Estimating under-reporting of COVID-19 cases in Indian states: an approach using a delay-adjusted case fatality ratio

Objectives The COVID-19 pandemic has spread to all states in India. Due to limitations in testing coverage, the true extent of the spread may not be fully reflected in the reported cases. In this study, we obtain time-varying estimates of the fraction of COVID-19 infections reported in the different states. Methods Following a methodology developed in prior work, we use a delay-adjusted case fatality ratio to estimate the true fraction of cases reported in different states. We also develop a delay adjusted test positivity estimation method and study the relationship between the estimated test positivity rate for each state and the estimated fraction of cases reported. Setting We apply this method of analysis to all Indian states reporting at least 100 deaths as of 10 October 2020. Results Our analysis suggests that delay-adjusted case fatality ratios observed in different states range from 0.47% to 3.55%. The estimated fraction of cases reported in different states ranges from 39% to 100% for an assumed baseline case fatality ratio of 1.38%, from 18.6% to 100% for an assumed baseline case fatality ratio of 0.66%, and from 2.8% to 19.7% for an assumed baseline case fatality ratio of 0.1%. We also demonstrate a statistically significant negative relationship between the fraction of cases reported in each state and the testing positivity rate. Conclusions The estimates provide a means to quantify and compare the trends of reporting and the true level of current infections in different states. This information may be used to guide policies for prioritising testing in different states, and also to analyse the time-varying effects of different quarantine measures adopted in different states.

Automated summary

By estimating the true fraction of reported COVID-19 cases using delay-adjusted case fatality ratios, significant variations are observed among different states, which can be used to guide policy priorities for testing and analyze the effects of different quarantine measures.