Death tolls of COVID-19: Where come the fallacies and ways to make them more accurate

ABSTRACT

The death toll of the coronavirus disease 2019 (COVID-19) sparked much controversy since its advent in December 2019. Underestimation because of under testing and deaths happening outside the hospitals were important causes. Bold revisions of the diagnostic criteria leading to dramatic changes in death tolls by different governments were observed in attempts to generate more accurate estimates. On the other hand, the influence, censorship and manipulation on case and death data from top political leaders of some countries could create important impacts on the death toll. Baseline mortality data of previous years may help make more accurate estimates of the actual death toll. The pitfalls and strategies during such processes could become valuable lessons to leaders and policymakers worldwide as more accurate statistics serve to navigate policies to combat this pandemic in the days and months to come.

KEYWORDS:

• COVID-19
• death toll
• mortality
• case definition
• infectious disease

Introduction

Confusion prevailed over the death toll due to the coronavirus disease 2019 (COVID-19) since its advent in December 2019. As the pandemic loomed, it became a universal challenge that baffled governments globally as this enigmatic tally of deaths induced much controversies and debates. There was no surprise that the death toll was also among the most vexing figures during the 1918 influenza pandemic, with huge variations in mortality across countries beyond 30-fold (Murray et al., 2006). Is it therefore possible to gain insights from the 1918 pandemic, the worst pandemic in recent human history, in face of the daunting mission to keep track of the case and death statistics so that we may emerge stronger from the pandemic of our times? This article examines the possible aetiologies of the inaccuracy of the death toll and review strategies for improvement that could be of use during the COVID-19 pandemic in the months ahead.

Statistics in pandemic times

Governments, institutions and the media are keen to obtain precise figures of deaths because these figures are fundamental in the calculation of the case fatality rate (CFR) and mortality. According to the Dictionary of Epidemiology, the CFR is defined as ‘the proportion of cases of a specified condition that are fatal within a specified time’, whereas a case can be defined ‘based on geographic, clinical, laboratory, or combined clinical and laboratory criteria or on a scoring system (Porta, 2014)’. The death toll used for the CFR calculation of COVID-19 is the number of deaths defined to be due to COVID-19. It is therefore directly influenced by how a case of COVID-19 is defined, for instance, by laboratory or clinical criteria before its progression to death. This death toll is the number most commonly reported by the authorities and media where controversy on its accuracy has arisen. The CFR is not to be confused with mortality, which is defined as ‘an estimate of the portion of a population that dies during a specified time (Porta, 2014)’. The mortality is influenced only by the overall death toll for all causes of deaths during this period, although its surge is most likely attributed to COVID-19 in this context. The CFR and mortality, hence both death tolls, reflect the impact and severity of the pandemic on the society and can be used to guide government responses and policies. The 1918 pandemic took place than a century ago, yet flawed and conflicting data continue perplexing epidemiologists in spite of the many studies done in retrospect.

The 1918 pandemic has been estimated to have infected 500 million people (around one-third of the world’s population of 1.8 billion at that time (Parmet & Rothstein, 2018)), with at least 50 million deaths worldwide (Centers for Disease Control and Prevention, 2019). This results in a CFR of approximately 10%. Most articles in the literature however stated a CFR for the 1918 pandemic of around 2.5% (Marks & Beatty, 1976; Nickol & Kindrachuk, 2019; Taubenberger & Morens, 2006). Such a figure appears to be falsely low, since with 50 million individuals dead due to the pandemic, the total number of infected people would have been 2 billion, which is impossible since the total world population was just 1.8 billion at that time (Jaber, 2020). Murray et al. also reported that mortality varied over 30-fold across countries during the 1918 pandemic, part of which would likely be accounted for by inaccurate death tolls among countries aside from other socioeconomic factors contributing to such variations (Murray et al., 2006). It reflects the difficulty to document precisely the number of cases and deaths in pandemic times, not to mention the COVID-19 pandemic, which is still rapidly evolving daily.

The World Health Organization (WHO), the European Centre for Disease Prevention and Control and Johns Hopkins University are among the groups that published COVID-19 statistics based on information provided by government sources. Nonetheless, their data could still widely differ up to 99,000 cases and 6800 deaths on a day (The Wall Street Journal, 2020a). Such discrepancy revealed the difficulty to obtain precise tallies and demonstrated this universal pitfall during this pandemic. The underlying reasons certainly merit scrutiny in times of escalating public discontent fuelled by the inaccurate death toll and the mounting pressure that governments worldwide are under until such inaccuracy could be accounted for (NBC News, 2020).

Inevitable errors during desperate moments

After reporting a cluster of pneumonia cases, subsequently confirmed to be COVID-19, in Wuhan to WHO on 31st December 2019, China has subsequently publicly shared the genetic sequence of COVID-19 on 12th January 2020 (World Health Organization, 2020). This has been crucial as diagnosis of COVID-19 on suspected cases have been made by means of reverse transcriptase–polymerase chain reaction (RT–PCR) assays on clinical specimens, which are based upon the genetic sequence of the virus.

Unfortunately, due to the overwhelming scale of the pandemic and insufficiencies of testing kits and hospital capacities in the first weeks of the pandemic, many suspected cases did not have the opportunity to be tested and recorded (The New York Times, 2020a). It inevitably led to an inaccurate death toll, which excluded deaths of many untested patients. The overburdened healthcare system in which hospital beds were unable to meet the enormous demands of thousands of suspected newly infected patients per day, so that many patients could only be sent home for self-quarantine exacerbated the problem (Chen et al., 2020). Many patients could have died at home undocumented (The New York Times, 2020b; The Washington Post, 2020).

Such a situation happened in Wuhan, the epicentre of the pandemic, at the onset of the outbreak. Nonetheless, as the reach of the coronavirus extended across continents, countries such as Spain, the U.S. and U.K. which became hardest hit by the pandemic were eventually left in a similar predicament (NBC News, 2020). These experiences illustrated that limitations in testing capabilities and hospital capacities resulted in delayed diagnosis, update of the case numbers and failure to account for cases of death to generate an accurate death toll during the initial and peak phases of the outbreaks when there were rapid surges of cases and deaths.

Effects of increased case detection on the death toll

The scarcity of testing reagents and hospital beds which contributed to the inaccurate death toll were eventually met with measures from the Chinese government. In a time, when the number of suspected cases far exceeded the testing capacity, the Hubei Provincial Health Commission in Wuhan revised on 13th February 2020 the diagnostic criteria of COVID-19 from laboratory-based diagnosis by means of nuclei acids assays to include also the clinical diagnosis based upon computerised tomography (CT) scans for suspected cases. It led to a record high of 14,840 new cases and 242 new deaths on one day (Xinhua, 2020). Makeshift (Fangcang) shelter hospitals were also rapidly established in Wuhan since early February so that most suspected patients could then be treated in designated settings, along with an improved documentation of death cases (Wan et al., 2020). The Wuhan authorities also publicly acknowledged on 17th April 2020 the omission of many fatalities due to COVID-19 for reasons discussed above, and subsequently revised the death toll upward by 50% after a detailed overhaul of various sources (The Wall Street Journal, 2020b).

Similar challenges were faced by many other states and cities affected by the coronavirus. In March, when New York, the epicentre of the outbreak in the U.S., ramped up its testing of the virus, there was a 38% surge of 20,875 new cases overnight (CNBC, 2020). Furthermore, as the capacity of laboratory testing was exhausted, the US Centers for Disease Control and Prevention (CDC) adopted the COVID-19 position statement issued by the Council for State and Territorial Epidemiologists to include both confirmed and probable cases in its tabulation of COVID-19 cases and deaths. The cases were classified according to clinical, laboratory and vital records criteria and epidemiologic linkage. Only cases with confirmatory laboratory evidence were counted as confirmed cases, whilst cases meeting other criteria or epidemiologic evidence were counted as probable cases (Council of State and Territorial Epidemiologists, 2020). This immediately led to a surge of 4,059 probable deaths on 16th April 2020 (CNN, 2020).

The death toll would be underestimated due to the deficiency of testing kits required for laboratory diagnosis if cases are only confirmed by such method, since it would miss a large proportion of cases and deaths resulting from those cases would be unaccounted for. The measures taken by different authorities revealed the necessity of wide scale surveillance and identification of cases by methods in addition to laboratory testing in order to accurately estimate the death toll during a pandemic. This experience may serve to support the use of broadened diagnostic criteria that are based not only on laboratory evidence but also include additional clinical and bedside information to define a case.

Political censorship and manipulation of statistics

At the time of writing, the US and Brazil are the two countries with the highest number of COVID-19 cases, over 2.6 million and 1.3 million of confirmed cases (Johns Hopkins Coronavirus Resource Center, 2020; worldometer, 2020), respectively. Statisticians, public health experts and epidemiologists in these worst-hit nations strove to refine accuracy of case and death tallies and overcome the issue of undercounts due to the aforementioned reasons, yet these efforts of bold revision of the statistics to generate more accurate estimates did not meet with support from local political leaders. Instead, evidence and accusations were there that government officials attempted manipulation of COVID-19 statistics and pressured healthcare systems to alter data towards an optimistic direction.

In the US, President Trump has publicly stated his belief on more than one occasions that the current COVID-19 death toll in the States documented by the CDC and hospitals were inflated and should be lowered, in spite of top infectious disease experts in the country pointing out that such figure was indeed more likely to be an underestimate and the disease has a more pervasive reach than that reflected from the data (The New York Times, 2020c). Meanwhile, as Brazil’s death toll soared to over 35,000 in June, the Brazilian president ordered cessation of public release of total COVID-19 case and death numbers and wiped clean months of cumulative data on COVID-19 from an official website (BBC, 2020; The Guardian, 2020).

Political censorship and reliability of the COVID-19 statistics from China, the initial epicentre, has also been doubted. Four of the authors of this paper are healthcare personnel based in Hong Kong and wish to offer a perspective to serve as food for thought. Currently, around 1.5 million Hong Kong and Taiwan citizens are long term residents working in China, which comprise slightly over 0.1% of China’s population. Meanwhile, COVID-19 death toll from China is 4641 (Johns Hopkins Coronavirus Resource Center, 2020; worldometer, 2020). This should theoretically translate into around five cases of COVID-19 death of such Hong Kong or Taiwan citizens. Should there be even one single case of COVID-19-related death for Hong Kong or Taiwan citizen living in China, it would be highly publicised because of political reasons. As a matter of fact, up to this date, there have been no such reports in Hong Kong or Taiwan. Therefore, while it is acknowledged that there was discrepancy in the COVID-19 death toll during the initial outbreak for reasons discussed in previous sections, our hypothesis herein suggests that the reported Chinese death toll would probably be not far from the actual number.

COVID-19 has brought grave impacts to global economy and induced dissatisfaction of citizens against their leaders as their normal lives and activities were wrecked by this pandemic which has shown no signs of a hiatus within a foreseeable future. Countries such as the US and Brazil which were the hardest hit undoubtedly were devastated by an unprecedented scale of social discontent and financial losses. Coupled with political agenda, these governments were the most tempted to abuse their authority to censor and manipulate data of the pandemic to their advantages, so as to reopen economy, restore citizens’ normal life and solicit for political support. We believe this is a dangerous and unethical manoeuvre. Accurate estimates of death, as discussed in previous sections, allows governments to gauge the severity of the disease, hence appropriate policies and interventions can be implemented to control the pandemic. Furthermore, citizens deserve the right of knowledge of the most accurate data so as to take precautionary measures accordingly. The false sense of safety derived from such erroneous data may jeopardise citizens’ compliance with public health measures, such as social-distancing, personal hygiene and mass masking, which further worsens the outbreaks and socioeconomic impacts. Therefore, we urge governments to refrain from such censorship and manipulation of the data and objectively review the genuine statistics and evidence which would guide them in execution of appropriate public health measures to curb disease spread.

Utility of the mortality data in future

Despite accounting for confirmed and probable COVID-19 cases with a reported death toll of 18,879 between 11 March and 2 May, the CDC revealed on 11 May 5293 excess deaths that were not previously reported for New York City. The agency determined such excess deaths by means of mortality data by the New York City Department of Health and Mental Hygiene. Excess deaths were calculated as the difference between the average death counts in the past five years and the total number of reported deaths in the same period this year (New York City Department of Health and Mental Hygiene (DOHMH) COVID-19 Response Team, 2020). When such method was applied to seven states in the U.S. hard hit by the pandemic, a similar surge of death tolls was observed (Centers for Disease Control and Prevention, 2020). In a recent study, Weinberger et al. applied a similar method across the United States and also found that the excess deaths were 28% higher than the official tally of COVID-19-reported deaths (122,300 vs. 95,235) during the period from 1 March to 30 May 2020 (Weinberger et al., 2020).

Such procedures of the CDC utilised the mortality rate as defined above. Assuming that COVID-19 is the most important factor contributing to a rise in mortality in this period as compared to the preceding years, use of mortality statistics could give valuable information to obtain a more accurate estimate of the COVID-19 related death toll. Since most of the countries around the world should have the data on previous and current mortality, the COVID-19 global death tolls can be estimated with reasonable accuracy by such a calculation and be used as an objective and unbiased method to reflect the death toll in parallel with direct documentation of patients who died with COVID-19.

Learning from the COVID-19 experience

This article serves to help explain controversies and clarify some of the myths regarding the COVID-19 death toll which sparked heated debates within the general population and among governments worldwide. The strategies to incorporate broadened clinical diagnostic criteria to facilitate case surveillance and use of mortality data could be implemented to generate a more accurate death toll in the future so as to guide governments the implementation of public health interventions to manage this outbreak. It is hoped that the international communities could learn from these experiences to arrive at the conclusion that the constant change in death statistics is only natural in times of this rapidly evolving pandemic.

Disclosure statement

No potential conflict of interest was reported by the author(s).