Unaccounted deaths or exceptional management of epidemic: what we can see from latest seroprevalence survey

 4th round of seroprevalence survey by ICMR in June-July 2021 has revealed that nearly 2/3rd of Indian population has antibodies. The presence of antibodies is assumed to indicate prior exposure (though not necessarily manifested with disease symptoms) to Coronavirus. Exposure to virus can be seen as indicative of low probability of Covid-19, at least for the short-term. Very high seropositivity, more than 70%, suggests low probability of occurrence of severe Covid-19 outbreak. 

Table 1

After feeling somewhat relived over national estimate, if we look at state-level results, there is a variation of seropositivity levels. (Table 1) Seropositivity is highest in Madhya Pradesh (79%) and least in Kerala (44.4%). It has been noted with hint of surprise that Kerala and Maharashtra do not have very high seropositivity despite bagging the attention for their high Covid-19 numbers for sustained duration. 

We must understand that one should read these state level numbers with some caution. Not all districts of any state were part of the survey. For example, 6 districts of Maharashtra, Beed, Nanded, Parbhani, Jalgaon, Ahmednagar, and Sangli are part of the serosurvey. Seroprevalence of these 6 districts is considered as Maharashtra estimate, which stands at 58%. It is very likely that Mumbai (city and suburbs), Thane, Palghar, Pune, Nashik, and Nagpur, all the districts associated with large cities, had greater seroprevalence in June 2021. 

In all the states covered in serosurvey, state sample consists of a large urban district along with other not so large urban districts or entirely of not so large urban districts (as in the case of Maharashtra). Hence, seropositivity estimates from 4th round of ICMR survey should be taken as lower bounds with expectation that prevalence of antibodies is higher than the estimates. 

With this caveat covered, let us understand the puzzle that 4th Seroprevalence survey leads to. The puzzle is how states with large seroprevalence have relatively less fatal epidemic. Consider the case of Madhya Pradesh. It has highest seroprevalence with nearly 8000 out of 10000 exposed to virus (seroprevalence rate of 79%). But only 1 person out of these 8000 has died as seen in Table 1. 

There are multiple explanations for this apparent variation in Covid-19 mortality across states. It is likely that more than one explanation holds true. 

1. Unreported Covid-19 deaths. 

2. Deaths avoided due to better management of epidemic 

3. Likelihood of mortality different across states 

4. Variation in epidemic trajectory 

Unreported Covid-19 deaths is the likeliest of these explanations. This article is not the first one to talk about unreported Covid-19 deaths in various Indian states. There is some debate going on about ‘excess deaths’ during Covid-19 epidemic in India. Rukmini S and Chinmay Tumbe have consistently worked on estimates of excess deaths. Excess deaths capture deaths which were not expected during ‘normal’ year and thus are likely to be attributable to epidemic. Claims of excess mortality have been contested on the ground of problems with information sources used for estimation.

Unreported Covid-19 deaths resolve the puzzle of varying epidemic fatality across the states if states with low IFR are likely to have high fraction of unreported Covid-19 deaths. If we assume that true deaths due to Covid-19 are not vastly different and higher than official count, then we must conclude that at least one out of remaining three explanation plays a dominant role for observed variation of epidemic across states.

Individuals from certain state in India are more or less likely to die from individuals in other state, either due to reasons related to nutrition/poverty or structural biological reasons is generally a tough position to defend unless we have considerable state-level demographic differences. Poverty and poverty induced nutrition is likely to play role in fatality. But then poorer states in India should have higher fatality, not the exceptionally lower fatality as 4th serosurvey and official Covid-19 statistics is showing. If we rule out poverty or income differences, then we must conclude that, for intrinsic biological reasons, Indian from some states are more likely to die from Covid-19 than Indian from other states. Such intrinsic possibilities are not implausible. There is historical evidence which have shown that colonial settlers in India were more likely to die from certain disease than local population. If some Indian states have such intrinsic biological reason based low mortality, then we must unpack this low mortality, not just for the sake of India, but for the sake of world. This article is not a place to go in further detail, so we will leave this discussion at this point. 

‘Variation of epidemic trajectory’ assumes how epidemic went through its growth, peak and decline or ‘wave’ is not same across the state. This is indeed the case. Period of second wave was of shorter duration for states like Uttar Pradesh, Madhya Pradesh, and Bihar than for states like Maharashtra, Kerala, or Punjab. These states with shorter second wave have relatively high seroprevalence. So, virus must have spread relatively rapidly and extensively through these states of high seroprevalence and short second wave. Yet, these states have low fatality. This is possible for two reasons: the intrinsic biological reasons and better management of epidemic. Intrinsic biological reasons mean Keralite, and Maharashtrians are more likely to have severe Covid-19 disease than a Bihari. I found this possibility deeply perplexing and if there is some chance of it being true, then it requires serious investigation. 

If we keep intrinsic biological reasons aside, then we must conclude that states with shorter second wave, high seroprevalence, and low fatalities have indeed managed the epidemic well. Prime minister was very well right when he praised chief minister of Uttar Pradesh for management of Covid-19. Even more praise should be there for Madhya Pradesh. 

Testing has been touted as an important tool in Covid-19 management strategy. Number of tests to population exposed to Covid-19 ratio tells us how extensive the testing has been. Kerala, which has lowest seroprevalence has this test to infection ratio of 1.46. For every 10000 infected (not necessarily tested) persons, Kerala conducted 14600 tests. Kerala ended up identifying 1856 cases out of these 10000 infections and 8 out of these 10000 succumbed. In contrast, Madhya Pradesh has 1800 tests for every 10000 infected persons which ended up identifying 118 cases and only 1 succumbed. That is a huge success story. With limited testing and limited resources, Madhya Pradesh has pulled of a miracle of making large fraction of population immune to virus without considerable coverage of vaccination. It is a feat that should be studied, distilled, and applied across the world. 

But if this possibility seems too much or leads to some uncomfortable counter-intuition, then we are back to unreported Covid-19 deaths. If our logic is not tamed by our political position, then we must see that that many Indians are likely to have perished in the second Covid-19 wave without becoming part of official count. We can further spin this possibility and reduce the discomfort but may be for a moment we can let the possibility, that human life and death can be too inconsequential to be even counted, stung us.