0/N #thread 👇I was a bit surprised to find Delhi's high sero-positivity rate (23.48%) in early July, and the low infection fatality rate (IFR) of 0.06%. My curiosity led to this short paper on #COVID-19 in #Delhi.
medrxiv.org/content/10.110…
medrxiv.org/content/10.110…
1/N Delhi's IFR of 0.06% is significantly lower than the estimates for China(0.66%),New York(1.45%), France(0.5%), Lombardia (1%),Spain (1.1%) and Geneva(0.64%). One reason to expect this is that COVID-19 impacts the elderly with comorbidities, and India's population is younger.
2/N Here are the age based IFRs for France from science.sciencemag.org/content/369/65… Applying the same to Delhi's population distribution would definitely give a lower IFR than France's. 
3/N Here's Delhi's population distribution in 2011 and the estimated 2020 population distribution. While Delhi is aging it is younger than Europe where about 20% above the age of 60.
4/N Applying the aged based IFRs of China, New York, France, Lombardia and Spain leads to this 👇. Delhi's IFR is still low by at least a factor of two. Are we doing better than Europe in every age group?
5/N Can I compare deaths and IFRs by age group? Delhi doesn't release that data! But, I find that the cumulative distribution of deaths by age group is similar for Mumbai, Bengaluru, Surat and the IDSP data for all of India.
6/N So, I'll use the death distribution of Mumbai instead. I'll also assume the 23.48% sero-prevalence is uniform across age groups. That lead to this 👇
7/N Delhi's deaths happen at a younger age than expected if I used the age based IFRs of the other countries. Is it due to comorbidities? But, shouldn't that affect the 60+ population even more.
8/N Let us compare deaths in age groups (assuming Mumbai's distribution) to what one would expect with IFRs of France, Lombardia and Spain. Seems like at least 1500-2500 deaths are missing (higher for the older 2020 distribution)
9/N If I use these estimates to calculate age based IFRs, these seem to be comparable for ages <60 but drop off significantly from the trend past 60.
10/N Possible explanations:
1. Sero-positivity is higher because of false positives, dengue antibodies (?) or survey design. Maybe 60+ age group has lower infection rate.
(See paper for sources/references for these.)
1. Sero-positivity is higher because of false positives, dengue antibodies (?) or survey design. Maybe 60+ age group has lower infection rate.
(See paper for sources/references for these.)
11/N Possible explanations:
2. Many deaths happened in homes (nobody knew that these were COVID deaths), or were suspected but not registered as COVID+, or were COVID+ at death but the deaths were registered under the comorbidity.
(See paper for sources/references for these)
2. Many deaths happened in homes (nobody knew that these were COVID deaths), or were suspected but not registered as COVID+, or were COVID+ at death but the deaths were registered under the comorbidity.
(See paper for sources/references for these)
12/N Speculative explanations:
3. The virulence of Indian strains is lower.
4. Indian have innate immunity because of BCG, past exposure to common cold corona viruses etc.
These could be true but Occam's razor suggests that the most parsimonious explanation is missing deaths.
3. The virulence of Indian strains is lower.
4. Indian have innate immunity because of BCG, past exposure to common cold corona viruses etc.
These could be true but Occam's razor suggests that the most parsimonious explanation is missing deaths.
N/N We need timely, transparent and accurate data to better understand the epidemic, to design better containment strategies, and to do better science.
Delhi (and all of India, actually) appear to violate this log-linear relationship between IFR and age. @muradbanaji @GidMK @MenonBioPhysics @PriyankaPulla @vishuguttal @Rukmini @abantika77 @rsidd120
Finally, one possible hypothesis for lower fatalities in India. This needs a good study.
