Consideration of Trade-offs Regarding COVID-19 Containment Measures in the United States

Background: The economic stimulus package in the United States, which totalled $2.48 trillion, was designed to soften the economic impact of sweeping containment measures including shelter-in-place orders that were put in place to control the COVID-19 pandemic. Methods: In healthcare, interventions are rarely justified simply in terms of the number of lives saved but also in terms of a myriad of other trade-off factors including value-for-money or cost-effectiveness. Cost-effectiveness analysis was therefore conducted as the cost per life-year gained (Cost/LYG) from the containment measures adopted based on several different projections of the baseline number of deaths in the absence of any containment measures. Reductions in premature mortality due to the shutdown (i.e. the difference between years of life lost relative to life expectancy under the shutdown and no shutdown scenarios) were used to calculate changes in health status. Given that men and women have different life expectancies, the analysis calculates premature mortality for men and women by age bracket. Results: The results showed seven different scenarios that reflect different death projections. It showed that as the projected number of deaths increases, the cost-effectiveness of the containment measures becomes more favourable i.e. providing better value-for money for US taxpayers. Cost-effectiveness ranged from $180,874 per life-year-gained for the high-end projection to $4,258,780 per life-year gained for the low-end death projection estimate. Conclusion: Incremental costs per life-year gained related to the economic shutdown can span a wide range depending on the baseline number of deaths in the absence of any containment measures. The results show that in the US, under no scenario for life-years gained does the stimulus package compare favourably to other healthcare interventions that have had favourable cost-effectiveness profiles. However, when comparing value-of-statistical-life-year (VSLY) threshold measures used in other sectors, it is plausible that the U.S. stimulus package could be viewed more favourably from a cost-effectiveness perspective. Given the wide-ranging impacts that COVID-19 has had on American life, it would seem that the comparison should be made to experiences in multiple sectors and on this basis, it appears that the shutdown is likely to represent good value-for-money.


Introduction
SARS-CoV-2 or COVID-19, a novel coronavirus, comes from a family of zoonotic viruses that can lead to severe respiratory symptoms. Past corona virus epidemics included Severe Acute Respiratory Syndrome or SARS-CoV resulting in 744 deaths and 8,098 cases worldwide and the Middle Eastern Respiratory Syndrome MERS in 2012 resulting in 862 deaths and 2,506 confirmed cases worldwide (NHS 2014;WHO 2020a). COVID-19 was first documented in China on December 31 st 2019 with the first death on January 11, 2020. Since then, it has rapidly spread globally and on March 11, 2020, the World Health organization (WHO) officially declared a pandemic. The last pandemic was the H1N1 flu in 2009 that resulted in 400,000 deaths. As of April 14, 2021, there were 137 million cases and over 2.95 million deaths worldwide and 31 million confirmed cases including 557,000 deaths in the U.S. due to COVID-19 (WHO, 2020b).
This highly-contagious spread of the virus has been unexpected and has left the global community unprepared. In an effort to control the spread, countries including the US has put forward stringent containment measures (e.g. social distancing, travel restrictions and quarantine provisions) in an effort to collectively slow down the spread of the contagion. In the US, all foreign nationals who exited China two weeks prior to arriving in the U.S. were banned from entering beginning January 31 st , 2020. The first U.S. death was reported on February 7 th in California followed by the more-often reported death in Washington on February 29 th , 2020. Further travel restrictions were put in place when the US banned all travellers from 26 European countries on March 11 th, 2020, and two days later, the U.S. declared the outbreak as a national emergency. By April 7 th , 42 states issued shelter-in-place orders with nearly 95% of Americans under lockdown. By April 20 th , protestors started anti-lockdown rallies throughout the US and in many other jurisdictions (Mervosh et al., 2020).
The instituted containment measures included lockdown and shelter-in-place orders, domestic and international travel restrictions, self-isolation, quarantine, closure or restriction of all nonessential businesses, school closures and limits on public gatherings. These have resulted in large economic consequences for the United States. The result is best exemplified by the contraction of the U.S. GDP by 4.8% during the first three months of 2020a contraction not seen since the Great Depression of the 1930s when the economy contracted by 13% in the wake of the stock market crash of 1929. U.S. unemployment increased to 14.7% by the end of April with a total of 33.5 million Americans filing for unemployment since March 13th when the state of emergency was declared (U.S. BLS, 2020). Further, these figures do not include the rise in underemployment owing to reduction in work-hours or increases in job-sharing. Many U.S. employees and their families also lost their health insurance coverage along with their loss of employment. There was and continues to be a particularly large impact on workers manning the frontlines of the restaurant, retail and transportation industries and an associated likelihood of permanent job loss due to sudden structural changes in the economy brought on by the pandemic. The impact also goes beyond the economy with school closures and its potential effects on students' educational trajectory and negative health impacts resulting from delayed elective surgeries and routine medical checkups including the inability for timely access to therapies and diagnostic tests and child vaccinations. Finally, there is the impact on mental health as the combination of the economic downturn, job losses, social isolation, stress and anxiety could have led to possible increases in rates of depression and in behaviors such as gambling, domestic violence, alcohol and drug addictions. In addition, those who had been treated in ICUs could have experienced some form of post-traumatic stress disorder and other physiological issues (McKie, 2020).

Trade-off Decisions
In an effort to counteract the damage to the economy as a result of these containment measures, the US government put in place a massive, unprecedented stimulus package totalling approximately US $2.5 trillion. It is important to note that these stimulus packages merely provide stabilization rather than long-term stimulus; as a result, there has been growing public pressure to reopen the economy even though there is no evidence of a sustained levelling off of the pandemic. Information regarding the magnitude of trade-offs between economic effects and health effects become crucial to determining the path forward as governments consider both strategies to loosen the current lockdown and to determine further economic measures needed to manage the current crisis. State governments may have reopened their respective economies prematurely prior to a sustained reduction in new daily cases and without proper monitoring or testing capacity needed to avoid a resurgence in the number of COVID-19 cases.
Trade-off decisions are not new to healthcare. When it comes to selecting optimal therapy for patients, trade-offs are an explicit part of all levels of decision-making. This includes decision-making at the bedside with or without patient involvement by the physician and decision-making by regulatory agencies to approve new interventions after weighing both efficacy and safety elements (Neumann et al., 2016;US FDA, 2019). Finally, some reimbursement agencies internationally evaluate costeffectiveness of interventions to determine whether they provide reasonable value-for-money before making funding decisions on coverage and formulary inclusion. Interventions that do not meet the required threshold for value-for-money often can be denied approval for funding leading to reduced access to therapies.

Methods
The economic impact of the pandemic could possibly computed with greater precision as complete data becomes available. However, there is value in assessing the value at the point when restrictions start to ease and projections are consequently adjusted upward. One method of measuring trade-offs could be in the form of a simple cost-effectiveness analysis (CEA), calculated as the cost per life-year gained (Cost/LYG) from the containment measures adopted and in force through the end of April, 2020. In this regard, there are no scientifically rigorous standards regarding thresholds that represent good value-formoney from a societal perspective (Garber and Phelps, 1997). Though current economic evaluation methods recommend evaluating therapies from a societal perspective, most CEA and associated methodologies have currently focussed on single interventions from a payer-perspective (Drummond et al., 2005). Attempting to conduct these forms of analyses outside of clinical trial settings is challenging as the impacts to society are wide-ranging with too many unknowns in order to model effectively (Weatherly et al., 2009). However, if the model is defined by what is known currently, it is possible to perform a high-level analysis using concepts of cost-effectiveness especially given that the costs and outcomes are defined within a short time window.
In its basic form, cost-effectiveness is a form of analysis that compares the difference of costs and effects between an intervention and baseline standards. The incremental cost-effectiveness ratio (ICER) is therefore as follows: For the purposes at hand in this paper, the intervention is considered to be the pandemic containment measures collectively in place and the changes in health status refers to reductions in premature mortality due to the shutdown (i.e. the difference between years of life lost relative to life expectancy under the shutdown and no shutdown scenarios). Given that men and women have different life expectancies, the analysis calculates premature mortality for men and women by age bracket.
To calculate the change in life years lost due to the shutdown, it was assumed that the distribution of deaths across society by age bands would have been similar under a no shutdown scenario compared to what was observed under the economic shutdown. In these calculations, the midpoint of each age category is taken as the age of all those who died in that category except an age of 20 years was assigned for the first age bracket (<1-24) and 85 years for the final bracket (>=85). These ages were also attributed to the distribution overlaid on the different projections for a no-shutdown scenario. The reduction in life-years lost between the shut-down and no-shutdown scenarios--based on current remaining life-expectancies for males and females--produced the effects due to the shutdown (i.e., lifeyears gained under the shutdown).
In this regard, there will be a degree of uncertainty in estimating baseline projected deaths in the = 3) in the absence of any containment measures (Ferguson et al., 2020). Another projection from the University of Nebraska Medical Center estimated 480,000 deaths in the absence of any containment measures (Zoellner, 2020). These estimates show the large range in potential mortality in the absence of containment strategies. To reflect a wider possible range given the uncertainty in baseline, the number of potential deaths used for baseline projections in the forgoing analysis ranged from 200,000 to 3 million to span the full range of possible scenarios.
To estimate premature mortality, projections of mortality under the shutdown need to be determined alongside the already discussed projections in the absence of a shutdown. The institute of Health Metric and Evaluation (IHME) projections favored by the Trump administration originally projected 73,433 deaths by August 4 th , 2020 (IHME, 2020). This projection, made at the end of April 2020, did not account for the announced easing of restrictions and assumed social distancing measures remained in place until R0 fell below one. (R0 is the measure of reproduction i.e. the average number of people who will contract COVID-19 from one infected person). This estimate with the overlay of age-sex category mortality will be used as a point estimate in estimating mortality experienced under the shutdown. The model has since been modified to project 134,475 deaths by August 4 th , 2020 and greater than 201,000 by October 1rst to reflect the continued relaxation of some of the measures in some U.S. states including the opening up of certain businesses and public spaces (IHME, 2020). Since the number of deaths and cases is continually increasing, this model uses a cut-off point of August 4th, 2020 when the IHME model suggests that deaths would have plateaued during the current wave had containment measures remained in place.

Data
The largest impact from these pandemic containment strategies is on the economy. Containment measures have resulted in closures of businesses, both small and large corporations and air and land travel. Quantifying exact costs of the impact on the economy is challenging given the ongoing pandemic and its wide scope. The best proxy for costs associated with the pandemic containment measures is therefore the costs of the stimulus package that was put forward by the U.S. government in an effort to minimize the damage to corporations, small businesses, and recently unemployed individuals affected by the shutdown. Even though these measures might still be viewed as insufficient, they are the best quantifiable proxy for lost economic productivity at the present moment. public services including hospitals, foodbanks, stockpiled equipment, child nutrition and veterans ( Figure   1). The package also included the one-time cash payment to all qualified Americans which technically represent a tax credit to offset future income taxes (U.S. Congress 2020a and 2020b). The detailed breakdown of the stimulus package is shown in Table 1 with a proportional breakdown illustrated in Figure 1.
--Insert Table 1 here---Insert Figure 1 here--Mortality data shows the age distribution of pandemic-associated decedents (Table 2) with a gender distribution of 56.6% male to 43.4% female. The gradient of mortality shows that decedents have primarily fallen in the over age 65 category with higher proportion of deaths in males in the younger age brackets (NCHS, 2020).

RESULTS
Using these data, we obtained high-level estimates for the cost per LYG based on seven different scenarios with different death projections ( --Insert Table 3 here- The most commonly referred cost-effectiveness threshold in the US is $50,000 which was the costeffectiveness calculated for end-stage renal dialysis when it was added to the U.S. Medicare program in the 1970s (Neumann et al., 2014). Inflating this amount to present times would imply a threshold of $150,000 which is currently used to set a value-based price in many cases (Neumann et al., 2014).
Thresholds of up to $500,000 have also been referenced for rare diseases (Garrison et al., 2019). Outside of the health sector, thresholds may also be substantially higher (Brennan, 2016;Hirth, 2020).
Using the conventional threshold of $150K/LYG in the health care sector to establish favourable costeffectiveness profiles, the results shown in Table 3  be considered to make more definitive statements regarding value-for-money of the economic shutdown-a topic that will be expanded upon in the following discussion.

Discussion
This paper helps to frame the advisability of whether a large economic stabilization program in the wake of a pandemic-such as that for COVID-19--represent value-for-money. Rather than simply looking at It is possible that the shut-down may have actually produced a net cost-savings to society rather than involving a trade-off of dollars for lives saved. Nationally, it is estimated that Americans who contract COVID-19 over age 60 could lose an average between 153 to 222 days of life expectancy while those under 40 would lose an average of two weeks with the total value of VSL lost without containment measures in place of between $8 to $60 trillion (Wilson, 2020) If the estimates of the stimulus package of $2.48 trillion stand as a good proxy for the costs of the shut-down, then it would appear that this initiative actually produced $5.5 trillion in savings along with at least 583,000 life-years gained (See Table 3); that is, there was not actual trade-off but a net actual cost savings to society from saving lives.
This result is consistent with the results of a recent draft working paper that estimated $5 trillion in net benefits from current containment initiatives (Thunstrom et al., 2020).
Given both the highly contagious nature of COVID-19 and limits on testing for the virus, the low-end estimate for the total number of deaths may be a vast underestimate with many COVID-19 decedents remaining unidentified. In addition, beyond the cost of the stimulus package, there may be other cost considerations that include additional investments made by the federal government not contained in the stimulus package, investments and assistance made by state and municipal and city governments, non-profit organizations, charitable groups including foodbanks, other financial investments made directly by corporations, the deployment of the military to build temporary hospitals, and other services-in-kind and charitable funding from foundations and private citizens. The analysis also does not take into account the decreases in quality-of-life and morbidity associated with hospitalizations and shelter-in-place orders that may likely extend well-beyond the time perspective of this analysis and involve extensive often-unmeasured costs to the health care system particularly near the end of life (Dao et al., 2014) There are other ramifications including mental and physical sequalae including benefits and costs associated with short-and long-term behavioural changes associated with the pandemic shock. Lastly, there is a large degree of uncertainty on the range of death forecasts given no intervention with an associated wide range of subsequent impacts upon society.
As the Ro trends downward and more high-quality data becomes available for each jurisdiction, it may be possible to conduct future research that focusses on the development of a population-based, longterm cost-effectiveness analysis from a societal perspective. A comprehensive analysis will enable the capture of both direct and indirect downstream costs including potential mortality effects of the economic downturn, health outcomes and quality of life (Ariizumi, 2012). It can also take into account the broader costs associated with the pandemic response including unemployment and lost business production. Such increased detail will determine whether the containment measures that resulted in the economic downturn were too broad or narrow to achieve optimal value-for-money.
In the end, this analysis is focussed on the short-term and therefore not all costs and effects are included in this analysis. The large question is whether the magnitude of missing effects is larger than the magnitude of missing costs or vice versa. If the former, then it might be possible that the true value-for-money of the stimulus package might be substantially more favourable than what this analysis shows; however, if the latter is true, then the cost-effectiveness profile of the stimulus package may be even more unfavourable than what has been presented.

Conclusion
The economic stimulus package under the U.S. CARES Act (2020) and the subsequent U.S. Paycheck Protection Program and Health Care Enhancement Act (2020) is unprecedented in terms of its magnitude suggesting that the U.S. economic shutdown was also unprecedented in terms of its cost to the U.S. economy. There are a variety of ways to justify the US shutdown, not least of which was to avoid overwhelming the hospital sector in the U.S. which has experienced significant pressures in March and April, 2020. Other measures of justification are also warranted based on trying to compare the costs with the benefits in terms of life-years gained from the economic shutdown compared to a scenario of no containment efforts. Based on this criteria, it is unclear as to whether the economic shutdown was a worthy endeavor depending on whether the shutdown was compared to interventions in the health sector or provisions that have been put in place to protect human life outside this sector.
Given the wide-ranging impacts that this pandemic has had on American life, it would seem that the comparison should be made to experiences in multiple sectors rather than just the health sector, and on this basis, it appears that the shutdown is likely to represent good value-for-money.     Congress. 2020