India is slowly recovering from the second wave of SARS-CoV-2 . It is still uncertain whether a third wave will happen and if so with what intensity. Assessing how many people actually died from COVID-19 , and why they died, can help governments and the health-care sector prepare their treatment strategies effectively.
When it comes to evaluating fatalities, the Indian Council of Medical Research has issued clear guidelines as to what constitutes a COVID-19 related death. Despite this, it is evident that there is an undercounting of deaths in India due to a variety of reasons. Even in rich countries, where almost every death is registered, undercounting has been observed based on international analyses of “excess deaths”. This reflects the “real world” problems of acquiring perfect data during a public health crisis.
Given that this is the case, there is no shame for either the Government of India or the country’s States to admit that the reported number of deaths due to COVID-19 is much less than the actual figures. The reticence of the public in reporting a death as COVID during the pandemic is also a factor. To estimate the actual numbers is far more honourable given the magnitude of the crisis, and the precision of planning required to overcome it. In this article, we concentrate on estimating COVID-19 fatalities in Chennai and whether the southern metropolis too falls prey to “undercounting” COVID-19 deaths that plagues India, and the world, currently. This adds to the previously published articles in this newspaper .
To estimate the “excess deaths” during the period of the pandemic requires accurate baseline data of deaths in the previous years. For Chennai, the data of age unstratified all cause daily deaths (ACD) are available from 2010 to 2019. However, the Greater Chennai Corporation was redefined in 2012, adding a significant population that came under its ambit. Hence we only consider the 2015-19 data. During this period the ACD steadily increased by about 1,850 per year (3% per year). The projected population of Chennai also increased at roughly the same rate. Hence, we assume that, if the novel coronavirus pandemic had not occurred, this steady increase in the ACD would have continued in 2020-21. Our estimated baseline is based on this assumption ( Figure 1 ). The average of ACD from 2018 to 2019 is used as the baseline for the age-stratified deaths ( Figure 2 ) since the age dependence of the increasing trend cannot be reliably extracted from the data that just spans two years. To summarise, we have used two different baselines to analyse the excess ACD in the pandemic period, 2020-21. The first (method I) for the total (age-unstratified) ACD using the 2015-19 data, including the effects of the observed increasing trend. The second (method II) for the age-stratified ACD using the simple average of the 2018-19 data.
The total baseline ACD ( Figure 1 ) shows a clear seasonality. There is a clear summer and a winter peak. We cannot ascribe reasons based on data for this just now.
The age stratified data ( Figure 2 ) clearly indicates that this seasonality is mainly due to deaths in the elderly age group (60+).
Examination of the ACD during the pandemic from January 2020 until June 2021, as compared to the baseline of ACD from 2015 to 2019, reveals two clear observations ( Figure 3 ). First, there is reduction in mortality prior to the lockdown in the first wave and also during the intermediate period between the two waves. There could be many reasons for this example, reduction in road accident fatalities and other unnatural deaths, reduction in deaths due to other infectious diseases, etc. As of now, we have no reliable data to ascribe causes and, the dip in mortality during these periods remains unexplained. Second, there is an increase in ACD during the first and second wave of reported deaths due to COVID-19 in Chennai.
Broadly, when the reported mortality due to COVID-19 was low, there was a reduction in ACD, and the converse when it was high. It is reasonable to conclude based on this data, that the pandemic had two opposing effects on the ACD. On the one hand, the ACD increased due to those who died due to being infected by COVID-19. On the other, it also decreased due to the deaths due to other causes (OCD) reducing due to lockdown effects. At the moment the reasons for this duality are not clear. The only way this issue can be resolved is by conducting a verbal autopsy with the families of the deceased. This has not been performed systematically in Chennai. In the absence of this confirmed data, the under-counting factors are defined to be (excess deaths)/(reported COVID-19 deaths).
The results on the approximate undercounting factor for Chennai, defined above during the different periods of the pandemic, are shown ( Figure 4 and Table ). These undercounting factors are probably underestimates. This is because our estimated number of deaths during the pandemic period inferred from the pre-pandemic data on ACD does not account for the hypothesised reduction in OCD due to the lockdown effects. It is clear that the undercounting of the deaths was much more in the second wave as compared to the first one.
The age dependence of excess deaths and undercounting ratio are shown ( Figure 4 ).
The ACD in the “juveniles” (0-17 years) and “young adults” (18-44 years) are reduced throughout the pandemic period (deficit deaths) except for the young adults during the second wave. The undercounting ratio in the first wave increases with age but the trend is not discernible during the second wave.
It has also been reported that the mortality rate due to COVID-19 among the elderly was significantly less than that of the international data. However, this analysis of data from Chennai indicates that this may simply be a consequence of the fact that the undercounting factor of the deceased among the elderly is much more than that of the younger age group during the first wave. The reasons for this undercounting are not clear just now. Is it a lack of medical certification of cause of death as COVID-19? Or other causes? This will only be known after a thorough examination of the causes of death in the future. However, it is important to acknowledge that there has been an undercounting that requires detailed quantitative analysis. Understanding the severity of this viral infection is important to devise effective policy for the immediate future.
T.S. Ganesan is Professor, Medical Oncology and Clinical Research, Cancer Institute (WIA), Chennai. R. Rajaraman is Professor, Homi Bhabha National Institute, Indira Gandhi Centre for Atomic Research, Kalpakkam. R. Shankar is Honorary Professor, The Institute of Mathematical Sciences, Chennai. The views expressed are personal