuhan, China [1], to the confirmation of 9826 cases of SARS-CoV-2 across 20 countries one month later [2], to the current pandemic of greater than 28 million confirmed cases and 900 000 deaths globally to date at the time of writing [3]. The onset-to-death distribution was used in recent approaches to modelling the transmission dynamics of SARS-CoV-2 to estimate the reproduction number Rt and other important epidemiological quantities such as the serial interval distribution [7,8,9,10,11,12].

Methods

Results

Discussion

Conclusion