[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
Purchase Options:
[Skip to Content Landing]
Views 23,028
Citations 0
July 8, 2020

Challenges Estimating Total Lives Lost in COVID-19 Decisions: Consideration of Mortality Related to Unemployment, Social Isolation, and Depression

Author Affiliations
  • 1Harvard T.H. Chan School of Public Health, Boston, Massachusetts
JAMA. 2020;324(5):445-446. doi:10.1001/jama.2020.12187

The coronavirus disease 2019 (COVID-19) pandemic has claimed hundreds of thousands of lives, directly and indirectly, and threatens to claim many more. Nations have made different policy decisions that have affected the rate of infection, mortality, the economy, and the life of the country differently. The choices between various alternative policies have led to different trade-offs between what are arguably incommensurable goods, such as survival, mental health, social connection, and economic growth. It can seem difficult or impossible to weigh these numerous factors, yet policy decisions must be made, with countless implications for society. In the early stages of the pandemic, and when information was limited, a cautious approach was arguably most appropriate. As further information becomes available, it becomes possible to make better-informed decisions. However, the inherent challenges involved in the very real, and very difficult, trade-offs remain.

One approach to weigh these different outcomes, which are difficult to directly compare, is to attempt to use a composite measure such as well-being–adjusted life-years.1 In such an approach, each year of life saved or lost is weighted by a self-assessed overall life-satisfaction score (range, 0-10). This approach, however, has several drawbacks. First, it is difficult to weigh aspects of well-being against survival. Second, many conceptualizations of well-being involve numerous components.2 While a number of well-being approaches prioritize life satisfaction,1 it is unclear why this aspect of well-being ought to be prioritized over others, such as having a sense of meaning in life. If different aspects of well-being were assessed (such as meaning and purpose), this would then lead to different assessments of well-being–adjusted life-years. Third, using well-being–adjusted life-years based on life satisfaction is also problematic insofar as it may deprioritize the lives of individuals who are poor, disabled, and vulnerable by giving their lives less weight because their life satisfaction tends to be lower.

An alternative would be to use a “total lives saved” approach that prioritizes life as the highest present good at stake and requires that decisions be based on lives saved alone. This has been proposed in recent ethical statements concerning clinical care3 but is arguably applicable more generally in policy decision-making. A total lives saved approach does not mean that economic, social, and well-being outcomes are to be neglected—in fact, these factors also affect mortality rates, sometimes substantially. Meta-analyses of covariate-adjusted longitudinal cohort studies indicate that unemployment, social isolation and lack of community, and late-life depression are all associated with increased all-cause mortality.2,4-7 The magnitude of associations may, in some cases, be sufficient to considerably shift the evaluations of lives saved or lost when considering different policies.

For example, if everyone younger than 60 years in a particular region who was healthy returned to work but an additional 500 people died of COVID-19 infection, is that a reasonable trade-off, and would it be offset by avoiding the mortality consequences of unemployment and social isolation?4,5 Another scenario highlights the challenges of trade-offs: In the fall of 2020, decisions will need to be made about children returning to school. Although deaths attributable to COVID-19 among children are exceedingly rare, children who become infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) could infect older individuals. Should millions of children be kept out of school, with substantial and likely lifelong consequences, to “save” 500 or 1000 or 10 000 lives? Do the consequences of keeping children out of school, and potential future adverse effects on health and longevity,2,4 outweigh the consequences of infection-related deaths, assuming that as much as possible is done to protect older adults, health care workers, and other individuals who ensure a functioning society?

The implementation of such a total lives saved approach faces potential, although not insurmountable, challenges. First, the associations of employment, social isolation, and depression with mortality are likely to vary by context. The association between unemployment and mortality may be more pronounced in midlife than in later life. Conversely, depression may have a stronger association with mortality later in life rather than earlier.

Second, while these meta-analyses were based on prospective cohort data, with multivariate-adjusted covariate control, the data are observational and subject to confounding. It would be important, when using meta-analytic estimates to inform policy decisions, to also consider sensitivity analyses that set the values of these relationships to somewhat lower levels than the estimates, to examine whether conclusions change, and to provide estimates that individuals can understand.

Third, and perhaps most challenging, it will be difficult to assess and compare the influence of various alternative self-isolation, social distancing, contact tracing, and other policies on employment, social isolation, and depression. While decades of data are available to inform the association of unemployment, social isolation, and depression with mortality,4-7 much less data are available regarding the relationships of pandemic policies with these social, psychological, and economic outcomes. Moreover, what is important to assess with regard to decision-making is not so much how the pandemic influences these social, psychological, and economic outcomes but rather how different policies for handling the pandemic may influence these outcomes. Even without strict isolation and workplace closure measures, the economy and social relations would still have been adversely affected by the morbidity and mortality related to COVID-19 and the substantial changes in individual behaviors.

These aforementioned challenges, however, are not necessarily an insurmountable barrier to this total lives saved approach. Within the past months, different countries, regions, and cities have made different decisions, essentially resulting in a series of “natural experiments.” From these experiences and decisions, it is possible to begin to assess the association between different policies and social, psychological, and unemployment outcomes, as well as SARS-CoV-2 infection rates and COVID-19 fatality rates. However, such data must be used carefully, because different countries, regions, and cities may differ from one another in a host of other ways; some of these factors may not be possible to control for; and societies may prioritize various outcomes in different ways. Nevertheless, a series of analyses, as rigorous as possible, to evaluate various policies, drawing on country-by-country, region-by-region, and city-by-city comparisons, may give considerable insight into the association between implementation of various policies with social, psychological, and economic outcomes.8 From estimates of these outcomes, it may be possible to extrapolate to the mortality consequences over time. Simply assessing current total excess mortality rates is insufficient because it will take time for the mortality consequences of unemployment, isolation, and depression to become manifest. More direct mathematical and theoretical modeling of these effects may also give further insight, as has occurred with modeling for infection rates and fatality rates. It will again be important, when using the estimates to inform policy decisions, to also consider how sensitive conclusions are when the strength of the relationships between various policies and outcomes are set to values different from their best estimates.

In the months ahead, while confronting the possibility of a second wave of the pandemic, these calculations of total lives lost, both from SARS-CoV-2 infection and because of social, psychological, and unemployment outcomes, may prove important in policy decisions. Other factors associated with mortality also will require consideration, such as delayed treatments, for example, for heart disease or cancer,9 and prolonged absence of preventive health care and vaccinations.10 However, there may come a point at which the number of lives lost from economic, social, and psychological consequences of different policy decisions will outweigh the number of lives lost from infection, and it will be crucial to consider the indirect mortality consequences of these policy decisions. When reasonable and rigorous sensitivity analyses and variations of the parameters indicate that this point has been reached, it would be a mistake to ignore these other considerations. Moreover, this approach of using only total lives saved or lost is effectively conservative in its deference to infection-related fatalities because it does not directly take into account social and other goods but rather places them subordinate to life. With new unemployment claims in the US alone reaching 40 million, and with unemployment associated with all-cause mortality, it may be time to more seriously, and quantitatively, take the social, economic, and psychological consequences of policies into account in decision-making when calculating total lives lost from the COVID-19 pandemic.

Back to top
Article Information

Corresponding Author: Tyler J. VanderWeele, PhD, Harvard T.H. Chan School of Public Health, 677 Huntington Ave, Boston, MA 02115 (tvanderw@hsph.harvard.edu).

Published Online: July 8, 2020. doi:10.1001/jama.2020.12187

Conflict of Interest Disclosures: Dr VanderWeele reported receiving personal fees from Aetna Inc and grants from the John Templeton Foundation.

Additional Contributions: I thank Howard Koh, MD, MPH, Harvard T.H. Chan School of Public Health, and Jan Vandenbroucke, MD, PhD, Leiden University Medical Center, for helpful comments and suggestions on the article. Neither of these individuals received any compensation for their contributions.

Layard  R, Clark  AE, De Neve  J-E, Krekel  C, Fancourt  D, Hey  N, O’Donnell  G. IZA DP No. 13186: When to release the lockdown? a wellbeing framework for analysing costs and benefits. IZA Institute of Labor Economics. Published April 2020. Accessed June 26, 2020. https://www.iza.org/publications/dp/13186/when-to-release-the-lockdown-a-wellbeing-framework-for-analysing-costs-and-benefits
VanderWeele  TJ.  On the promotion of human flourishing.   Proc Natl Acad Sci U S A. 2017;114(31):8148-8156. doi:10.1073/pnas.1702996114PubMedGoogle ScholarCrossref
Joint statement: moral guidance on prioritizing care during a pandemic. Public Discourse. Published April 5, 2020. Accessed June 8, 2020. https://www.thepublicdiscourse.com/2020/04/62001/
Roelfs  DJ, Shor  E, Davidson  KW, Schwartz  JE.  Losing life and livelihood: a systematic review and meta-analysis of unemployment and all-cause mortality.   Soc Sci Med. 2011;72(6):840-854. doi:10.1016/j.socscimed.2011.01.005PubMedGoogle ScholarCrossref
Holt-Lunstad  J, Smith  TB, Baker  M, Harris  T, Stephenson  D.  Loneliness and social isolation as risk factors for mortality: a meta-analytic review.   Perspect Psychol Sci. 2015;10(2):227-237. doi:10.1177/1745691614568352PubMedGoogle ScholarCrossref
Li  S, Stampfer  MJ, Williams  DR, VanderWeele  TJ.  Association of religious service attendance with mortality among women.   JAMA Intern Med. 2016;176(6):777-785. doi:10.1001/jamainternmed.2016.1615PubMedGoogle ScholarCrossref
Wei  J, Hou  R, Zhang  X,  et al.  The association of late-life depression with all-cause and cardiovascular mortality among community-dwelling older adults: systematic review and meta-analysis.   Br J Psychiatry. 2019;215(2):449-455. doi:10.1192/bjp.2019.74PubMedGoogle ScholarCrossref
Goodman-Bacon  A, Marcus  J. Using difference-in-differences to identify causal effects of COVID-19 policies: DIW Berlin discussion paper No. 1870. Published May 18, 2020. Accessed June 8, 2020. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3603970
Sud  A, Jones  M, Broggio  J,  et al.  Collateral damage: the impact on outcomes from cancer surgery of the COVID-19 pandemic.   Ann Oncol. Published online May 16, 2020. doi:10.1016/j.annonc.2020.05.009PubMedGoogle Scholar
Stephenson  J. Sharp drop in routine vaccinations for US children amid COVID-19 pandemic. JAMA Health Forum. Published May 12, 2020. Accessed June 26, 2020. https://jamanetwork.com/channels/health-forum/fullarticle/2766119
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    3 Comments for this article
    Estimating Total Lives Saved and Lost from COVID-19
    Michael McAleer, PhD (Econometrics), Queen's | Asia University, Taiwan
    The cogent and informative Viewpoint by a respected academic on the challenges in estimating the total lives saved and lost based on economic, social connection, psychological, and medical considerations is essential reading for everyone involved in public health care policy decision making of the extent of the damaging effects of COVID-19 on modern society.

    Economic, social, psychological, and medical differences arise according to different cultures, countries, regions, and cities, and in how to deal with the first and second waves of the COVID-19 pandemic.

    Difficult and necessary trade-offs among economic, social, psychological, and medical considerations in protecting the physical and
    emotional health of individuals, versus opening up society and the economy, are essential challenges to enable society to function effectively rather than descending into chaos.

    There are serious problems associated with the development of a composite measure of "well-being–adjusted life-years", which depends on arbitrary and untestable assumptions regarding the quality of life and life satisfaction, especially for the poor, disabled, and those who are least capable of looking after themselves.  

    Similar difficulties are associated with a “total lives saved” approach, in that questionable assumptions must be made regarding economic, social, psychological, and medical measurements, which can be problematic in themselves regarding their accuracy.

    Community responses are essential in determining the acceptable trade-offs among economic, social, psychological, and medical options, which would lead to a range of assumptions in the associated models, before sensible, informed, and acceptable public policy considerations can be reached and implemented.

    This may be difficult during the COVID-19 pandemic, but as the new norm it is essential that any underlying assumptions in alternative modelling approaches be easy to understand, interpret, and communicate so that the associated predictions and prescriptions will not be dismissed by the general public, social media and (especially) politicians as yet another unrealistic set of ill-considered recommendations by academic researchers.
    Lives vs Money (and Lives!!!)
    Gary Ginsberg, DrPH, MSc (Econ) | Braun School of Public Health, Hebrew University, Jerusalem
    As the authors correctly point out, its not only health vs economy, but it's actually health vs (economy and health). Israel (population 8.7 million), has had a relatively low death toll from COVID of 343 souls. Having successfully implemented a lockdown to almost eliminate the first wave (at one stage down to 5 new cases a day), we are now in the midst of a huge second wave (around 1,000 cases a day) and a further lockdown remains a possibility. Data from Scotland (Clemens et al. EJPH 2015,25,115) showed age-adjusted relative risk of mortality in unemployed males and females to be 1.85 and 1.51 respectively. Applying these data to the Israeli population, I estimate that if a lockdown reduces employment by 70% then 1,700 lives will be lost to unemployment-related factors during the next 12 months. This exceeds the estimated deaths from COVID that are liable to be prevented.

    If employment is reduced by 50% then there will be around 505 extra unemployment-related deaths. I know it's hard to think across sectors, but we have to at least (as the authors suggest) start combining cross-sectorial data on which to make our decisions.
    Widespread and Prolonged Unemployment Rates as Risk to Democracy: Lessons from the 20th Century
    David Gurwitz, Associate Professor | Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel
    This is a superb opinion article that is a 'must read' for public health strategy planning teams globally and others in the health policymaking chain. This includes Parliament/Congress/Senate members in all democratic nations. The aspects discussed in this article must be addressed and carefully taken into consideration when deciding on lockdowns and on financial support systems for assisting unemployed individuals and affected private sector businesses.

    Alas, given our incapacity to predict when an effective and inexpensive vaccine would become widely available, as well as future SARS-CoV-2 mutations and their effects on COVID-19 fatality rate, it is next
    to impossible to calculate numbers of lives saved vs. lives lost due to prolonged lockdowns.

    Moreover, a key aspect entirely absent in this otherwise elaborate article is that widespread and prolonged unemployment rates tend to destabilize democratic nations, which is what fueled the rise of the Nazi Party to power during the period leading to the Third Reich. Such events may lead to far higher numbers of lost lives compared with COVID-19. Sadly, there are already signs that in some nations such events are conceivable.