Excess Deaths in the COVID-19 Pandemic in India

Abstract

Background: Excess death estimates, reflecting a combination of covid- 19 and non-covid-19 factors, are required to understand the full impact of the pandemic. Relatively little is known about excess deaths in low and middle income countries (LMIC) owing to weaknesses of official statistics on number of deaths and causes of death. India has the second largest number of covid-19 infections globally and is thus an important LMIC. We contribute to the measurement of two key facts: all causes excess deaths and an upper bound for the IFR. Methods: Consumer Pyramids Household Survey (CPHS) is a panel survey dataset across India carried out three times a year. The survey has demographic, economic, and health information about 170,000 households and 600,000 individuals. We use this to estimate excess deaths in the April - August 2020 period across India. We examine correlations between member characteristics such as age, sex, health conditions, urban or rural residence, and economic profile, with the probability of death. We juxtapose this estimated excess death rate with estimates from a seroprevalance survey for the same household dataset, in the state of Karnataka (which has a population of 64 million), to estimate an upper bound for the IFR. Findings: The average annualised crude death rate for May - August 2015-2019 was 8.293 per 1000. This increased to 15.02 per 1000 in May - August 2020. This suggests an annualised excess death rate of 6.7 per 1000. There is a certain divergence of correlations with mortality with member characteristics such as gender, age and income class, as compared with what may be expected with covid-19 related deaths. The estimated upper bound for the IFR in the state of Karnataka is 1.3%.Interpretation: This rise in all-causes deaths, and the socioeconomic characteristics associated with mortality, can help shape improved responses on the part of health policy in understanding the vulnerable and modifying policies to address the problems. These methods can be applied with some panel datasets in other LMIC in order to estimate similar quantities. These results hint at a significant role for non-covid-19 factors in the increased deaths.Funding Statement: None.Declaration of Interests: The authors declare no conflict of interest.Ethics Approval Statement: The work does not involve primary data collection. No ethics approval was therefore taken.