key: cord-0864030-gtp01rna
authors: Wetzler, H. P.; Wetzler, E. A.; Cobb, H. W.
title: COVID-19: Dying is Bad--Losing Life is Worse
date: 2020-06-12
journal: nan
DOI: 10.1101/2020.06.08.20050559
sha: 1c0e04572e6b8bbdb953b4b6f1f37ef8f1afad9a
doc_id: 864030
cord_uid: gtp01rna

Background: COVID-19 was the leading cause of death in the United States over the three-month period March through May 2020. Another perspective is COVID-19s toll in terms of years of life lost. We calculated years of life lost for COVID-19 and other leading causes of death over those three months in the US. We also predicted years of life lost for COVID-19 and ischemic heart diseases (which includes heart attacks) for March through August 2020. Methods: Years of life lost are the sum of differences between life expectancy at age of death and age at death. Average years of life lost, years of life lost divided by the number of deaths, were also calculated. We used the COVID-19 Projections Using Machine Learning model to predict years of life lost from COVID-19 through the end of August 2020. Results: COVID-19 caused 12,035 more deaths than ischemic heart diseases during March through May 2020 but ischemic heart diseases years of life lost were 1.5% greater than those for COVID-19. Average years of life lost were 10.8 and 12.4 for COVID-19 and ischemic heart diseases, respectively. At the end of August, COVID-19 may overtake ischemic heart diseases as the leading cause of deaths and years of life lost in the US. Conclusion: Each COVID-19 death causes more than a decade of lost life in the US. We are reminded of a Danish Proverb that states "Prediction is difficult, especially when dealing with the future." We suggest that while dying is bad, losing life is even worse.

During the months of March, April, and May of 2020, COVID-19 became the number one killer in the United States (US). 1 Not since the great influenza pandemic of 1918 has there been a contagion with such a nationwide impact. 2 The US recorded its 100,000 th COVID-19 death on May 27. Although most models predict a slowing in the death increase in the coming months and the reemergence of heart problems as the mortality leader, COVID-19 could be the number two cause of death for all of 2020, after ischemic heart diseases (IHD). Moreover, Osterholm suggested that we are in the early phase of the pandemic, and there could be 800,000 COVID-19 deaths in the US over the next 18 months. 3 In 1947 Dempsey stated that deaths do not tell the entire story regarding the seriousness of a disease, and she introduced the concept of "potential years of life lost". 4 A year later Greville stated the justification in simple terms "from the standpoint of society, the death of a person in the prime of life constitutes a greater loss than the death of one of advanced years." 5 Since then years of life lost (YLLs) have been the subject of considerable research perhaps culminating in the Global Burden of Disease studies. 6, 7, 8 Furthermore, the initial metric, years of life lost, has been modified to include disability-adjusted life years (DALYs) and years lived with disability (YLDs) in an attempt to account for the fact that not all life years are equal due disability.

This study had two goals: tally deaths and YLLs for major causes during March through May 2020 (January through March 2020 for influenza and pneumonia due to seasonality) and calculate YLLs for IHD during March through August 2018 and compare them to those projected for COVID-19 over the same six months in 2020.

We used the Centers for Disease Control and Prevention (CDC) WONDER Multiple Cause of Death Online Database, which includes mortality estimates for the United States stratified by age, sex, and cause of death. 9 The latest year in the WONDER Database is 2018. Only deaths from March through May 2018 (January through March for influenza and pneumonia due to seasonality) were used for comparison to COVID-19 deaths because, except for 2020, aggregate causes of death are quite stable from year to year. The maximum year to year change for 2014-2018 was 11.5% . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 12, 2020. . https://doi.org/10.1101/2020.06.08.20050559 doi: medRxiv preprint

Since COVID-19 is basically a single cause of death, we used more discrete categories for comparison instead of the broad ones such as Diseases of Heart or Malignant Neoplasms that are customary. We also included influenza and pneumonia because early on some policy-makers claimed the coronavirus was no worse than the flu. 11 The categories used and their UCD -ICD-10 Codes are listed below.

• COVID-19 (U071)

• Ischemic Heart Diseases (IHD) (I20-I25) The COVID-19 Projections Using Machine Learning model is a susceptible, infectious, and recovered (or deceased) (SEIR) model that uses artificial intelligence to minimize the error between the projected outputs and the actual results. 15 The model's 95% confidence interval . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 12, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 12, 2020. . https://doi.org/10.1101/2020.06.08.20050559 doi: medRxiv preprint decedents in those two categories. Conversely, the average years of life lost are markedly higher for both categories of accidents indicating relatively more deaths among young people. In these data the correlation coefficient between working years of life lost, years lost before age 65, and average years of life lost is 0.95. COVID-19 and ischemic heart diseases together account for over a fifth of the total YLLs. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 12, 2020. . https://doi.org/10.1101/2020.06.08.20050559 doi: medRxiv preprint Discussion COVID-19 killed more than 100,000 people in the US in less than 90 days. Although we do not know the extent to which COVID-19 is displacing other deaths, we do know that most of these were excess deaths that would have taken about 21 months to occur absent the virus. 16 , 17 And, as Mary Dempsey wrote in 1947, deaths fail to tell the entire story. 4 Age at death matters and YLLs measure when deaths occur. IHD has been the leading cause of YLLs in the US since 1990. 18 and that estimate changes slightly every day. That does not make the model wrong given its purpose;

we use the 95% confidence interval of what is likely to occur. Three months of excess deaths and YLLs are bad enough but the model suggests there is a 33% chance of over 200,000 COVID-19 deaths in the US by the end of August. 15 Thus, there is about a 30% chance that by the end of August COVID-19 YLLs will equal or exceed those for IHD over the previous six months. Although we anticipate that IHD YLLs will be the leader for all of 2020, there may be a second COVID-19 wave and we are reminded of Osterholm's forecast that there could be 800,000 COVID-19 deaths in the US by the end of 2022. 3 Several factors may influence our YLL estimates and comparisons. The first factor is the accuracy of underlying cause of death numbers. The tallies of COVID-19 deaths have been criticized for under-reporting. For example, the government counts 26,000 COVID-19 deaths in nursing homes. But because the federal government is not requiring nursing homes to report deaths prior to May, the under-reporting of deaths at nursing homes could be in the thousands. 20,21 On the other hand, there is reason to question whether COVID-19 deaths are being over-counted because of higher payment for COVID-19 diagnosis in Medicare recipients. 22 In comparison, WHO reports that the relative uncertainty for deaths from IHD is about ±12% for high-income countries. 23 The second factor is using 2018 data for causes of death other than COVID-19. Although year to year cause of . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 12, 2020. . https://doi.org/10.1101/2020.06.08.20050559 doi: medRxiv preprint death counts are quite stable, there may be some displacement of deaths in 2020. We will not be able to accurately assess displacement or other changes for six months or more. The third factor is variation in life expectancy which is a random variable but with small standard errors (e.g., approximately 0.04 years at birth in 2017). 24 The final factor is assumed stable age distributions of COVID-19 deaths.

Regarding the accuracy of the data, we do not know the exact numbers for any cause of death. For example, influenza deaths in the US are estimated each year. 25 Cumulative COVID-19 deaths on May 31, 2020 ranged from 98,600 to 106,432 on two prominent tracking sites. 26,27 The Global Burden of Disease Study reported 544,800 IHD deaths in the US in 2016 whereas CDC WONDER included only 363,452-33% less. 18, 9 Yet to dismiss counts, analyses, and models because they lack exact accuracy is a mistake and undermines their purpose. We are well advised to pay attention to Spiegelhalter who wrote "my cold, statistical approach is to wait until the end of the year, and the years after that, when we can count the excess deaths." 28 However, it is not merely an accounting exercise to understand the severity of the current health crisis even if such models are not totally accurate. Timely evidence-based analysis should support institutional policy-making and personal decision-making else those decisions are, by definition, ill-informed. We acknowledge the limitations of this analysis, and hence, will update it in September after the August mortality counts stabilize. We will also include quality of life and comorbidity adjusted YLLs.

COVID-19 killed more than 100,000 people in the US in less than 90 days and we know that most of these were excess deaths that would have taken about 21 months to occur absent the virus.

For at least three months, March through May 2020, COVID-19 was not only the number one cause of death, but it was number two only to ischemic heart diseases in years of life lost. If there is a significant second COVID-19 wave in the fall, 2020 will be the year ischemic heart diseases are dethroned as the number one killer as well as the number one cause of years of life lost since 1990 in the USA. In this context we are reminded of a Danish Proverb that states "Prediction is difficult, especially when dealing with the future." 29 We suggest that while dying is bad, losing life is even worse.

. CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 12, 2020. . https://doi.org/10.1101/2020.06.08.20050559 doi: medRxiv preprint

The Great Influenza: The Epic Story of the Deadliest Plague in History

Infectious disease expert says we're only in the second inning of the pandemic -CNN

Decline in Tuberculosis The Death Rate Fails to Tell the Entire Story

Decline in Tuberculosis: The Death Rate Fails to Tell the Entire Story

Loss of Working Years in Accidents

Premature Mortality in the United States: Public Health Issues in the Use of Years of Potential Life Lost

Global Burden of Disease Study 2017 (GBD 2017) Cause-Specific Mortality

GHDx

Multiple Cause of Death Data on CDC WONDER

Provisional Death Counts for Coronavirus Disease (COVID-19)

Comparing COVID-19 Deaths to Flu Deaths Is like Comparing Apples to Oranges -Scientific American Blog Network

United States Life Tables

the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted

Lives-Saved" From COVID-19 | Avalon Health Economics LLC

COVID-19 Projections Using Machine Learning

Covid-19 excess deaths in the United States through

The state of US health, 1990-2016: Burden of diseases, injuries, and risk factors among US states

Interim Final Rule Updating Requirements for Notification of Confirmed and Suspected COVID-19 Cases Among Residents and Staff in Nursing Homes

Physicians Say Hospitals Are Pressuring ER Docs to List COVID-19 on Death Certificates

On the accuracy of life expectancy

Influenza Surveillance Report (FluView). Centers for Disease Control and

Prevention. Publ March. 2020;6:2020. https://www.cdc.gov/flu/weekly/#S2 Accessed June 2, 2020.. CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)The copyright holder for this preprint this version posted June 12, 2020. https://quoteinvestigator.com/2013/10/20/no-predict/ Accessed June 4, 2020.. CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 12, 2020. . https://doi.org/10.1101/2020.06.08.20050559 doi: medRxiv preprint