Representative Estimates of COVID-19 Age-Specific Infection Fatality Rates from Four Locations in India: A Cross-Sectional Observational Study

Abstract

Background: India has the second-highest number of SARS-CoV-2 cases in the world. Age-specific infection fatality rates (IFR) are critical parameters for public health policy, but little is known about IFR in low- and middle-income countries. This paper estimates age- and sex-specific IFR of SARS-CoV2 in four regions in India, with a combined population exceeding 150 million.Methods: We estimated infection levels from large, population-representative prevalence/seroprevalence surveys. We drew data on the corresponding number of deaths from official reports and estimated IFR as the number of deaths per infection. We accounted for test inaccuracies, survey design, and potential biases with a battery of alternative models. We calculated the slope of the IFR-by-age function, representing increased risk associated with age.Findings: Among males aged 50–89, measured IFR was 0·037% in Tamil Nadu (95% CI: 0·035%, 0·040%), 0·12% in Karnataka (0·09%, 0·15%), 0·53% in Mumbai (0·52%, 0·54%), and an imprecise 5·64% (0, 11·16%) among migrants in Bihar. Estimated IFR was approximately twice as high for males as for females, heterogeneous across contexts, and rose less dramatically at older ages compared to similar studies in high-income countries (HIC).Interpretation: Estimated age-specific IFRs in India are generally lower than in HIC, but vary widely across the country. Low estimated IFR and geographic variation within India may partly be due to under-testing of suspected COVID-19 deaths, though reporting errors are unlikely to explain cross-country disparities. The elderly in India are at an advantage relative to peers in HIC. Low elderly IFR may partly be due to survivorship bias in a country with lower life expectancy.Funding: Emergent Ventures Grant #466.Declaration of Interests: Authors declare no competing interests.Ethics Approval Statement: This study was exempted from ethical approval because it involved no human subjects. All data were secondary. Ethical approvals and protocols for collecting primaryseroprevalence data have been published in separate papers.