Study in Science Advances Rejects Media Reports of Excess Mortality in 2020 as Based on Unacceptable Estimates
Some media reports have highlighted the findings from a paper published today in an academic journal Science Advances on life expectancy during the Covid-19 pandemic in India in 2020. These are based on untenable and unacceptable estimates.
While the authors claim to follow standard methodology of analyzing National Family Health Survey-5 NFHS-5), there are critical flaws in methodology. The most important flaw is that the authors have taken a subset of households included in the NFHS survey between January and April 2021, compared mortality in these households in 2020 with 2019, and extrapolated the results to the entire country. The NFHS sample is representative of the country only when it is considered as a whole. The 23% of households included in this analysis from part of 14 states cannot be considered representative of the country. The other critical flaw is related to possible selection and reporting biases in the included sample due to the time in which these data were collected, at the peak of the Covid-19 pandemic.
The paper erroneously argues for the need for such analyses claiming that vital registration system in low and middle income countries including India is weak. This is far from being correct. The Civil Registration System (CRS) in India is highly robust and captures over 99% of deaths. This reporting has constantly increased from 75% in 2015 to over 99% in 2020. Data from this system shows death registration has increased by 4.74 lakh in the year 2020 compared to 2019. There was a similar increase of 4.86 lakh and 6.90 lakh in death registration in the year 2018 and 2019 over the respective previous years. Notably, all excess deaths in a year in the CRS are not attributable to the pandemic. Excess number is also due to an increasing trend of death registration in CRS (it was 92% in 2019) and a larger population base in the succeeding year.
It is strongly asserted that an excess mortality of about 11.9 lakh deaths reported in the Science Advances paper in 2020 over the previous year is a gross and misleading overestimate. It is noteworthy that excess mortality during the pandemic means increase in deaths due to all causes, and cannot be equated with deaths that were directly caused by Covid-19.
The erroneous nature of the estimates published by the researchers is further corroborated by data from India’s Sample registration System (SRS). SRS covers around 84 lakhs population in 24 lakh households in 8842 sample units spread across 36 States/UTs in the country. While the authors take great pains to show that results from NFHS analyses and Sample Registration Survey analyses for 2018 and 2019 are comparable, they completely fail to report that SRS data in 2020 shows very little, if any, excess mortality compared with 2019 data (Crude death rate 6.0/1000 in 2020, Crude death rate 6.0/1000 in 2019) and no reduction in life expectancy.
The paper reports results on age and sex, that are contrary to research and program data on Covid-19 in India. The paper claims that excess mortality was greater in females and in younger age groups (particularly 0-19 year old children). Data on about 5.3 lakh recorded deaths due to Covid-19, as well as research data from cohorts and registries consistently shows higher mortality due to Covid-19 in males than females (2:1) and in older age groups (several fold higher in > 60 years olds than in 0-15 year old children). These inconsistent and unexplainable results in the published paper further reduce any confidence in its claims.
In conclusion, the all-cause excess mortality in 2020 compared with the previous year in India is markedly less than the 11.9 lakh deaths reported in the Science Advances paper The paper published today is methodologically flawed and shows results that are untenable and unacceptable.