How accurate are forehead thermometers, really?

I've seen several articles (like this one from the New York Times in February) suggesting that the now-ubiquitous forehead thermometers (aka infrared thermometers, temporal scanners, or just "temperature guns") are quick and easy but not actually very accurate. Now, that article also talked about how useless masks are (this was February, remember?), but my physiotherapist said the same thing just this week, so it's clearly a pretty widely-held view. Is it actually true, though?

Well, yes and no, but mainly no. There are several different types of thermometer, and they can give slightly different readings. Rectal temperature readings are generally considered the most accurate, although the drawbacks are obvious, and this method is rarely used in practice. The more usual oral thermometer tends to give a reading about 0.3-0.6°C (0.5-1.0°F) lower than a rectal thermometer, and the oral reading is the most often used (yielding the familiar 37.0°C or 98.6°F "normal" temperature). Tympanic or ear thermometers give very similar readings to rectal thermometers. Armpit thermometers tend to give temperatures 0.3-0.6°C (0.5-1.0°F) lower than oral thermometers. And so, as it turns out, do infrared forehead thermometers. 

So, while they are not 100% accurate, they are still pretty good. They are unlikely to give a reading that is wildly inaccurate in either direction. And, if they are predictably half a degree lower than oral thermometers, it is easy to make an adjustment for the discrepancy. 

Of course, whether a fever is a reliable marker of COVID-19 is another question entirely: the presence of a fever is a good (but not definitive) indicator of the virus, but the absence of a fever does not equate to the absence of the virus.

Forehead thermometers remain the easiest, quickest and least invasive test for COVID-19, and so should not be dismissed out of hand when you encounter them at a barbers or a nail salon or a school or a restaurant. But they are not reliable or definitive enough to serve as the arbiter of health at an international airport in my opinion. They are a useful initial screening tool, but no substitute for a proper COVID-19. But how accurate are THEY?