Exergen Corp.'s latest radio ad claims no-contact thermometers are totally inaccurate and have no science supporting their use. Is it true?
I don’t think the ad is available online but you can find similar sentiments, for example, here:
“The fact that so many people already know that non-contact thermometers are inaccurate is reassuring. Yet the fact that so many others still mistakenly believe they are accurate is deeply troubling, as they have been proven time and again to be completely inaccurate. Companies and organizations that use them for public screenings are being negligent at a time when nothing matters more than taking an accurate temperature reading,” said Francesco Pompei, Ph.D., CEO of Exergen Corporation. “Screening with a non-contact thermometer is pure theater, and will cause many to refuse to enter the establishment.”
(Exergen is a thermometer company, specifically they sell temporal artery [forehead] thermometers. And for people who normally resort to Wikipedia, caveat lector: I’m pretty sure Exergen has been editing that site to promote itself.)
Based just on what I see on Exergen’s website, I would be leery of their claims. For instance, they state that other methods of determining body temp including oral, axillary and in-ear thermometers are unsatisfactory because they involve “significant discomfort”. Oh, really?
The CEO’s quoted statement that no-contact devices are “completely inaccurate” is itself bogus, according to people who don’t have a vested interest in promoting temporal artery thermometers. For instance:
Without doing a deep dive into the literature I don’t know if accuracy has been evaluated comparing a health professional taking a reading to the hostess at your local Applebee’s, but I suspect that the latter provides a somewhat less accurate temp on average.
Unless I am missing something (which is quite possible), the Cleveland Clinic article seems to draw a conclusion unsupported by the research it actually cites. See post #2.
Having now reviewed all of the FDA’s citations on that page, except the ISO and ASTM standards (which are not free and seem to be more about manufacturing standards than evidence of accuracy), I found absolutely no evidence of the accuracy of no-contact thermometers and no scientific studies supporting their use.
The evidence cited in that review is mixed, but it certainly doesn’t support a sweeping claim that no-contact thermometers are “totally inaccurate”.
By the way, this is the review I meant to link to earlier.
“Thermal screening has reasonable diagnostic accuracy in the detection of fever, although it may vary with changes in subject characteristics, setting, index test and the reference standard used. Thermal screening has a good NPV (negative predictive value) even during a pandemic.”
PubMed offers links to other relevant published research.
One consideration is that accuracy in the lab may not translate to accuracy in the field due to user error. The thermometer I have came with instructions saying to hold it a certain distance from the temple. That’s likely how it was calibrated in the lab. But common use typically seen in public is the thermometer held a variable distance to the center of the forehead. So even if the thermometer itself is accurate, incorrect use may lead to incorrect readings.
When all of the COVID stuff first hit, the company I work for bought a bunch of pulse oximeters and no-contact thermometers, and anyone who wanted one was free to take them home, so I ended up with one of each. I was curious how accurate they were. I don’t have a way to independently measure my blood o2 levels so I couldn’t verify the oximeter (other than the pulse rate, which was accurate) but the no-contact thermometer is pretty accurate.
I realize that this is just a sample size of one, but anyone who claims that they are totally inaccurate is full of shit.
It does have its limitations. It’s measuring your skin temperature so if you have done something to temporarily raise or lower your skin temperature it will be a little off. Similarly, if you drink something hot or cold you can throw off an oral thermometer just as easily.
Overall, it’s quick and easy to use, and doesn’t require cleaning or the use of a disposable sleeve. I don’t expect it to be accurate enough to be traceable back to NIST standards, but for what it is, it seems to do the job well.
I have also used older versions designed for electronic testing. Again, not accurate enough that you’d want to calibrate it to NIST standards, but good enough for diagnostic work in electronic stuff. Works on the same principle as human thermometers, and the technology has been around for quite some time.
“Significant discomfort” is a reach, but I think it’s fair to say that those other methods are more time-consuming and/or more intrusive. Less of an issue in a doctor’s visit, more of an issue if you’re screening people for entry into a restricted venue.
What do the authors mean when they speak of a thermometer’s specificity? Is it a percentage value of the reference temperature (from eg: an oral thermometer)?
I just tried mine. The instructions say 1 to 5 cm distance. It remained accurate within 0.2 deg F from a distance of 0 (physically touching my forehead) out to about 5 inches (12 cm or so). At about 5 1/2 inches it started reading low and kept getting more inaccurate with increased distance.
Apparently varying the distance doesn’t matter much as long as you keep it reasonable.
The distance can vary significantly without affecting accuracy. Like a camera, the thermometer’s sensor has a certain field-of-view angle. So at distance X, it’s seeing a spot with diameter Y. At distance 2X, it’s seeing a spot with diameter 2Y. As long as the larger circle is the same temperature as the smaller circle, the thermometer will report the same temperature in either case. It only gets to be a problem when you’re so far away that the sample spot size starts seeing things other than your forehead that might be at a different temperature (your hair, the wall behind you, etc.).
After checking wiki & a couple dictionary type sites, they seem to be using a definition like “sensitivity: % chance a person with fever is detected as having fever, specificity:% chance a person with no fever detects as having no fever”. So, not an absolute kelvin-calibrated temp, just correctly showing whether or not you have a fever. Which makes sense in a way - “is your body at higher than healthy temperature “ is more relevant than “is your body at 98.6000 °F”
From what I know about sensors, IR, etc, the sensor “sees” some solid angle in front of the tip. As long as all it sees is skin, it tells you the average temp of all the skin. Really close to a match head or hot square mm of skin, you’ll get an unusual high reading. Far away and you are averaging the temperatur of the skin, and all the air around skin that it sees, so a low reading.
It is a comparatively old & known technology, so assuming average quality diode & one of a variety of easy calibrations, I’d expect an accuracy of maybe hundredths a degree F, with displayed precision of tenths of a degree. For a quick check, boil some water, point thermo at water. Should be 212 minus a bit for ambient pressure and mineral content of the water?
One complication with IR thermometers is that they aren’t measuring how hot something is like a contact thermometer. Rather, they are measuring the radiant heat coming off the object. It’s kind of like how “bright” the heat is. The difficulty is that not all objects radiate heat the same amount at the same temperature. For example, if steel and aluminum are at the same temperature, the steel will radiate more heat than the aluminum. The IR sensor will see the steel as brighter than the aluminum even though they are both at the same temperature. Lab IR thermometers will have a way to indicate what material you are testing so it can correctly translate the brightness to a temperature. Forehead temperatures are fixed and calibrated for skin radiant values. That is, the temperature it shows is for the brightness values of skin at those temperatures. If you point them at other materials that radiate heat differently than skin, the temperature will be off. So pointing a forehead IR thermometer at water likely wouldn’t be accurate. In addition, forehead thermometers typically only show temps in a small range around skin temperature.
And then looking at Figure 4 (p.36) and its interpretation (p.14) (PPV=positive predictive value, NPV=negative predictive value)
At an arbitrary prevalence of 1%, the PPV for detection of fever was 9.2% for NCITs and 9.7% for thermal scanners. This means that out of every 10 patients detected febrile by thermal screening, ~one actually turned out to be febrile. Interestingly, in contrast to PPV, there was only a comparatively smaller fall in the values of NPV- 2.3% (from ~100% to 97.7%) for NCITs and 2.1% (from ~100% to 97.9%)- even as the prevalence of fever increased 105 fold (Figure 4). This would mean that, even at a fever prevalence of 10% during a pandemic, a patient who is detected to be afebrile by thermal screening has over a 97% probability of being truly afebrile by the reference method.
A quick Google on, say, the number of COVID cases in Florida per million residents is 162,724; 162724/1000000=0.162724; a quick search for the % of COVID cases that have fever gives me this site’s claim that “only 4.6% of people infected with COVID-19 are likely to have a fever while contagious”; 0.162724*0.046<0.01, so I think it’s fair to rely on the 1% prevalence (9.2% PPV) rather than the 10% (eyeballing figure 4, close to 100% PPV). The apparent conclusion, (please explain why I’m wrong,) is that if I use a no-contact infrared thermometer I should expect a 90% false-positive rate. But notably a negative reading is very reliable.
