Thermometers Are Hot Garbage

Germs are in the air again: Indicators indicate that the winter wave of flu and COVID is finally under way. Are you on the verge of getting sick? Am I? My 5-year-old does feel a little warm to me; his sister seems okay. Maybe I should take their temperature?

Maybe I should not. Here’s my resolution for the year ahead: I will not take their temperature. No parent should be taking temperatures. Because doing so is next to useless. Home thermometers are trash.

The thermometer I have is the kind you point at someone’s head. Clearly it’s a scam. At times, I’ll pull the trigger and the number that I get seems almost right. At other times, the readout is absurd. I know when it’s the latter case because, as a human being, I possess a sensate hand. Evolution has deployed a field of thermo-sensing cells on the glabrous surface of my skin, and I’ve found that when these are laid against the forehead of my child, they may produce the following diagnosis: He is hot. Or else: He seems normal. No further probing is required.

I bought my noncontact fever gun in 2020, during what was, in retrospect, a fever-screening fervor, when thermal bouncers were deployed at concert halls and other venues to test your forehead from however far away. I think we all knew in our hearts that this was silly, even those of us who thought the fever guns might be better used in other circumstances. But to call the practice “silly” may have been too kind.

The published evidence on fever guns is damning. One study from the FDA compared their readings, as produced under ideal conditions, with those from oral thermometers; it found that they were often grossly out of whack. The very best-performing models, according to this research, were able to detect a threshold fever—100.4 degrees Fahrenheit—about two-thirds of the time; the very worst could never make the proper diagnosis. Another study, led by Adrian Haimovich, who is now an assistant professor of emergency medicine at Beth Israel Deaconess Medical Center, identified visits to emergency departments in which patients had received both forehead temperature checks and readings with oral or rectal thermometers. The forehead guns were successful at identifying fevers in fewer than one-third of cases.

But let’s not single out the gun, which was to some extent a product of its COVID moment. The standard infrared tympanic probe—which takes a temperature quickly in the ear—is also, in important ways, a waste of time. “I was an ER doc practicing full time when these tympanic thermometers came out,” Edmond Hooker, a professor of health-services administration at Xavier University, told me. He quickly came to think they didn’t work: “I would have a kid come back who was so hot, I could fry an egg on their forehead, and the tympanic thermometer had said 98.6 or 99.” So he started running tests. A paper from 1993 found that the ear thermometers were missing children’s fevers. Another of his studies, conducted in adults, found that the devices were dangerously miscalibrated. (More recent research has raised similar concerns.)

Oral thermometers are fairly accurate, but they present some challenges for use with small children. Rectal probes are the most precise. As for armpit readings, those are also pretty unreliable, Hooker told me. I began to ask him about another means of checking temperature, the light-up fever strip that my parents used to lay across my forehead, but he wouldn’t even let me finish the question. “Absolutely worthless! Your mother was better,” he declared. “That’s what my other study showed: Mom was pretty damn good.”

His other study: Having demonstrated that ear thermometers were ineffective, Hooker decided to compare them with human touch. A parent’s hand—nature’s thermometer—did pretty well: It correctly flagged some 82 percent of children’s fevers, versus the tympanic probe’s 75 percent. Parents’ hands were more prone to overdiagnosis, though: Among the kids with normal temperatures, nearly one-quarter felt warm to their parents. (The false-positive rate for ear thermometers was much lower.)

Many such experiments have been conducted now, in health-care settings all around the world: so much effort spent to measure our ability to diagnose a fever with nothing more than touch. (In medical lingo, this practice is properly—and ickily—described as “parental palpation.”) As a rule, these studies aren’t large, and they may be subject to some bias. For instance, all of the ones that I reviewed were carried out in health-care settings—Hooker’s took place in an emergency department—so the participants weren’t quite “your average kids who might or might not have a cold.” Rather, it’s likely that those kids would have had a higher baseline rate of being feverish, and their parents might have been unusually prone to thinking that their children were very sick.

Some researchers have tried to look at all the little studies of parental touch in aggregate, and although this can be an iffy practice—pooling weak research won’t make it any stronger—these studies do yield about the same result as Hooker’s when taken on the whole: Parents’ hands have a solid sensitivity to fever, of nearly 90 percent, but their specificity is low, at about 55 percent. Put another way: When a kid does have a fever, his parents can usually detect it with their hands, but when he doesn’t, they might mistakenly believe he does.

The latter isn’t great, given that a kid with a fever is supposed to stay home from school or day care. In that case, a thermometer could provide a helpful (moderating) second opinion. But taken as a measure of the risk to your child’s health, palpation must be good enough. The very hottest children—the ones whose infection may be most imperiling—are also the least likely to be misdiagnosed by touch: If your kid’s head feels like the side of a convection oven, then you’d almost certainly say he’s sick, and you’d almost certainly be correct. (And you’d be correct to call his doctor.) As for the borderline conditions—a temperature of, say, 101 or 99 or 100.4—your hand won’t name his fever with as much precision as a good thermometer would. But the added benefit that thermometer provides, both to your child’s health and to your peace of mind, is next to nothing.

It’s important to remember that the very definition of a threshold fever is arbitrary and subject to the ancient scientific law of Hey, that sounds like a nice, round number. Converted into Celsius, 100.4 degrees Fahrenheit comes out to an even 38 degrees. The established “normal” temperature of 98.6 degree Fahrenheit maps on to 37 degrees Celsius. (In truth, the temperatures of healthy, older adults will range from 98.9 to 99.9 degrees throughout the day, as measured with an oral thermometer.) Under normal conditions, a measured fever is nothing more than a single aspect of a broader picture that informs the course of treatment, both Haimovich and Hooker told me. An elderly patient with symptoms of a urinary-tract infection might receive a more comprehensive course of antibiotics if she also has a fever, Haimovich said; the heightened temperature suggests that an infection may have spread. But a kid who has a mild fever and is otherwise okay won’t need any treatment. Some evidence suggests that a light fever may even fortify an immune response; so in principle, slightly elevated temperatures should be left alone unless your child is uncomfortable, in which case, maybe ibuprofen? (Conversely, a kid whose temperature is “only” 99 but who seems listless and confused should probably be seen.)

Haimovich said he has small children, so I asked him how he checks their temperature—does he ever feel their head? “Oh, yeah,” he said. He told me that his wife seems better at detecting fever than he is, which fits with known neurophysiology: Some research suggests that women’s hands are more sensitive to warmth than men’s, on average. One study, though, done at a hospital in Canada, found that dads are just as good as moms at detecting fever with their hands. (The moms were much more likely to believe that they possessed this skill.) Other research has examined whether having multiple children—and thus perhaps having more experience feeling heads—might also be a factor. The answer is no. This suggests that sussing out a child’s fever is not so much a practiced art as a basic fact of our perception.

As for Hooker, he said he doesn’t even own a thermometer. He has four kids, and he used to feel their heads all the time. “They’re now all grown adults,” he told me. “They all survived me and my lack of concern for fever.” He advises parents not to waste their money on fancy thermometers that probe the ear or forehead. “Just buy an ice-cream cone for your kid; it’s a lot better,” he said. “And if you really feel you need to know your child’s temperature—if it’s an infant—go up their butt” with a rectal thermometer.

Infants are a special case: Tiny babies with any sort of fever could need treatment right away. But for parents who are beyond that stage, your plan of action will be easy: I will not take their temperature. No parent should be taking temperatures. Just place your hand against their forehead, or use your lips instead. Perhaps your child has a fever. Or maybe he just needs a kiss.