Stephen Macedo and Frances Lee raise profoundly important questions about whether U.S. public health agencies and political leaders met the epistemic and moral challenges presented by the COVID-19 pandemic. On their analysis, the answer is a resounding “no.”

They are most persuasive on failures to live up to the ideals of evidence-based decision-making and reliable science communication. There is no doubt that our information ecosystem struggled during the pandemic, especially in the early months, in ways that affected policy decisions and public choices.

Public opinion on important matters of fact, such as the efficacy of antiviral medications, became highly polarized along irrelevant ideological lines. Preliminary and speculative information shared in scientific preprints, public health statements, news reports, and in some cases published research articles spread widely on social media, even after such information had been refuted or corrected by subsequent research. Mainstream news sources did not always retract claims based on research results when those results were later revised following expert criticism and peer-review.

Even public health officials with training both in science and science communication struggled to separate their best evaluation of scientific evidence from moral judgments about how individuals and government should behave in a pandemic. Several states advised that it was safer to go to a doctor’s office, to work, or to church than to go to the beach or a playground, even though transmission was far lower outdoors than indoors. It also seems clear, in hindsight, that certain common interventions, such as closing schools, had severely negative impacts, the full magnitude of which we are only now beginning to appreciate.

Even so, we think the U.S. COVID-19 response was not as thoroughly flawed as Macedo and Lee conclude. First, the jury is still out on whether non-pharmaceutical interventions (NPIs) of the sort that Macedo and Lee criticize, such as stay-at-home orders, were effective and worthwhile. Several prominent studies of NPIs in the United States and elsewhere show that some NPIs significantly reduced viral spread. To be sure, assessing the benefits of these interventions is complex and controversial; confounding factors, such as differences in demographics, compliance, and culture, may make NPIs more or less effective. At the same time, the true costs of pandemic policies, such as the long-term economic and well-being consequences of learning loss, may not be measurable for decades. More work is needed to fully understand these issues.

Suppose we grant Macedo and Lee’s premise, however: that the answers are clear cut, and that the benefits of NPIs were not worth the cost. Still, there is a difference between retrospectively finding that other actions would have yielded better outcomes and concluding that a decision was a bad one at the time. At best, Macedo and Lee’s arguments about the effect of NPIs on mortality establish that some COVID-19 decision-making led to suboptimal outcomes (with school closures a persuasive example). But this does not establish that public health authorities made bad choices at the time—or that they were driven primarily by political expediency. Sound decision-making requires a balance between costs, benefits, uncertainty, and tolerance for risk within the context that the decision is made. Taking that context into account yields a different perspective on COVID-19 policy.

Early in the pandemic, decision-makers across the world were working with extremely limited information about the nature of the disease, its rate and means of spread, its fatality rate, and its health complications. Data out of Wuhan, China, where the disease was believed to have originated and where the first major outbreak occurred, was very limited, but policymakers elsewhere knew that the disease had prompted local public health officials to initiate a total lockdown for several weeks. Other sources of information, such as data from the cruise ship Diamond Princess in February and early March 2020, suggested that the disease spread rapidly and that more than 1 percent of infections resulted in death.  

Many people’s worst fears appeared to be confirmed in early March, when outbreaks in northwest Italy and New York City overwhelmed hospitals and other health infrastructure. Under these circumstances, there seemed to be little choice but to institute a version of China’s lockdown policy, which at the time appeared to have been relatively effective at preventing the chaos that had struck other cities, which seemed to have been effective to stop the spread on the Diamond Princess. As COVID-19 reached other parts of the United States, Europe, and elsewhere, local authorities in most places implemented similar decisions.

Several factors should be considered when assessing these decisions. One was the information available. Macedo and Lee emphasize that expert opinion, prior to the pandemic, was that quarantines and stay-at-home orders were ill-advised. They criticize public health officials for rejecting their own recommendations. But in March and April 2020, the evidence on which those prior recommendations were based appeared to contradict both the emerging situation on the ground, where radical NPIs did have an effect, and contemporary modeling work performed by epidemiologists using the best-available data about the disease. Both the data and the modeling involved a great deal of uncertainty, but it all pointed in one direction—a direction that apparently superseded the evidential basis for the prior recommendations.

In addition, the measures taken were popular at the time. According to Kaiser Family Foundation polling, even prior to widespread outbreaks in the United States, approximately two-thirds of poll respondents reported making lifestyle changes—such as changing travel plans, abstaining from large gatherings, and stocking up on food or medications—out of concern about COVID-19. By late April, when stay-at-home orders were in place across much of the country, 80 percent of respondents approved of the measures. In other words, the public demanded a strong reaction, including NPIs. Macedo and Lee argue that this was in part due to a failure of public communication about science, especially the risks and benefits of both the disease and the interventions. But against a backdrop of confusion, most people found the known economic costs more palatable than the uncertain health risks, and politicians and other policymakers responded to these judgments.

We also disagree with Macedo and Lee’s portrait of public discourse. As policymaking decisions were made, robust and healthy debate took place around the world. While many countries followed a similar playbook, there were exceptions, such as Sweden. Likewise, while most U.S. states issued stay-at-home orders, some, such as Arkansas and Iowa, never did, and there was considerable variation in how long those orders lasted. By summer 2020, leaders in different regions could make more informed decisions, based on better data, more experience, and, most important, input from their constituents about how best to balance competing values.

Macedo and Lee think this variation reflects preexisting political polarization and had little effect on mortality outcomes. (Was it really so little? In 2021, the five states with the highest age-adjusted fatality rates were consistently Republican-leaning states, whereas the five states with the lowest death rates were Democratic-leaning states. The averages of fatality rates in these groups differ by a factor of three.) But this variation might equally reflect differences in preferences, risk tolerance, and political values—precisely the sorts of cultural differences that might lead different communities to elect different political parties. This is even more apparent in the politics of school closures. By late 2020, these and many other decisions were made at the county or school district level, with input from local parents, teachers, and administrators.

In the end, one extremely important measure of whether political leaders properly balanced the competing considerations that mattered most to their constituents is what happened when they next faced voters. On that measure, the answer is clear. Trump’s chaotic federal response likely cost him the presidency in 2020. But in the first gubernatorial elections after the initial pandemic response, which occurred in 2020 and 2022 depending on the state, only one governor was voted out of office: Nevada’s Steve Sisolak, who narrowly lost to Joe Lombardo. Macedo and Lee may believe that these governors greatly erred in their COVID-19 responses, but voters apparently disagreed.