By some accounts, Amazon has struggled with virus infections at more than half its warehouses, some of which have even closed for several days to be disinfected. As a result, the company is now installing thermal cameras to screen for workers with elevated temperatures. Amazon had previously been using forehead thermometers, and are now upgrading to IR thermal cameras, costing between $5,000 and $20,000 each.
There is some doubt as to which is more accurate – remote scanning IR cameras or direct forehead temperature readers. The benefit of the thermal cameras is they can be continually scanning many people, whereas thermometers obviously require a person to stop at a testing point while having their temperature taken.
Other companies are rolling out similar systems – for example, Carnival Cruise Lines say they’ll test the temperature of every passenger when they board their cruise.
In the early stages of the virus outbreak, some (but not all) countries were doing temperature screening of some (but not all) arriving passengers from international flights, using IR cameras that simply scanned everyone as they walked from the plane to the immigration and customs arrivals hall. The same systems have been used with earlier virus outbreaks such as SARS.
So why do we say this is meaningless and even dangerous, when it has been adopted at a national level in some countries, and by large companies such as Carnival and Amazon? What is it we know and they don’t know?
There are a number of problems associated with using temperature checks to determine if a person has a coronavirus infection or not. These are not secrets, and are neither subjective nor ambiguous. Certainly, some aspects of this have only become known as more has been learned about the virus, but anyone with any degree of interest in the virus and responsibility for the health and wellbeing of their employees or customers should be much more aware of these issues than me.
The Problem
The problem is simple and easy to understand. There is a period of time between when a person becomes infected by the coronavirus and when they start to display symptoms of their infection. An elevated temperature is the most common of those symptoms (appearing in almost 90% of all symptomatic cases), others like a dry cough and tiredness may also be present.
During the incubation period between infection and visible symptoms appearing, an infected person’s temperature is normal. Checking that person’s temperature will not show if they already have an early stage infection or not.
Typically, it takes about 4.5 days for symptoms to start to appear. But for some people it can take longer, which is why the usual guideline for a person who may have been exposed to the virus is to have them self-isolate/quarantined for 14 days.
Indeed, it has now been shown that some people will take considerably longer than 14 days before the virus makes its presence felt. Some cases have seen incubation periods of 21 days, 28 days, even possibly up to 40 days, prior to symptoms appearing. But these are “edge” cases. Most people who develop virus symptoms do so in 4 – 7 days, and just about everyone else in less than 14 days.
The point here is that temperature checks are missing all these people who have not yet started to develop a fever. Does that matter? Yes, for two reasons.
Temperature Checks Rapidly Become Invalidated
The first of these two reasons is that if you test a person’s temperature when boarding a one week cruise, and do not continue to regularly test their temperature several times a day, every day, while on the cruise, there’s the probability that any people who boarded with the infection but not yet with a high temperature will become symptomatic some days later, while on the cruise.
Indeed, if you work through the numbers, if you’re on a cruise in a large Amazon warehouse, both cases with 4,000 or more people around you, there’s a possibility among those 4,000 people that someone goes from no fever to getting a fever within a single day (in a workplace environment, during a single 8 or more hour shift).
Possibly the new IR camera systems Amazon is deploying are continually scanning employees all the time. But those are not on cruise ships or in most other locations due to their high cost.
Many People Never Show Any Symptoms
Now for a much bigger problem. It seems that for every infected person who is showing or will subsequently show symptoms, at least one more person will go through the entire cycle of their infection without displaying any symptoms at all. Some studies suggest there might be as many as five or even as many as 50 people who don’t show symptoms for each person who does. The exact ratio is not yet certain, and may vary regionally and demographically, but what is certain is that way more than half of all people who get the Covid-19 virus never show any symptoms at all.
So when you are screening for a high temperature, you’ll miss all the people who have not yet developed a high temperature, you’ll miss all the people who have a high temperature and are so unwell they’re staying at home or in hospital, and you’ll miss more than half – maybe more than 90% of all infected people entirely – because they’ll never develop a high temperature.
Don’t get us wrong. If temperature screening could reduce the virus transmission rate even by “only” 10%, that would be a small victory and get us closer to bringing the reproduction rate of the virus down to the point where it is no longer growing in numbers, and instead shrinking.
But let’s now move on to the truly big weakness of temperature testing.
People Are Most Infectious While Asymptomatic
The coronavirus is a very “clever” spreader. It is common for respiratory type viruses to spread through coughing and sneezing. The “good news” part of that is that the process makes it obvious when a person coughs or sneezes, allowing both the infected person and people around them to modify their behaviors.
But unlike most other viruses, the coronavirus appears to be most effective at spreading itself and infecting other people before any symptoms appear. Viral particles are simply breathed out in every breath an infected person expells – no cough or sneeze required.
Adding to the severity of this risk is the particles are not droplets. If they were droplets, they would quickly fall to the ground. But they are smaller than droplet size and are “aerosolized”. In their aerosol form they hang, suspended in the air, for five or ten minutes, maybe even longer.
So, a person gets infected one day, and within probably a day, the virus is sufficiently present in their lungs as to be releasing lots more copies of itself every time the person breathes out. But for the next 3.5 days or longer, the person doesn’t register on any thermal scan. By the time there is a detectable high temperature, the virus has done its job and replicated itself into another two or three people, ensuring the virus not only perpetuates itself but also grows its spread to more and more people. Temperature testing has been useless.
What about the majority of people who never show any symptoms at all, and have such a mild dose of the infection as not to even realize they have it? Can they also pass the virus on to other people?
The answer seems to be generally stated as “Yes, they can”. (There are many other articles all saying the same thing.)
What Percent of Virus Spreaders Are Detected by a Temperature Check?
So, let’s think about how effective these temperatures checks might be. We know that for the first three or more days between becoming infectious and displaying symptoms, a person is at their most infectious and does most of the passing of the virus to other people.
We also expect that once a person gets a full fever and probably other related problems, they’ll realize they’re unwell and stay at home or even go to hospital because they’re too sick to work. If they did go to work, their malaise would likely become apparent to coworkers who could then respond with increased social distancing and/or urge their colleague to go home.
Let’s also consider that many more people get the disease, become infectious, and never display symptoms at all.
So really all the temperature checks are doing is advising us of perhaps a day or two between when some people start to feel unwell and when they reach the point where they know something is definitely wrong and need to stay home or seek medical help to recover.
If we say three quarters of all transmission is done before symptoms appear (assuming people ever display symptoms at all), if we say that half the people who do get a fever then take themselves out of public exposure on their own accord, and if we say that two out of every three infected people never display symptoms at all, that means temperature checks are at very best no more than about 4% effective. For every infected person detected by temperature checks, another 25 people have walked past the thermal scanners, in an infectious condition, but undetected.
So the temperature checks are almost useless. But, as you can see, there’s more to this article. Let’s now see how temperature checks actually become worse than useless.
The Social Dangers of a False Negative
(A “false negative” is when the test says “You’re okay, you’re not infected” even though you actually are.)
One of the weak arguments against encouraging mask use is that it encourages over-confidence on the part of the mask wearers.
This is similar to what was observed when cars began to be fitted with anti-lock brakes. People started driving more aggressively, meaning the accident reduction factor from anti-lock brakes was much lower than initial studies had suggested. Indeed, one study by the Insurance Institute for Highway Safety shows that rather than having fewer fatal crashes, drivers with anti-lock brakes are actually 65% more likely to be in a fatal crash.
A similar concept applies to “official temperature testing”. It is a test that of course is asserted as being useful and meaningful – after all, why would companies be spending tens of thousands of dollars on such testing if it was useless?
So, when people have been validated by the temperature check as being “safe”, they start to relax their social distancing for two reasons. First, because they no longer worry about the people around them who have also been validated as being “safe”, and secondly, even after leaving the building/ship/whatever, they feel that they themselves are not a potential threat to other people. This makes them more likely to acquire the virus from someone else in the “safe” areas and more likely, if already infected, to pass it on to someone else, anywhere and everywhere.
If the test was actually correctly detecting 99% of at-risk people, this compensatory behavior might be acceptable. But remember the test might only be detecting 5% or fewer of the at-risk people, meaning the relaxing people will inevitably do is inappropriate and on balance adds to exposure/transmission risk.
If an unambiguous safety-enhancing system like anti-lock braking on cars has resulted in drivers overcompensating so as to increase fatalities by 65%, how much more dangerous is temperature screening, something that is flawed right from the get-go?
The Problems of a False Positive
There’s another problem, too, the “false positive” problem, which causes us to examine the assumption of “better safe than sorry” and its modern day more obsessive version, “out of an abundance of caution”.
The basic false positive means when the temperature check shows you to have an elevated temperature, but the reason for the high temperature is not due to the coronavirus.
High temperatures and fevers can be caused by many different types of infections, both from viruses and bacteria. Now maybe you might say “a person with a fever should be home in bed, no matter what their illness, because they’re obviously unwell”. There’s even truth in that, sometimes.
But who are you to dictate to someone else what they can and can’t do? Are you saying a person who is recovering from strep throat or a “tummy bug” or an ear infection shouldn’t be allowed out in public?
More than that, if you’re screening people somewhere and discover a person with a high temperature, you are probably not only turning them away right then, but also requiring them to either stay away for two weeks or to get tested and cleared before they can return.
So you have just changed a person’s life for two weeks. And as for the “get tested and cleared” requirement, while it seems logical, the reality continues, even today (Saturday 18 April) that there is a desperate shortage of tests available in the US (and many other countries). In some areas, getting tested is almost impossible to arrange, and even if you do ultimately get tested, it could take a week for a result to be returned.
Meanwhile the person has been refused permission to travel, refused permission to work, refused entry to various stores and leisure centers, all “out of an abundance of caution”, but with no actual virus risk present.
That False Sense of Security, Again
We cringe when we read of organisations who now boast about temperature testing. Because, all too often, that is all they do, and they use the temperature testing as an excuse for not adopting or continuing more difficult and effective measures.
Are cruise lines now telling us that temperature testing when we board the cruise ship guarantees our medical safety on board the ship? Is Amazon telling its workers that its expensive infra-red camera warehouse surveillance makes it safe to work there?
Sure, by all means take our temperatures, but make sure you understand, and you in turn make sure we understand, that your temperature testing probably misses 95% of all dangerously infected people.
Summary
Testing a person’s temperature is simple and inexpensive. But it is unlikely to find much more than 4% of people who might be infectious with the coronavirus.
Sure, even a 4% reduction in passed on infections is nothing to be sneezed at (bad joke!) but the offsetting problems of this testing – the “false positives” and the false sense of safety from people who are dangerously infected but pass the temperature test – probably outweigh the marginal benefit to be gained.
The key thing is to educate everyone that temperature testing is almost meaningless, so as to ensure people who “pass” a temperature test don’t start relaxing other more valuable approaches to minimizing both their risk of becoming infected and also their risk of infecting other people.
Completely agree with your points about temperature measuring.
I am excited by the news that special disease sniffer dogs are being
considered as a possible solution (I believe in the UK). With their advanced
sense of smell, they have been able to detect things like various cancers,
malaria, etc. They can also screen up to 750 people an hour.
If this is possible, wouldn’t it be great? If someone is identified, then they
can take the swab or blood tests. You do fewer tests, focus on those that
need to be tested, and can do it at low cost and for large numbers of people.
I hope this is possible, and that the US jumps on it, if it is!
Hi, Peter
The sniffer dogs would be nice. There’s something very “human” about simply being sniffed by a friendly dog, isn’t there! But there’s not yet any promise that they will be capable of smelling the virus, and one also wonders about their own risk of contracting it if they are smelling for it. There are some unclear studies about virus transmission from people to dogs and from dogs to people.
Also the 750 people per hour number might be a bit optimistic. There’s a reason we don’t see more dogs in airports. The training is complex and lengthy, and the dogs have short attention spans and need to be given plenty of breaks.
But anything extra we can do has to be good, and I’d be as delighted as you if we start to develop a canine resource. I’d much prefer to be sniffed by a dog than to have a stick poked way up to the top of my nose!