Is India’s Covid-19 death rate higher than Italy’s?
The common perception is that the Covid-19 fatality rate in India is probably the lowest in the world. At first glance, it seems broadly valid in the sense that India’s rate, at 2.8% is much lower than, say, Italy’s at 14.3%, and its concurrent cumulative case fatality rate (CCCFR) certainly looks to be much lower than that of many countries, as seen in Figure 1. The CCCFR is the ratio of cumulative deaths to cumulative cases on a given date. Since death, if it occurs, happens days after a case is detected, CCCFR understates the probability of dying, given that one is infected, especially if the number of cases is rising. Ideally, the fatality should be calculated for every cohort, i.e., people grouped by when the infection was detected. But this is difficult to measure and so CCCFR is globally used as a summary measure of deaths due to Covid-19.
One can make some adjustment for this by taking the lagged CCFR (LCCFR), which is the ratio of cumulative deaths on a given date to cumulative cases on a prior date (in this case, 15 days earlier). This will increase the CCCFR, by a lot, if cases have been rising rapidly, as in India’s case or a little, if they have stabilised, as in Italy’s case. But even by this measure, India comes out looking good, with India at 5.9% still much below many others, and certainly much below Italy’s 14.9%.
However, one of the few certainties about Covid-19 is that it is more likely to be fatal as the age of the infected patient increases. Age-specific CCCFRs are not easy to find, but are now being published, e.g., in Onder, et. al. (2020) for Italy and China. Table 1 presents key numbers from the two countries. In India, we do not have published national age-specific CCFRs, but the government of Maharashtra, in a display of remarkable transparency in the face of the rising impact of the pandemic, provided age-specific cases and deaths from which age specific CCFR could be calculated (the age classes are off by one year, i.e., 31- 40 instead of 30-39, compared to the data for Italy and China) until May 8.
India does, however, have the age-specific number of cases for April 30, published in a recent article authored, inter alia, by the director general of the Indian Council of Medical Research (ICMR). This provides the number of cases in India by the same age-class as the data for Italy and China in Onder, et. al. (2020) and the Chinese Center for Disease Control (CDC). This allows us to estimate the number of deaths that should have occurred in India, if the same age-specific CFRs seen in China and Italy prevailed in India. If the actual number of deaths in India is less than these estimates, then India is doing better than these countries, and vice versa. However, given the fact that China carried out a one-time readjustment of its Covid-19 deaths in mid-April, we use only Italy’s numbers, though the Chinese age-specific CCCFR is provided for comparative information.
In Table 1, the critical number is in column (8). This is calculated by multiplying Italy’s age-specific CFR in column (3) to the age-specific number of cases in India in column (7). The estimated numbers of death that should have occurred, if the age-specific death rates of Italy were to prevail in India, is 535. The official number of deaths in India as of April 30 was actually twice that number, at 1074.
It is possible to reconcile the fact that India’s CCCFR is lower than Italy’s in Figure 1 with the deaths in India being twice the expected number, if one examines the distribution of cases by age. It can be seen, Figure 2a, that more than half the patients in India and Maharashtra are under 40 years, while Italy (the total case figures are estimated using the CCCFR and the number of deaths) has less than one-seventh of cases in this age group, and 56% in the age group above 60, where India has less than one-seventh. Indian Covid patients are much younger and thus, would have been less likely to die in Italy or China, which, as seen in Table 1, have very low CCCFR for patients below 60.
This is seen more clearly in Maharashtra, the only state to release age-specific CCCFR. Maharashtra’s CFR for those below 60 is well over four times higher than Italy, as seen in Figure 2b, even though Italy’s aggregate CFR of 7.2% is much higher than Maharashtra’s 3.8%. If it had Italy’s age-specific CCCFR, Maharashtra would have reduced its deaths by two-thirds and applying Maharashtra’s CCCFR to India would increase the national number of deaths by 50%.
The bottom line is our aggregate CFR is low because we have more young patients than other countries who are not expected to die. However, in India, they are dying at a rate much higher than expected, which means that India has a much higher CCCFR than it should given the experience of other countries. Instead of being among the lowest, India’s age-adjusted death rate is actually higher than Italy’s, which, as seen in Figure 1, has among the highest aggregate CCCFRs.
Why is this so? Is it because our hospital system is not equipped for managing (there is no consensus on treatment for Covid-19, only supporting the patient, while his or her immune system tries to fight the infection) Covid? Or because we lack facilities like oxygen support, etc. Or because our immune system is compromised by lack of nourishment and bad water and air, or because of numerous comorbidities?
It will be difficult to disentangle these factors, but without recognising our problem, we won’t even start asking such questions, and continue to believe we got a lucky break. States can begin this discussion by releasing age-specific CCCFRs in the public domain. ICMR too needs to release age-specific case loads. Only then can we take corrective action in time.