Our vaccines are rightly prioritized based on risk (defined as significant, irreversible harm). We’re currently targeting most vulnerable groups: 60+, 45+ with specific comorbidities. So, it’s natural to ask what % of vulnerable have already been covered. Here’s an attempt at numeracy on this topic, admittedly by a non-expert. For simplicity, I stick to 60+ in my city of residence.
First, denominator.
Health ministry estimates 10% of Indians to be 60+. This seems reasonable as 2011 census had this number at 8.6%. It’s been creeping up over time (5.6% in 1961). According to Longitudinal Aging Study in India (LASI-2020), Maharashtra is similar to India in terms of % of 60+. Assuming same 10% for Mumbai, I get to 1.25 million people aged 60+ in Mumbai (defined as what comes under BMC, not all of MMR).
Next, numerator. In two parts.
Mumbai’s elderly acquired immunity through two routes: virus, vaccine.
Virus-route is best estimated using sero-survey data, not reported case counts that vastly understate infection. Mumbai’s second sero-survey, published in October-2020, estimated 48% of 60+ in slums and 13% of 60+ in non-slums to have antibodies (works out to ~35% overall). Since detected cases were only 5% of antibody prevalence at that time, 95% of these people had no idea that they had covid. Between that sero-survey and now, Mumbai’s reported cases have doubled. It’s likely that true infections have also risen, not necessarily in the same proportion. With recent infections mostly from non-slums, infection increase was likely higher in non-slums and lower in slums. Since I’m aiming for a lower-bound, I am going with 60% and 25% for slums and non-slums respectively as percentage of 60+ with some immunity to covid. This works out to 45-50% of elderly in Mumbai with some immunity via virus-route.
As I write this, 0.45 million people aged 60+ in Mumbai have got first shot of vaccine (per Cowin dashboard). That’s 35% of 60+ population. Before adding (b) to (a), I need to correct for double-counting. Some vaccines would’ve gone to people who already had antibodies. Assuming those with and without antibodies are equally likely to get vaccinated (since 95% didn’t know they had antibodies), I count 50-55% of vaccinations as first-time immunity. This works out to an additional 20% of elderly in Mumbai with some immunity only via vaccine-route.
Adding both routes, around two-thirds of Mumbai’s 60+ cohort have some degree of acquired immunity to covid. Assuming proportionate vaccination turnout for slums and non-slums, three-fourths in slums and half in non-slums have some degree of acquired immunity.
The genesis of my analysis lies in a sharp divergence between top-line and bottom-line in recent covid uptick. As I wrote earlier (https://buggyhuman.substack.com/p/arbit-not-random) “data seems to indicate meaningfully lower virulence than before”. Since early-Feb, Mumbai’s active cases have grown 6x while critical cases have only increased 1.7x. Daily cases are running at 2x of 2020 peak, while daily deaths are at 5% of 2020 peak.
I don’t claim to present an answer, since messy world isn’t so easily reductionist. I’m merely tinkering with public data pertinent to one explanatory variable, hoping that I’m not way off. I wish those with better context, data or methods attempt a better answer to the question in my essay-title, for all of India. Happy numeracy.
Postscript: A few related questions
What about antibody fading?
This complicates using antibody-detection as surrogate for immunity. In Mumbai, sero-prevalence in slums actually declined from first to second sero-surveys. Preliminary results of third sero-survey show similar prevalence in slums as in second survey (higher in non-slums, at 22%). However, there’s supposed to be some level of residual immunity even after antibodies become undetectable in tests. It’s above my pay-grade to wade into this topic using words like T-cells. I use the phrase acquired-immunity rather than antibodies to account for this. Beyond a point, my choice of 60% for slums is arbitrary, though not unreasonable. Especially since first sero-survey showed 57% and virus spread has continued to a point where slums account for ~10% of incremental cases (despite accounting for >60% of population). I believe my estimate to be low, not high.
How’s vaccination progress in Mumbai?
While headlines lament 3% vaccination coverage in India, it’s a nonsensical number. Why use 100% as denominator when we are only targeting 10-15% in this phase. Covering 35% of Mumbai’s elderly in three weeks isn’t bad. On a good day, we’re adding 2.5% to this number (half of it going to those without prior antibodies). With continuing scale-up, we’re a few weeks away from vaccinating a majority of Mumbai’s most vulnerable. (India’s approaching 20%, vs Mumbai’s 35%, on this metric)
But, second dose isn’t done?
Of course. However, with trial data revealing meaningful immunity even after first dose, disease can get mitigated even before second dose is done. More generally, I find it better to think of the world as grey rather than binary. Immunity, via vaccine or otherwise, is about substantially lower odds of dire outcomes, not 100% protection from any infection. This view helps avoid panic when an occasional post-vaccination infection draws disproportionate media sensationalism.
How about 45+ with comorbidities?
For simplicity and data-availability, I ignored 45+comorbid. Across India, 45-59 year-olds were ~13% of population in 2011 census (vs ~8% for 60+). However, only a minority of 45-59 year-olds have comorbidities. LASI-2020 suggests that prevalence of diabetes and coronary-disease in 45-59 cohort is around 10% and 20% respectively. However, a majority of this is un-diagnosed if not un-treated. Vaccine eligibility specifically mentions ‘on treatment’, making eligibility lower than prevalence. If 10% of 45+ are eligible, 45+comorbid category is one-sixth as large as 60+. Vaccination turnout nearly mirrors this ratio, with 45+comorbid turnout at one-fifth of 60+. While my estimate is for 60+, I don’t think picture will be very different for 45+comorbid.