Skip to Main Content

Consensus seems to be emerging that businesses can safely reopen by checking the temperatures of people entering them and by putting in place environmental controls — from social distancing and wearing masks to improved air circulation — to reduce the spread of SARS-CoV-2 in the workplace.

This consensus is rational. It is convenient. And it is driven by the well-founded desire to bring life back to normal by reopening businesses, schools, places of worship, and more. The consensus is also reinforced by guidance from the Centers for Disease Control and Prevention and the White House on the steps needed to reopen the U.S.

I worry that this consensus is built on optimism, and reopening in this way will provide people with a false sense of security and a false impression they are reducing their risk of infection that is not supported by evidence.


There are two main types of risk-reduction efforts in the workplace: those designed to keep infected individuals out, and those designed to limit the spread of SARS-CoV-2 within the workplace, given the likelihood that some infected individuals will get in.

Since we can accurately and rapidly measure the sensitivity of screening measures for SARS-CoV-2, it is easier to measure the potential efficacy of interventions designed to keep infected individuals out of the workplace. The published scientific literature and data on testing accuracy can inform workplace interventions, since the primary question is: What works in identifying who is infected? It is much more difficult to establish the evidence for preventing the spread of SARS-CoV-2 within the workplace, since environmental studies on masks, air recirculation, and sanitation will need to isolate the key factors and account for the variations across environments, and these studies have not been done for this virus.


Keeping infected individuals out of the workplace

Without a way to identify individuals infected with SARS-CoV-2 who have no symptoms or mild symptoms before entering the workplace, American business may be forced to close soon after reopening.

Measuring temperature to allow employees into the workplace has become a common strategy. This approach is being used to screen employees by Amazon, major car producers, technology employers, airlines, and others. Some employers eschew thermometers and opt for the more efficient and precise-sounding “thermal scanner,” basically a camera that looks for heat.

This approach has limitations, as Ford recently learned. It had to shut down production at two plants after reopening when employees who had been checked for fever were later found to be infected with the coronavirus.

Chances of missing an employee with a Covid-19 infection % missed
Temperature screening/”thermal scanning” with best-in-class scanner >86%
Symptom checks with best-in-class symptom screener >46%
Universal antibody testing with best-in-class FDA-approved test ~36%
Universal PCR testing with typical FDA-approved test ~2%


Temperature screening provides false assurance to employees entering the workplace. One recent study showed that about 70% of patients sick enough to be hospitalized for Covid-19 did not have fevers. Coupled with the fact that most people infected with the coronavirus do not have symptoms, screening for temperatures will miss at least 86% of infected individuals, and likely miss an even higher percentage.

A “barrier” that allows nearly 9 in 10 infected individuals to enter a workplace or business is not one that should be used to provide reassurance to employees.

A slightly better method for keeping infected individuals out of the workplace is daily symptom screening coupled with clear instructions to stay home if symptoms appear. The primary weakness of this method is that only a minority of individuals infected with SARS-CoV-2 develop symptoms. A nursing home study in Washington state showed that 56% of infected individuals were asymptomatic, while broader studies in Wuhan showed 69% were asymptomatic. Studies among pregnant women in New York and tourists and crew on an Antarctic cruise ship showed that more 80% of infected individuals were asymptomatic.

If we use the optimistic scenario and rely on expanded symptom screening done daily, this approach identifies only about half of infected individuals, though likely only several days after they became contagious, allowing for workplace spread in the interim. More realistically, based on recent data outside of nursing homes, symptom-tracking will miss upwards of 80% of infections.

This approach is compounded by the risk of underreporting symptoms, especially when employees are not paid when they are cannot come to work.

The only approaches that can reliably reduce the number of people with active SARS-CoV-2 infections coming into the workplace involve testing. There has been a great deal of debate on various tests. While antibody tests may play a role in some settings, the greatest limitation to their use in risk reduction is that tests do not reliably turn positive in infected people until about a week after being infected (and thus missing a key window for transmission risk). Tests that look for the virus, typically via PCR (but also possibly other emerging modalities such as loop-mediated isothermal amplification [LAMP)], CRISPR, next-generation sequencing, and antigen testing), are the most useful in reducing the spread of the virus in the workplace.

Tests that look for the virus identify infections early on and with high sensitivity. With testing, employers can reliably identify almost all infected individuals and keep them from entering the workplace. The emergence of self-administered testing from several companies via a swab placed in the front of the nose or saliva are making this kind of testing more convenient, while also eliminating the need for protective equipment and exposure of health care staff.

By using testing protocols based on local epidemiology of Covid-19, personal and workplace risk factors, and tracking symptoms and contacts, employers can calibrate how often they need to test workers to greatly reduce the risk of workplace transmission. These protocols need to be updated frequently to account for changes in epidemiology and in science, like the protocol a colleague and I have published with input from several leading experts.

Employer-based testing won’t be easy. The annual per employee cost of making testing part of risk reduction — while less than one day’s wages for the average worker — may be a deterrent to some organizations. The data on test reliability and testing types are evolving quickly. Without employer-led strategies to make testing convenient for asymptomatic workers, many will not pursue testing. And even the most sophisticated employers would not wish to invent their own testing or clearance protocols, but will need to depend on evidence-based reference protocols.

Gathering test results from diverse labs where workers will be tested in their communities, and from employer-facilitated lab testing partners, is a technical challenge, matched only by the technical challenge of communicating the relevant information to the employer in a manner that preserves worker privacy. How will an employer make sense of the testing information? Employers will need to know who is and who is not at risk of being infected, and that is not simply just who has or has not been recently tested negative.

Despite these challenges, employers that implement testing based on protocols will be in good positions to reduce risk by keeping enough infected people out of the workplace to avoid a workplace epidemic.

Limiting spread from infected individuals in the workplace

Even the most complete testing will not create an impermeable barrier to SARS-CoV-2, so employers must assume that infected individuals will enter the workplace and implement measures to limit the spread of the virus from infected to uninfected individuals. The challenge with assessing these measures is that they have not yet been reliably studied in the workplace.

Sensible best practices complement efforts to keep infected individuals out of the workplace but cannot replace them. These practices include:

  • Wearing masks indoors
  • Reducing the number of people in the workplace and spacing out those who are there
  • Encouraging frequent hand sanitizing
  • Limiting meeting sizes and the number of people in a room
  • Taking steps to avoid or reduce physical contact, including at meals and entrances
  • Frequent cleaning of spaces
  • Increasing air circulation

Some employers have also sought to keep individuals who may be at higher risk for serious illness from Covid, out of the workplace, asking older workers, those who are obese, those with diabetes, and others to remain at home. These practices, however, may be discriminatory, and should be instituted with caution.

An evidence-based approach to return to work

Workplace risk-reduction strategies need to do exactly what their name implies: reduce risk in a demonstrable manner and to a level that would be acceptable to employers and employees. Disease modeling suggests that a risk-reduction protocol that includes both evidence-based testing and environmental controls can meaningfully bring down the risk of workplace transmission of Covid-19.

Employers should be sharing with employees the expected impact of risk reduction strategies. That’s a fair and necessary exchange of transparency for asking employees to be subject to additional screening designed to protect them. Efforts to keep infected individuals out of the workplace will backfire on employers who provide misguided or overstated reassurance.

Workplaces and public spaces across the nation were shut down to slow the spread of disease so it wouldn’t overwhelm our health care systems. It was also done to buy us time to develop at least one of the three options that would enable the country to reopen safely: widespread testing so we could contain the virus, treatments that could sharply reduce the threat of the virus, and a broadly distributed vaccine that could limit infection with SARS-CoV-2. While there is some promise in vaccines and treatments, they will take longer — and possibly be less effective — than we want to believe.

We have developed the testing capacity, workplace protocols, and the data framework to enable organizations across the U.S. to sharply reduce workplace Covid-19 risk. If we do not take advantage of the emerging ability to use testing and workplace protocols to safely re-open our businesses and schools, then we will have squandered our massive investment to buy time over the last several weeks.

Rajaie Batniji is a physician and co-founder and chief health officer of Collective Health.

  • What about schools in the Fall? Can they get regular PCR testing onsite? Seems like a safe, standardized way to do population surveillance. Through the kids we can access a broad range of socioeconomic and ethnic/racial groups, promoting contact tracing and self-isolation.
    They are doing it in parts of Germany. Will need federal funds but it seems like the economy simply cannot work with kids staying at home for 2020-1.

  • This material from Dr. Batniji ought to be shared nation-wide, with as many employers as possible. PCR testing is clearly the only effective “catch”, and the cost will be reflected in price for the consumer. Covid HAS changed how we live and operate, and will continue to do so until there are good drugs and vaccines. Sloppy checking = missing cases = spread = shutting down again = the next wave = exponentially higher as it is built on a much broader base. The hospital and health care system WILL overflow – with quite possibly as biggest shortage : health care providers.
    The onus for safe return to business & work & income also depends greatly on employEE commitment (at and outside work) : because the virus spreads air-borne: wear a mask, and wear it properly (tight seal) – and keep social distance. For return to business and work to be successful, this Covid-prudent behavior must be much more consistently adhered to.

    • With respect “Sloppy checking = missing cases = spread = shutting down again = the next wave = exponentially higher as it is built on a much broader base. The hospital and health care system WILL overflow” = Speculation absent data or back up which sadly 1) Was used to justify the flattening of the economy and the futures of many low income folks and 2) Undermines credibility… See “Boy Who Cried Wolf.”

  • What’s left out here is the potential for pooled testing, appropriate for background surveillance of nominally healthy people. If each employee can spit into a clean straw at the start of each shift, and all the samples are combined, you get a very cheap test.

    Even if the real value false negative rate is 10%, that’s OK, because you make it up in volume.

    Efforts like have a role also. With better data we can get a better sense of who’s lower risk of poor outcomes, and plan appropriately.

    Wearing a mask for an entire shift is hard. That’s just not a viable solution long term.

    We’re also missing case data. Only a select public health agencies release case data indicating how an individual was infected. Without that data, we’re not being optimistic, we’re flying blind. We don’t really know what activities are risky and not.

  • I have to disagree with Dr. Batniji about one thing – it is not optimism driving the return-to-work strategy, it is desperation.
    Having bungled the shutdown, making it far more destructive than it need have been – and I understand they did not know everything about the disease, but still they bungled what they did know – now the economic and emotional damage has become so great, they will have a bad reopening too, and it is not avoidable any longer.
    The evidence seems to suggest cases will continue to go up, that is, more newly diagnosed people every day in all but the states originally hardest hit -in most of the country – and around 100,000 more people dying by the end of summer, then, far more, until the vaccines arrive – but terrible as it is, we have to reopen.
    We had, in theory, three choices- 1. with intelligent leadership and public cooperation, do carefully targeted things to reduce the epidemic while minimizing economic and emotional harm. 2. with bad leadership and an indifferent public, shut down the country, with no clear path to reopening safely, but saving the country from a huge death/maiming toll in the short term, hoping good treatments are found. 3. With bad leadership and an indifferent public, do not fully shut down the country, and reopen it haphazardly in an often needlessly harmful way, but get it going so that we are not economically ruined.

    We chose # 3, beginning some time in January, maybe earlier. But now there are no other choices.

    • #3 is the choice by design or default.

      Imagine had a different strategy been used: mask the supermarket workers, open markets for 48 hours, require everyone stock up for exactly 15 days, then isolate just about everyone other than police. A harder shutdown could have been a shorter shutdown, giving more breathing room, knocking down more cases without the economic disruption.

      The Swedish model, while the jury is still out, looks better and better the longer we go on.

      We should also lay blame on the choice to use unemployment insurance to ride out the gap. Firing everyone so they can collect unemployment is just about the stupidest way to keep an economy going.

      From here out the only way forward is to try and figure out what’s really risky, and what’s not very risky.

  • Considering that pessimism not evidence was used to drive the lockdown strategy, perhaps we could do with a little optimism! Seriously, the lockdown was built on bad data (or no data) and evidence has continued to mount that the virus is 10-40x LESS deadly than we thought it was and it has remained extremely selective in who it generally kills.

    At the end of the day, the lockdown was to serve one purpose: to slow the spread enough so hospitals don’t become inundated…it was NOT to stop the spread wholesale in order to wait for some vaccine which may never come (the virus spreads too easily, it doesn’t kill enough people, and it is already endemic). The good news is, for the most part (outliers aside), hospitals never became inundated and likely most of them never would have.

    There is a case to be made that history may judge the lockdown as a massive overreaction which we’ll be paying for for years if not decades. The jury is still out and anyone who says otherwise is fooling themselves: the efficacy of this strategy will be debated for years if not decades. (..and I’m saying this as a professional data scientist with 20 years in the pharmaceutical industry…not a right-wing conspiracy theorist.)

  • The 2% of cases missed with PCR is likely not correct. The sensitivities listed in the link are likely analytical sensitivities which relate to detection of the virus when it is known to be in the specimen tested.

    This is not the same as clinical sensitivity. There are very few studies examining the clinical sensitivity which depends on many other factors (sampling technique, presence of virus in the spot tested). This is the parameter of interest for the question of percentage of cases missed with PCR testing. the few studies examining this question indicate sensitivity of 60-70%, not 98% as this table indicates.

    The percentage of cases missed with PCR testing is likely 30-40%, rather than 2%.

    • You raise an important point, that we have also grappled with quite a bit. For each of the interventions considered (temperature checks, symptom screening), we are considering what you may call analytic sensitivity. There simply are not real world studies on their implementation, so what you see is the maximum sensitivity assuming perfect implementation. Each of these, including PCR testing, may perform worse in the real-world. That said, the data on PCR testing do indicate a higher sensitivity, closer to the analytic sensitivity reported here, when looking at the improved and recently authorized sample collection methods, like anterior nasal swabs, which reduce user error when compared to NP swabs.

  • With respect there seems to be a straw man argument by stating that “businesses can safely reopen by checking the temperatures of people entering them and by putting in place environmental controls.” Actually the argument is more like 1) We closed the economy to avoid overrun medical facilities which didn’t happen and 2) The closed economy is crushing the finances and lives of tens of millions with low income hit the worst and 3) Adults under 60 w/o underlying illness are understood to have low actual risk of significant illness. Daily worksite testing and beyond (prior to boarding a plane, etc.) would be ideal but for now the focus should be on protecting the vulnerable and “above all due no harm.”

  • Employers at high risk work sites, i.e. meat packers, nursing homes, close work station manufacturing, etc., will need to test and retest their workers. It’s the cost of doing business until a vaccine or therapeutics are available. Smaller work site( offices of 10-15) can manage with testing already available offsite. With resuls transmitted to employers.

Comments are closed.