The all-important question in thepandemic that has wrapped the world in its grip for over a year has been this: how many people have actually died from COVID-19? The answer to that, alas, appears to be, in the immortal words of Bob Dylan, “blowin’ in the wind”.
A measure of how unreliable the death toll has been is proved by the fact that there is probably not a single government in the world whose official toll can be relied upon with a degree of confidence. Even in advanced countries such as the United States and Japan, the official toll is reckoned to deviate significantly from reality. This is not just because of governments’ suppression of facts. Instead, these have arisen from a complex array of factors. The most notable aspect of the COVID-19 pandemic has been its sudden and fierce impact, which has caused a massive and sudden spurt in casualties that health-reporting systems have been unable to cope with.
A recent report released by the Institute for Health Metrics and Evaluation (IHME), based at the University of Washington in Seattle, the U.S., provides an estimate of the casualties. Using an updated methodology to assess “excess deaths” , the IHME estimates that nearly seven million people have died worldwide as direct victims of the virus. This is about double the number of casualties reported on the basis of official or verifiable sources.
The estimates of casualties for India, now in the grip of a tidal wave of infections, are staggering. The IHME estimates that actual Indian casualties may be 0.654 million (6.54 lakh), not the official count of 0.221 million (2.21 lakh) as on May 6 when the report was released. That is a whopping three times the official numbers, an indicator of the extent of under-reporting.
Official numbers have lost credibility
As the second wave reached new heights, reporters in the Indian media have struggled to find data that reflect what they see and hear happening around them. Counts of bodies burning in cremation grounds or even in parking lots or, in fact, in any available open space in urban locations have served as proxies for journalists trying to reconcile official numbers on fatalities with the reality around them.
Across Indian cities, journalists have tried to reconcile counts of cremations complying with COVID protocols with the numbers provided by local and State government officials and have invariably found huge variations. Some journalists in Gujarat compiled a long list of the obituaries in city editions of newspapers and compared these with the official tally. One map enthusiast even took satellite images of fires at cremation ghats, which indicate the rough number of funeral fires being lit, and then compared these with previous images of the same location to demonstrate the spike in casualties. What is clear in all these desperate attempts is the reality that the official numbers have utterly lost their credibility in the face of the biggest humanitarian disaster in independent India. One time-tested method of assessing the toll in such tragedies is to estimate the number of “excess” deaths in order to arrive at a figure that is closer to reality. For instance, Chinmay Tumbe, a researcher at the Indian Institute of Management Ahmedabad, recently made an attempt to estimate the toll in the great pandemics during British Rule: cholera in the 18th century, plague in the 18th and 19th centuries and influenza about 100 years ago. The saving grace is that while it took a long time to arrive at reliable and comprehensive estimates of the casualties in the epidemics of the past, reliable estimates of figures for the biggest pandemic of our times are arriving much quicker. The IHME’s estimates are a good example of this progress.
The underestimates may not necessarily arise from just political interference. Factors such as poor health care infrastructure, limited access to health care facilities and the general under-reporting of deaths may “magnify this challenge”, observes the report.
According to the IHME’s analysis, the actual toll in the U.S. may be 0.905 million instead of the official count of 0.574 million and in Mexico, 0.617 million instead of 0.218 million. Significantly, it shows that the actual toll in Brazil, another major COVID hotspot, may not be very much higher than the reported toll: 0.596 million compared with 0.408 million. The actual toll in Russia, however, is likely to be almost six times higher than officially reported: 0.594 million instead of 0.109 million.
The report’s revelations about Japan are striking. The country, which has been touted as an example of “good management” ever since the onset of the pandemic, appears in much poorer light in the report. Japan reported just 10,230 deaths due to COVID between March 2020 and May 2021. However, the IHME estimates fatalities in Japan to have been 10 times that figure, 1,08,320.
The extent of the underestimate of the toll in Italy and the United Kingdom, two other major hotspots, is much lower than in many other countries; in Italy, fatalities are estimated to be 0.176 million instead of the 0.121 million reported, and in the U.K., the actual toll is estimated to be 0.210 million instead of the 0.151 million reported. The IHME study reveals that in several countries—notably Egypt, Kazakhstan and in countries of Eastern Europe—the reported fatalities were several multiples lower than the actual figures.
The analysis finds that COVID-19 deaths are “significantly under-reported in almost every country”. “As terrible as the COVID-19 pandemic appears, this analysis shows that the actual toll is significantly worse,” said Dr Christopher J.L. Murray, the IHME’s director. He explained that getting right the “true number” of COVID-19 casualties would help the world “appreciate the magnitude of this global crisis”. Moreover, it would provide “valuable information to policymakers developing response and recovery plans”, he observed.
A new method
Estimating the “excess” deaths in a population is an indirect method of assessing the actual impact of diseases. The idea is to use demographic data of the past and estimate how many more people died during the episode in order to arrive at the number of “excess” deaths that can be attributed to the cause that is the focus of the study. The IHME took a similar approach but after making several modifications.
It pointed out that since there was great diversity in the credibility and availability of statistics across the world, and even within countries, it adopted a new methodology. In fact, it accounts for the fact that“actual” COVID fatalities as a proportion of all fatalities also vary considerably across geographies, and even within countries. Moreover, in more advanced societies, where the population of the elderly is higher, many COVID deaths among this segment of the population went unreported during the initial days of the pandemic. It is well known that actual fatalities among the elderly in the U.S., for instance, especially of those housed in institutions, was much higher than the reported numbers. In many other countries, most notably, Russia and in Eastern Europe, official fatality counts were known to be gross underestimates. Counting “excess” deaths thus offers a solution to all these estimates made in widely varying circumstances.
But getting true estimates of fatalities is not just a matter of settling our consciences. “Estimating the total COVID-19 death rate is important both for modelling the transmission dynamics of the disease to make better forecasts, and also for understanding the drivers of larger and smaller epidemics across different countries,” the report states.
Six drivers of ‘excess’ deaths
Explaining its methodology, the report points out that “all-cause mortality” since the onset of the pandemic has been influenced by six factors as a result of the physical distancing norms and mobility restrictions that have been adopted since last year. In order to compute “excess” deaths, it lists six “drivers”, starting with the number of deaths from COVID-19 infections. The second driver of “excess” deaths is “delayed or deferred” health care during the pandemic. The third is due to mortality arising from mental health disorders or due to alcohol or drug use. The fourth factor, a mitigating one, is due to lower fatalities caused by accidents because of the restrictions on mobility during the pandemic. The fifth driver, another mitigating factor, is the lower number of fatalities caused by other viruses such as influenza and measles. The sixth driver of “excess” deaths arises from the fatalities among those who were already chronically ill with cardiovascular or respiratory ailments who succumbed to COVID instead of their other chronic medical conditions. Thus, in order to arrive at a correct estimate, the report notes that all six “drivers” of excess deaths have to be incorporated in the model.
The analysis starts with marshalling statistics from regions and countries that have reported all-cause mortality since the beginning of the pandemic last year, which is available for 56 countries and subnational units. The difference between the actual mortality rate and the “expected” death rate gives an estimate of the COVID toll. But the analysis does not stop here. Using other studies and evidence, the IHME estimates “the fraction of excess mortality that is from total COVID-19 deaths as opposed to the five other drivers that influence excess mortality”. It then proceeds to build a statistical model that that “predicts” a ratio of total COVID deaths to reported COVID deaths for regions that do not have reliable COVID fatalities. This ratio is then used to estimate COVID deaths in all locations.
A few caveats
Although the methodology appears to rest on strong foundations, there may be problems associated with the application of the model across geographies and demographies. Although the overall estimate may be robust, the fidelity of the estimates at a disaggregated level may be problematic, particularly because of the extremely uneven quality of the data. Thus, the variability in the quality of the data—both on all-cause mortality and the officially reported COVID mortality—may not offer the fidelity to do nuanced cross-country or even regional comparisons of projections.
We know, for example, that throughout the pandemic detection of COVID infections has been extremely poor and uneven. In fact, the dip in testing rates in recent days, even as the surge has continued, implicitly caps the case count. Naturally, when the case count is artificially lowered, it will influence the count of casualties attributed to COVID. To complicate matters, the wide variation in test rates across regions has a differential impact on the extent to which the case count and the mortality due to the disease deviate from the real numbers. Thus, if, as we know, Bihar’s test rates are far lower than, say, Delhi’s or Kerala’s or Maharashtra’s, the case count and the death toll in Bihar are likely to deviate from the real numbers far more widely than in the other States. Unless the methodology is able to incorporate this degree of nuance into its models, which is difficult simply because of the lack of data, it is likely to miss the extent of the undercount.
Instead of ranking countries by the number of casualties, it would appear that a more meaningful method may have been to group them either along income or geographical lines. For instance, it is striking that most of the countries where the estimated COVID death rate is unconscionably high are countries in Eastern Europe. Azerbaijan’s COVID death rate is estimated to be almost 650 deaths per 1,000 (almost 15 times its official death rate), Bosnia and Herzegovina’s 587 and Bulgaria’s 544.
If the IHME’s estimates appear outlandish to sceptics, these are surpassed by The Economist’ s projections. On May 15, the British publication, using its own model, estimated that the worldwide toll ranges between 7.1 million and 12.7 million with a “central estimate” of 10.2 million. “The official numbers represent, at best, a bit less than half the true toll, and at worst only about a quarter of it,” it said. With reference to India, The Economist said its model “suggests the country is seeing between 6,000 and 31,000 excess deaths a day, well in excess of official figures around the 4,000 mark”. It said this suggests that “around 1 m [million] people may have died of covid-19 in India so far this year”.
Lessons for India
For India, the estimates confirm what was expected—that the actual COVID mortality in Maharashtra and Kerala is higher than the official figure, but the extent of the undercount is much smaller than in the rest of the country. The relatively much more mature health reporting and surveillance systems in these two States obviously explains their performance.
But the IHME’s Dr Murray warns that India is likely to “suffer considerable mortality” between now and September. “Of the well over two million deaths worldwide by then, half of them are going to be in India alone,” he said. “Our understanding of the pandemic has profoundly changed,” said Dr Murray. This is because the global toll is not uniformly distributed across the world. “It is a function of testing and, in some regions, under-reporting has been truly profound.” In particular, Dr Murray referred to countries of Eastern Europe, Central Asia and Egypt.
Dr Soumya Swaminathan, Chief Scientist at the World Health Organisation, observed on Twitter that it is “important to document excess mortality due to Covid and other causes”. This, she said, “will help direct policy responses”.
Also read: Lessons from the ‘first wave’
An important lesson from the report is that accurate and quick testing is the key to getting a grip on the pandemic. Expanded testing is necessary to get not just a measure of the disease and the dynamics of its spread but also a true measure of the mortality.
The recent performance of testing in India is not encouraging; the highest ever number of tests conducted (19.5 lakh) was on April 30; despite the surge since then, the total tests conducted fell to 18.6 lakh tests on May 11. As the health economist Rijo John pointed out, at the end of the first wave there were 63 tests per case, but now when cases have surged there are just 4.5 tests per case. The test positivity rate then was 1.6 per cent compared with 22.3 per cent now, almost 14 times greater.
This report is a stark reminder to the Indian government to get its act together. The Narendra Modi government will ignore it only if it wantonly wishes to heap further misery on the country. Time, of course, will tell whether these estimates were close to reality, but waiting for time to tell its tale would be a monstrous blunder.