I've put down some estimates of COVID-19 infection fatality rate (IFR) in Mumbai during 2020. To estimate IFR you need: (1) the number of deaths (the numerator) and (2) the number of infections (the denominator). The results... https://medrxiv.org/cgi/content/short/2021.04.08.21255101v1
Median estimate of COVID-19 deaths during 2020: 18K (11.7-24.3K)
Median estimate of prior infection by the year-end: 60% (50-71%).
Median IFR: 0.23% (0.15-0.33%)
The wide confidence intervals reflect big uncertainties. The estimates are useful for thinking about two questions...
Median estimate of prior infection by the year-end: 60% (50-71%).
Median IFR: 0.23% (0.15-0.33%)
The wide confidence intervals reflect big uncertainties. The estimates are useful for thinking about two questions...
Q1) Is the death rate from COVID-19 in India surprisingly low?
Mumbai's data says no. And Mumbai is one of the few places we have enough data to comment on this question.
Previous thread on this: https://twitter.com/muradbanaji/status/1371410878524968962
Mumbai's data says no. And Mumbai is one of the few places we have enough data to comment on this question.
Previous thread on this: https://twitter.com/muradbanaji/status/1371410878524968962
Q2) What can explain the current surge in Mumbai? Could earlier spread have been overestimated?
The answer is: not greatly. I go through how the city's serosurveys fit together and possible biases.
Previous thread on the current surge: https://twitter.com/muradbanaji/status/1379703247083036673
The answer is: not greatly. I go through how the city's serosurveys fit together and possible biases.
Previous thread on the current surge: https://twitter.com/muradbanaji/status/1379703247083036673
Two key references: the meta-analyses of O'Driscoll et al, and Levin et al, with which Mumbai's data is broadly consistent:
O'Driscoll: https://www.nature.com/articles/s41586-020-2918-0
Levin: https://pubmed.ncbi.nlm.nih.gov/33289900/
(n/n)
O'Driscoll: https://www.nature.com/articles/s41586-020-2918-0
Levin: https://pubmed.ncbi.nlm.nih.gov/33289900/
(n/n)