Representative Image Only

Data on COVID deaths in India: Lessons for building a robust healthcare system

The Sample Registration System (SRS)’s death rate data for 2020 exposes the weakness of recording deaths in India when a massive pandemic strikes. The SRS estimates are the closest to reality, but they too could be underestimates because of their conservative approach.

—–

Why is the SRS 2020 data important?

THE routine annual Sample Registration System (SRS)-Statistical Report for 2020, like each year, provides the demographic dynamics of India.  It passed off without much fanfare after being published. But the 2020 SRS Report is critical from the perspective of deaths because 2020 was the first year of COVID; we saw death all around, and it was expected that the number of deaths for the year would be substantially higher as compared to the previous year.

The officially reported COVID death figure for 2020 is 1,52,539 for the period ending December 31, 2020. How authentic is this data? The SRS crude death rate data for 2020 was released a few days ago. What do these figures tell us about deaths in 2020?

Also read: Several Studies Using Diverse Methods Point to Undercounting of COVID-19 Deaths, Govt in Denial

How does the report’s crude death rate data for 2020 indicate underreporting of COVID deaths?

When we look at the trend in death rates (Table 1) over the last seven years, we see a declining trend from 6.7 deaths per 1,000 mid-year population (population estimates from the Union Ministry of Statistics and Programme Implementation) in 2014 to 6.0 deaths per 1,000 in 2019. Each year, the actual number of deaths decline. For instance, in 2017 the estimated deaths (6.3) amounted to 84,33,810 deaths; in 2018, it (6.2) totalled 83,86,120; in 2019, it (6.0) added up to 81,98,400. So, the decline from 2018 to 2019 was 1,87,720 less deaths.

Even with this conservative estimate, the official deaths attributed to COVID of 1,52,539 is three-fourths of the estimated excess of 2,01,705, indicating that COVID death reporting was under-estimated by at least 25 per cent.

Table 1: Trends in Crude Death Rates per 1,000 midyear population

Year Total Male Female
2014 6.7 7.1 6.2
2015 6.5 7.1 5.4
2016 6.4 6.8 5.9
2017 6.3 6.7 5.9
2018 6.2 6.6 5.7
2019 6 6.5 5
2020 6 6.6 5.4

Source: SRS Statistical Reports 2019 and 2020

If 2020 were a normal year, we would have seen a similar decline in deaths estimated. Using the average of decline from 2017-2018 and 2018-2019, we get the figure of 1,17,705 less deaths for 2020. However, SRS 2020 tells us that the death rate in 2020 remained constant at 6.0 per 1,000, which translates into 82,82,400 deaths, which is 84,000 more deaths in 2020 as compared to 2019. Add to this the expected decline of 1,17,705, and we get excess deaths amounting to 2,01,705; this is a conservative estimate because given the national lockdown in 2020, some types of deaths such as accidents would have reduced substantially. But it is also argued that because of the lockdown, access to healthcare declined, several people lost their livelihoods and/or went into depression; this would have contributed to increased mortality.

Nevertheless, it is clear that even with this conservative estimate, the official deaths attributed to COVID of 1,52,539 is three-fourths of the estimated excess of 2,01,705, indicating that COVID death reporting was under-estimated by at least 25 per cent. Of course, in 2021, the excess deaths are going to be much higher, and we will know the full details only when SRS releases the 2021 Report.

Also read: COVID-19: Local Reports Reveal Massive Under-counting of Deaths in North India

What are the other COVID death trends noticeable in the data?

Further, when we compare 2019 and 2020 deaths disaggregated by gender, geography and age (Table 2), we find that from 2019 to 2020, male death rate increased from 6.5 to 6.6, female death rate remained the same at 5.4, rural areas saw a decline from 6.5 to 6.4, and urban areas saw a jump from 5.0 to 5.1; this clearly shows how COVID played out, with men and the urban population being affected more severely.

Table 2: SRS Crude Death Rates per 1,000 mid-year population

2020 2019
Category India Maharashtra Kerala India Maharashtra Kerala
Total 6 5.5 7 6 5.4 7.1
Male 6.6 6.1 8.3 6.5 5.9 8.5
Female 5.4 4.8 5.9 5.4 4.8 5.9
Rural 6.4 6.2 7 6.5 6.2 7.1
Urban 5.1 4.6 7.1 5 4.4 7.2
Per cent deaths by age group
< 1 9.1 4.3 1.2 9.9 4.9 1.2
1 to 4 0.9 0.7 0.4 1.3 1 0.5
5 to 14 1.2 1.1 0.9 1.4 0.8 0.3
15 to 59 31.8 31.9 26 30.5 33.9 22.6
60 + 57 62 71.5 57 59.3 75.4

Source: same as Table 1

The age distribution of deaths show that while the child and adolescent age groups showed the expected secular decline in per cent share of deaths, the 15-59 age group witnessed a rise from 30.5 per cent of deaths to 31.8 per cent, and the 60+ age group share remained constant at 57 per cent. Again, this clearly indicates that the working adult population and the elderly were more affected.

In Kerala, interestingly, the death rates in 2020 were lower than 2019 across all disaggregates; this reflects that in the first phase, Kerala controlled COVID, especially COVID deaths, effectively.

Maharashtra and Kerala were the two most affected states and were probably also states with a more robust reporting of cases and deaths. In the first phase, Kerala managed COVID exceptionally well with minimal mortality, so its trends are a little different too. In Table 2, we see that Maharashtra saw an increase in death rate from 5.4 in 2019 to 5.5 in 2020; in the case of males, from 5.9 to 6.1 deaths per 1,000, and in urban areas, from 4.4 to 4.6.

For Maharashtra, this translates into 6,59,880 deaths in 2019, and 6,78,150 in 2020, an excess of 18,270 deaths, which indicates that Maharashtra’s recording of COVID deaths was closer to the estimated number of deaths by SRS. Earlier in the year, the death registration data for Maharashtra also showed a huge increase in the number of deaths registered in 2020 over 2019, from 6,93,800 in 2019 to 8,08,783 deaths in 2020, reflecting a more robust death registration system.

On the other hand, Kerala, which already has a good death registration system and a much stronger public health system, showed a different trend in 2020. In 2019, SRS estimates Kerala’s death rate at 7.1 or 2,49,210 deaths, which in 2020 went down to 7.0 or 2,47,100 deaths. In Kerala, interestingly, the death rates in 2020 were lower than 2019 across all disaggregates; this reflects that in the first phase, Kerala controlled COVID, especially COVID deaths, effectively. Of course, in the later phases of COVID, Kerala was also hit badly, and we will only know the excess mortality when SRS releases its 2021 data.

Also read: Lakhs of preventable Covid-19 deaths happened due to lack of planning: Anand Grover

What is the major takeaway from this analysis?

Overall, this clearly exposes the weakness of recording deaths in India even when a massive pandemic strikes. The SRS estimates are the closest to reality, but they too could be underestimated because of their conservative approach. While SRS is an important contributor to knowledge on demographics, COVID has demonstrated the need to have a strong and universal death reporting system to be put in place immediately. There are huge learnings from the COVID pandemic for documenting morbidity and mortality, and these need to be extended to all morbidity and causes of death so that we can get more robust morbidity and death statistics, which can in turn help plan a more robust healthcare system in the country.

The Leaflet