To the editors:

A famous aphorism, often attributed to Voltaire, states that “the perfect is the enemy of the good.”¹ This is usually understood to mean that one should not fail to take actions that have positive effects while waiting for an ideal solution that may never arrive. In their essay, Jay Bhattacharya and Mikko Packalen make a very different argument: contact tracing is imperfect and therefore useless. This position is not just wrongheaded; it is potentially deadly.

Bhattacharya and Packalen’s main claim is that contact tracing is not worthwhile in all but unique circumstances because, even when contact tracing is employed, transmission still continues. They imply that the primary purpose of contact tracing is disease eradication and that contact tracing is futile if transmission is widespread. This is a misunderstanding of the purpose of contact tracing and reveals an error by the authors in their thinking about transmission.

It is true that transmission will continue, even in the context of contact tracing, when infection is already widespread and programs are imperfect—and they all are. It is also true that contact tracing is hard. It requires difficult work from public health teams, as well as dialogue and collaboration with the communities they serve. These challenges are not unique to contact tracing. Indeed, they are also true of most public health interventions aimed at reducing morbidity and mortality. Bhattacharya and Packalen’s argument suggests that public health interventions should only take place if that intervention alone will stop all transmission and if it will also be easy to carry out. If this is truly the case, authorities should neither pursue contact tracing to reduce risk of SARS-CoV-2 infection, nor invest in the deployment of any SARS-CoV-2 vaccine. The latter will almost certainly not provide perfect protection for the community as a whole and will lead to numerous logistical challenges. Such an argument also implies that some of the largest global public health initiatives are essentially futile, including vaccination programs against measles and polio, bed net interventions to prevent malaria, and condom use to prevent sexually transmitted infections.

The evidence from these programs is clear: imperfect interventions make vast improvements in public health every day. For this reason, it is curious that public health experts would argue against contact tracing due to its imperfections. Bhattacharya and Packalen may be unfamiliar with this evidence. Technical inaccuracies in their description of how contact tracing typically works suggest a limited understanding of these approaches. The authors claim that if contacts take a PCR test and the test is negative, those contacts can leave quarantine even though the test may miss many infections. But the United States’ Centers for Disease Control guidelines about quarantine do not allow people to test out of quarantine due to the limited sensitivity of PCR in the early stages of infection. Bhattacharya and Packalen further argue that PCR’s false-positive problem, that persons infected may sometimes continue to test positive after they are no longer infectious, may also limit effectiveness. This claim misses a crucial point: such persons were likely recently infectious, and the focus of contact tracing is not the person being tested, but contacts who were unknowingly exposed.

Another unsupported claim made by the authors is that contact tracing could, in fact, make matters worse by discouraging testing. There is no evidence to suggest that contact tracing programs have ever increased the incidence of any disease. The authors argue that since a positive COVID-19 test can also lead to enforced separation and quarantine for one’s friends and family, people will attempt to avoid these negative consequences by refusing testing, even if they become ill. One’s contacts may indeed be asked to quarantine as the result of a positive test. But if they are not notified of their exposure, they may unwittingly infect their own friends and loved ones, a circle that could include people with an increased risk of severe illness, or even death. These are strong positive incentives. Who would not be concerned about infecting someone else, particularly a friend or family member, when that infection may prove fatal? Rather than a decline in demand, evidence from across the United States and elsewhere suggests that demand for fast testing frequently outstrips supply.

Contrary to the claims made by Bhattacharya and Packalen, contact tracing is not a monolithic intervention and can be adapted to the different stages of an epidemic. When case numbers in an epidemic are near zero, extreme measures to identify every possible exposure may be warranted and effective in preventing a potential resurgence. When case numbers are high, it may be more effective to focus on quarantining the highest risk contacts and retracing their movements to identify settings that are responsible for amplifying transmission. In each case, the role of contact tracing differs, but it nonetheless remains a critical element of the overall response.

The arguments put forth by Bhattacharya and Packalen are troubling for numerous reasons. Bhattacharya is one of the primary signatories to the Great Barrington Declaration, which argues that the general population should resume their daily lives and that SARS-CoV-2 should be allowed to spread widely in the community. This is a misguided rush to achieve herd immunity. To attain this outcome with minimal loss of life, the signatories argue that special attention should be paid to protecting the most vulnerable. In this essay, Bhattacharya argues the opposite, recommending that communities discard one of the primary tools for fighting a circulating virus and protecting vulnerable groups. His overarching argument seems to be that any inconvenience associated with public health responses to the COVID-19 pandemic is not worth the cost. Instead, the community should simply sit back and let nature take its course. Such a position evinces little regard for the millions of deaths that would likely occur as a direct result.

The fight against COVID-19 will be a long one. No single tool, even a vaccine, will end the pandemic and the most appropriate responses will change constantly as the situation unfolds. Taking contact tracing, or any other approach, off the table is unwise and reduces the chances of controlling the crisis. Instead, more effort should be invested in developing precisely calibrated methods to control epidemics that spare the public from shutdowns and social distancing. Test-trace-isolate programs are one of the best examples of these focused approaches and have proven their effectiveness over decades of outbreak responses. Far from being deemed futile and discarded, they should form a core pillar of the containment measures.

Emily Gurley, Justin Lessler, Andrew Azman, and Lauren Kucirka

Jay Bhattacharya and Mikko Packalen reply:

In their letter, Emily Gurley et al. concede that contact tracing cannot control the epidemic, but they defend it on the basis that it is still worth doing if it can slow the spread of the pandemic. The authors make many problematic and incorrect claims.

Gurley et al. assert that because we think that a perfect contact tracing program is impossible, we therefore believe it is futile to implement an imperfect but still useful program. Gurley et al. mistake our argument. In settings where the epidemic is widespread, we show that contact tracing efforts will not yield much gain in slowing the epidemic’s progress and may actually make this epidemic worse.

The first part of our argument has been proven by events during the fall and winter wave of COVID-19 cases. All of the contact tracing success stories cited in our original article—South Korea,² Germany,³ and China⁴—have seen sharp increases in disease spread. At the time of writing, Germany is experiencing the largest growth in cases. In all three countries, public health officials have testified to the futility of employing contact tracing to keep the epidemic under control when there is a large and growing number of cases. Though a successful contact tracing program was supposed to keep a country from the need for lockdowns, all three countries have reimposed lockdowns and restrictions on activities this fall and winter.

Modeling work by biologist Christophe Fraser et al. provides a simple explanation.⁵  In a report by Nature, Fraser explained that

in a single day, 70% of cases need to isolate and 70% of contacts need to be traced and quarantined for the outbreak to slow (defined as each infected person passing the virus to fewer than one other, on average).⁶

This level of efficiency and dispatch is well beyond the capacity of real-world contact tracing programs. In the US, the Coronavirus Aid, Relief, and Economic Security (CARES) Act allocated $25 billion to hire tens of thousands of contact tracers nationwide.⁷ In both Israel and Germany, soldiers have been active in contact tracing efforts.⁸ Despite immense expenditure and logistical efforts, contact tracing programs worldwide report being overwhelmed by rising cases during the fall and winter surge.

In upstate New York, delays in processing cases mean that 8 or 9 days can elapse after the initial PCR test before contact tracers are even made aware of a positive case.⁹ When the number of cases is sufficiently high, there is no way for the tracers to complete their assigned tracing activities each day, creating a backlog. This is the very definition of futility. Compare this outcome against the requirements outlined in the theoretical work of Fraser et al.¹⁰ One can readily understand why the head of New York City’s bureau of communicable diseases wrote to Anthony Fauci in January 2021, asking that the federal government permit the city to spend its contact tracing money on vaccine distribution efforts instead.¹¹

Gurley et al. claim that we erred in citing the possibility that someone quarantined as a contact could be set free if they test negative. On this point, they are simply incorrect. The US CDC guideline says that the two-week quarantine period can be cut short with a negative test: “Quarantine can end after Day 7 if a diagnostic specimen tests negative and if no symptoms were reported during daily monitoring.”¹²

Gurley et al. also claim that we misunderstand the impact of a functional false-positive PCR test—an unfortunately common occurrence—on the efficacy of contact tracing.¹³ They are mistaken. A functional false-positive test result will keep a person in quarantine longer than necessary, with no benefit in relation to infection control.¹⁴ Since a functional false-positive test result can occur several months after recovery, the contacts of a person identified on this basis will commonly have no more risk of being infected than anyone else in the population.¹⁵ The possibility of false-positive results undermines the efficiency of contact tracing programs.

In our essay, we hypothesized that the possibility of contact tracing would provide a disincentive for a person with suspected SARS-CoV-2 infection to seek testing in the first place. Gurley et al. assert that there is no evidence for this hypothesis. Instead, they offer an economically naive theory about the personal benefits of testing. We view their theory as naive because it does not consider the personal costs incurred from taking an imperfect test.

Our hypothesis is rooted in the literature of economic epidemiology, which finds much evidence that people will act on private health information in individually beneficial ways, even when external public health consequences would recommend a different course of action.¹⁶ In the case of COVID-19, our hypothesis is consistent with the evidence that many people are simply not interested in cooperating with contact tracers.

In San Diego County, California, contact tracers speak with an average of only 2.2 people for each person who tests positive.¹⁷ This is because a large fraction of people called by contact tracers will not answer the call, or will refuse to provide any information about their contacts to public health officials. As we noted, in Los Angeles County, California, 60% of people contacted by tracers refuse to provide any information about their contacts, despite being offered a payment of $20 to cooperate.¹⁸ The reasons provided are all too human: “Some people are embarrassed, others fear for their jobs, and some don’t trust the government.”¹⁹

Gurley et al. mistakenly assert that we think of contact tracing as a “monolithic intervention” and that we do not understand how it might be adapted to be useful in settings of high disease spread. They believe that a form of backward contact tracing can be “a critical element of the overall response” to slow disease spread. On the contrary, our essay provided descriptions of both backward and forward contact tracing and described situations where such programs might be useful.

In our view, Gurley et al. seem naive about the actual results of contact tracing operations in times of high disease spread, when long delays in case reporting and lack of cooperation by contacts is the norm. They also seem naive about the ability of rigorous contact tracing to prevent high disease spread. Despite implementing a well-publicized contact tracing program, described as the “envy of Europe” in spring 2020,²⁰ Germany has faced an explosion in cases and COVID-19 deaths throughout the fall and winter. Lockdowns in Germany have since been extended until at least March 28, 2021.

In their letter, Gurley et al. also mischaracterize a separate essay written by Jay Bhattacharya, entitled “The Great Barrington Declaration.”²¹ To begin with, this argument is a non sequitur: we could still be right about the futility of contact tracing even if every word in the Great Barrington Declaration were incorrect.  

Gurley et al. misrepresent the declaration; it does not call for nature to take its course. The declaration advocates for focused protection of the people at highest risk of mortality from COVID-19 infection, a group which includes the elderly and some chronically ill patients. Gurley et al. assert instead that the declaration calls for a herd immunity strategy. To misrepresent the Great Barrington Declaration in this way belies a fundamental epidemiological mistake. Herd immunity, achieved by a combination of vaccination and natural infection, is the natural endpoint of the epidemic, no matter what strategy a country follows.²² The scientific dispute concerns the policies and strategies that will minimize mortality and human suffering until that state is achieved.

Gurley et al. argue without justification that Bhattacharya believes there are no interventions where the benefits exceed the costs. None of the signatories to the Great Barrington Declaration have ever advocated such an idea. In fact, the Great Barrington Declaration and associated writings put forward an extensive set of interventions aimed at shielding the vulnerable. We believe they are certainly worth the expense.²³

Gurley et al. assert without evidence that implementing the focused protection approach described in the Great Barrington Declaration would result in millions of COVID-19 deaths. One possible interpretation of this statement is that Gurley et al. believe that focused protection of the vulnerable is impossible. If this is indeed their belief, it is contradicted by the evidence from seroprevalence studies, which indicates that some countries have more effectively shielded the vulnerable than others.²⁴ Focused protection is possible if a country adopts policies that prioritize protecting the vulnerable. A false sense of confidence in contact tracing has undermined the desire of public health officials to seek these solutions. A reliance on contact tracing to do the job is thus an exercise in futility.

Finally, Gurley et al. start their letter by describing our essay as “potentially deadly.” In our view, the COVID-19 pandemic has led to widespread censorship of scientific debate.²⁵ Indeed, we believe there has been censorship on a scale that has not been seen since Trofim Lysenko’s suppression of genetics in the Soviet Union during the 1930s.²⁶ In an unfolding epidemic, the work of science unavoidably encompasses matters of life and death. If we are right, accepting the thesis advanced by Gurley et al. would be deadly. It would also support the continuation of a contact tracing policy that provides an illusion of control over disease spread that we simply do not possess, create disincentives for people to get tested, and distract us from adopting more useful policies like directly protecting the vulnerable. The question scientists should be concerned with is whether there is good evidence in favor or against a given hypothesis, not whether the hypothesis itself is dangerous. Tactics and rhetoric designed to silence debate have no place in scientific discussion.