A common belief that deaths due to the pandemic are being underreported has led to several reassessments of deaths in India by researchers, journalists and others. While there is a clear agreement on the undercount of Covid deaths, the magnitude of deaths suggested by a range of professionals is far from being accurate. The problems in counting deaths are two-fold: First, counting the death and second, assigning the cause of death as being on account of Covid-19.
The history of birth, death and marriage registration in India goes back to the Registration Act, which was promulgated in 1886 throughout British India on a voluntary basis. Realising the importance of birth and death registration, the Registration of Birth and Death Act was enacted in 1969 to register and compile the statistics under the Office of the Registrar General and Census Commissioner of India. According to the last civil registration annual report 2019 published in 2021, the level of birth registration in India is close to 93 per cent: This means nine out of 10 births in India are registered. The same report also says that the completeness of death registration has also reached 92 per cent in 2019. If close to 90 per cent of births and deaths are registered in India, why do researchers and policymakers, both at the state and the national level, never make judicious use of such data and vouch for full registration of births and deaths in India?
Since 1969, the Registrar General has also collected medical certification of cause of death and produced a report on this measure. The latest available report of 2019 says that 20.7 per cent of total registered deaths have this information. Therefore, prior to Covid-19, only one out of five registered deaths had a medical certification of cause of death. In most cases, on examination of the death registration data, the cause of death is reported as “old age” or “unknown”. Moreover, this particular information is not filled by the person reporting the death. This is the state of affairs of the medical certification of the cause of death prior to Covid-19.
Counting vital events like births, deaths, marriages and, of late, migration, is the primary engagement of the discipline of demography. Since India is far from recording all births and deaths, alternative methods and approaches were put in place to make a near-accurate assessment. For instance, with the admission of incompleteness of death registration and its extent, we were able to obtain the real magnitude of the count of deaths. For a robust check of such deaths, the simple balancing equation of population growth was sufficient, which is deaths deducted from births with a positive/negative attrition of migration.
A renewed interest in the counting of deaths and its accuracy has emerged on account of the need to assess excess deaths due to the pandemic. However, the excess deaths are estimated by epidemiologists and epidemic modellers with no formal training, who are coming up with bizarre estimates largely based on an inaccurate and incomplete database.
A pandemic database in terms of reported cases may result in misleading and presumptuous estimates of deaths. Information on assumption of transmission, risk of severity of the ailment, access to proper healthcare and, above all, the stage at which care-seeking happened, can improve accuracy. The estimates currently in circulation are not only frightening but also far from reality. Their reliability is based on the accuracy of the inputs used and the assumptions made.
The real endeavour on this front is to have a general idea regarding the number of deaths that could have occurred without the pandemic. This is perhaps not that difficult, given the numerous information bases at hand like the Civil Registration System (CRS) datasets and the extent of incompleteness along with the survivorship summary offered by the periodic life tables of Sample Registration System (SRS). The periodic life tables have an advantage of offering the age cumulation of deaths and, therefore, will be useful in terms of age-sex characterisation of deaths. In addition to this, there are alternative information sets in periodic large-scale surveys such as NFHS with information on deaths in households, which could also serve towards validating a death count otherwise generated. Once we know the magnitude of pre-Covid-19 deaths, we can think of tweaking the observed age-specific death rates in incremental terms.
If the focus is on an accurate count of Covid deaths based on Covid-19 related information base, the outcome will always be suspect owing to a variety of reasons. The stigma and associated protocols of Covid mortality are very likely to lead to misrepresentation of deaths at the point of death itself. Additionally, there could be a reasonable share of deaths on account of failing to access critical care in Covid-19 times, which could have been avoided otherwise. Finally, the real designation of Covid-19 deaths will be possible only for institutional deaths registered under strict guidelines for categorisation of death due to Covid-related complications.
In the midst of all these anxious attempts at assessing Covid-19 mortality, associating all excess deaths (if at all estimated with demographic principles) with the pandemic will not only be naïve but also simplistic. Given our provisioning of Covid-19 care and the wild spread of infection, obtaining an accurate count of pandemic causalities is next to impossible. The most feasible one is to obtain excess mortality as a surplus over the routine count of mortality. But then, all those deaths cannot be singularly attributed to Covid because a certain share of them has to be accounted for under constrained access to healthcare in Covid times. While the undercount of pandemic-associated casualties is not unusual, its adverse impact on survival is only observable when the pandemic is over. This is not to say that an inquiry on excess deaths need not be attempted, but it should ideally be assessed as a deviation from the normal course of mortality backed by demographic principles and sound information base rather than epidemiological models based on suspect inputs and numerous assumptions.
This column first appeared in the print edition on July 17, 2021 under the title ‘How to count the Covid dead’. Rajan is chairman, The International Institute of Migration and Development (IIMAD) and Mishra is professor, Centre for Development Studies, Thiruvananthapuram.