This is a question about how doctors carry out diagnoses in practice, not one looking for medical answers.

My question is, statistically most cases are horses, but how do you find the zebras? According to Indiana University School of Medicine, a rare disease is defined by having less than 200,000 cases, and approximately 1 in 10 Americans has a rare disease. The discrepancy here is that while it is exceedingly unlikely to have any particular rare disease, there are enough of them that the odds of having any one are not that low. Evidently, doctors and nurses spend the vast majority of their time dealing with common and generally quite obvious conditions. Sepsis, vehicular accidents, heart issues, complications in childbirth, arthritis, falls, cancer, these are daily events, but rare diseases, while common in the aggregate, are each most likely to be seen by a given health care provider once or twice in their lifetime, and yet clearly there's still need to diagnose them effectively.

When presented with a patient who insists something is deeply wrong with some supporting evidence, seems impaired yet not under the influence of drugs or alcohol, and initial tests come back normal, what next?

How are rare diseases (or less-common-than-common diseases even) recognized and diagnosed?

  • @IanCampbell fair point, I would edit but anongoodnurse beat me to it :) Commented Aug 17, 2021 at 6:34

2 Answers 2


My question is, statistically most cases are horses, but how do you find the zebras?

The answer is... complicated. I will answer first from my experience as a (US) physician, and secondly, from a general standpoint in (US) Medicine.

While we spend the vast amount of our time as physicians treating "horses" (as it were), we spend the vast amount of our training (medical school and some of Residency) learning about zebras.* So, we know a lot about zebras, and the fear of missing a zebra is what keeps a lot of us on our toes. (There are lots of publications that feature a rare case in every issue or periodically.)

Also, there is a not terribly well known and informal source of further assistance: online groups for physicians consisting of physicians where difficult cases are discussed in the hopes that a solution might be found.

But, we can't recognize all the zebras, and this is where the line of referral comes in: a primary care physician or primary care provider refers to a specialist they think might help; if the specialist can't figure it out, they refer to a subspecialist, and on it goes up the chain. It may take time (in some cases, up to a decade), but eventually, one hopes the patient meets a specialist that has seen a zebra they have never forgotten and it is the zebra that the patient happens to have. A friend of mine finally got a diagnosis after 7 years of referrals. It's a long time, but it eventually got diagnosed. (Their condition is literally 1/1,000,000.)

*I recall one lecture on Ewing sarcoma (2.9/1,000,000) where, at the end of the lecture, the professor said, "But don't worry, you'll never see this in your entire lifetime." And as it so happened, I saw two.

Sometimes pure luck plays a roll. I saw a patient who had seen 4 specialists in the two weeks prior to his visit with me (a generalist). I performed a test that no one else had done, which led to a diagnosis that was so rare, only 5 people in the US had it before him.

In residency, our service had a patient with a fairly rare disease (1/100,000) which went undiagnosed, and the attending thought it was psychogenic, which led to embarrassingly atrocious care for the patient. Luckily there was a senior resident who researched the symptoms diligently and came up with the correct, physiologic diagnosis.

  • At the end of the journey, is there really any difference between how the rare disease and the common disease are diagnosed? Don't they both end with checkmarks on a differential diagnosis? The only difference being the things listed on the differential dx.
    – Carey Gregory
    Commented Aug 17, 2021 at 5:15
  • 1
    @CareyGregory - I'm not quite sure what you mean. If you mean, "Isn't every diagnosis just a matter of having the right pick on the differential?" Well, yes, I guess, but that's like saying driving a VW Beetle is basically like driving a racecar: you step on the gas and steer. I can do the former but not the latter. I need great good luck, or practice (having seen it before) to diagnose a true zebra. I've done it a few times, but it's not the same to me at all. Commented Aug 17, 2021 at 12:08
  • It's all good. I wasn't trying to diminish the accomplishment of diagnosing a rare disease. It was just an observation that once you get past the lucky break, the lecture you heard 10 years ago, or whatever it was that led you to the key, the rest was standard diagnostics. It was a comment aimed at the OP and audience, not so much your answer.
    – Carey Gregory
    Commented Aug 18, 2021 at 3:34
  • 1
    @CareyGregory _ Ah, understood, and as I said before, that's true. I wasn't trying to toot my own horn, btw. It's just... I feel very grateful when I can make an uncommon diagnosis, especially if the diagnosis was missed by people who should not have. I'm not proud of my "acumen", I feel grateful for it. I know all too well how easy it is to be wrong to be fooled into feeling proud. A proud doc (imo) is a dangerous doc. The "test" the other specialists didn't do? A thyroid exam. That simple. Thyroid was huge; got a CXR -> huge mediastinal mass. CT -> poorly organized. Dx on bx. (Not by me!) Commented Aug 18, 2021 at 15:51

The UK National Health Service estimates that 80% of rare diseases have an underlying genetic etiology¹. Thus, rare diseases diagnosis now and in the near future is mostly a function of:

Having a screening test with the appropriate sensitivity and specificity to consider multiple rare diseases simultaneously

There are probably more than 7000 such genetic diseases², and no one clinician can possible be aware of all of them other than superficially. Thankfully, there is a screening test which accurately identifies the vast majority of known genetic diseases, exome sequencing. Moreover, exome sequencing has favorable sensitivity and specificity³.

Thus, it seems most likely that diagnosis of the great majority of rare diseases mostly relies on being evaluated by a clinician comfortable with the application of broad genetic testing.

  • Although you've not said anything untrue, it sounds too straightforward imo. Many rare diseases take a decade or more to diagnose, in spite of seeing the appropriate (sub)specialist. It often takes being seen by more than one physician in the same subspecialty to diagnose a true "zebra". That's why it's so very frustrating for the patient; it seems to them that they're just being "shuffled around", when that "shuffling" is just trying to locate ta person familiar with that rare disease. Commented Aug 18, 2021 at 16:00
  • @anongoodnurse It turns out there is more literature on this subject than I would have thought. It turns out more than 3/4th of rare disease is genetic and so my former second point is even more important.
    – Ian Campbell
    Commented Aug 18, 2021 at 16:21
  • 1
    @anongoodnurse Since exome sequencing has become more available and affordable only relatively recently, I suspect not so many clinicians in the field have incorporated it into their routine practice yet -- would you agree?
    – Armand
    Commented Aug 18, 2021 at 17:09
  • @IanCampbell - Yes, I agree that most rare diseases are genetic, but my point isn't about genetics. It's about diagnosis. As Armand correctly stated, most MDs don't order genetic testing yet. They would refer to a geneticist (yet another subspecialist.) Commented Aug 18, 2021 at 17:32
  • 1
    @IanCampbell - "...because I know how to order and interpret an exome." Exactly. Until it's taught in residencies and then med school, it's not at all common knowledge. Commented Aug 18, 2021 at 17:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.