Those of us who grew up in the north can sometimes tell how cold it is just by looking out a window. It is so noticeable that it can reveal continuity errors in films (one that comes to mind is Men in Black, where the seasons seem to change from scene to scene). But it is also subtle, something that is often hard to nail down. "It looks cold out." Why? What in particular makes it "look cold"? Is it a character to the light, the texture of the steam from exhaust, chimneys, and manholes? How accurate is that assessment?
In medicine we are often faced with such gestalt opinions. The practice of office-based medicine requires a high tolerance for uncertainty. Unlike treating hospitalized patients, the doctor rarely has immediate access to lab results, complex imaging studies, and frequent measurements of vital signs. Knowing the difference between someone who "looks sick" and someone who doesn't is a critical skill.
Are we good at it? Like all things that rely on human intuition, probably not, or at least we likely over-estimate how good we are, so in medicine we try as hard as possible to develop objective criteria and decision tools to help us along. Take pneumonia for example. Community-acquired pneumonia (CAP, contrasted with pneumonia acquired in health care settings) is a common bacterial illness. Once known as The Captain of the Men of Death, its rank has been diminished by the invention of antibiotics. But it is still an importaint illness. There are about 4 million cases diagnosed in the US each year, with more than a million hospitalizations, and around 53,000 deaths.
Pneumonia is diagnosed by a good history and physical exam, and an X-ray showing lung tissue filled with fluid (mostly pus). We know which organisms usually cause CAP and can treat it successfully most of the time with an empirical antibiotic choice. The harder choice is whether to treat someone at home or in the hospital. The choice is often made on the basis of how "sick" the patient looks, but this strategy has obvious drawbacks. Perhaps as many as half the people admitted to the hospital with pneumonia could be safely treated at home. Being in the hospital adds to the cost and the danger of treatment, increasing the risk of hospital acquired infections such as C. dificile colitis. Of course, failing to hospitalize someone who is ill also has potential consequences.
So doctors have come up with various criteria to help them decide when a patient can be safely treated at home or when to send them to the hospital. Most of these criteria explicitly include the severity of the patient's clinical condition, the risk of death and complications, and other less tangible characteristics such as how likely it is that the patient will take medication properly at home. Online calculators can help put the data together.
But sometimes, whatever the objective data, someone just "looks sick", and that sort of determination is at the heart of medicine. If the decision tools tell me someone can probably be treated safely at home (and it's "probably", not "definitely"), and they look too sick to me, I'm putting them in the hospital. This is hard to teach, and doctors probably get this wrong from time to time. Still we do the best with what we have.
Sometimes I envy engineers.