There is a useful thought experiment I like to share with my students that goes like this:

Imagine that you are an emergency paediatrician, seeing a six-year old child with a fever and a sore, red throat. What do you do?

Most of these children have viral illnesses, but about five per cent have strep throat; you might consider doing a throat swab to find out. Unfortunately, about 10 per cent of all children, including those with a viral infection, carry strep as a "colonizing bacteria". That is, the strep is causing no illness; rather the bug is living happily in the throat, causing no problems whatsoever.

A throat swab can't discriminate; colonizing strep triggers a positive result just as readily as strep throat. So do the math: five per cent with strep throat, 10 per cent with a viral infection and colonizing strep, the vast majority that remains with viral infection and no colonizing strep. That means 15 per cent of the population will have a positive throat swab. Only a third of those, however, will actually have a bacterial infection, the remainder will have a viral infection, and a false positive result for strep throat.

In other words, the combination of fever, sore throat and a positive throat swab means the odds are 2 to 1 against the child having strep throat.

Now, experienced emergency docs don't swab every kid with a red throat.

Rather, we use our clinical skills to weed out those likely to be viral, and swab the rest. There are limits to how well we can do this, though, and even in the best of hands as many as 1/3 of all children "diagnosed" as strep throat will actually have something else causing their symptoms.

Students get a bit bent out of shape by this example; many simply refuse to believe it, nodding at me politely while wishing I would get on to teaching them something important. Parents find this even more difficult. If there is a fundamental rule of emergency pediatrics, it is this: parents want to know what's going on. We all want to believe that with persistence and science at our disposal, we can get to the bottom of things, that we can, in other words, diagnose.

I am not so sure. Diagnosis, you see, just isn't what it's cracked up to be.

A throat swab is a good example of a proxy test. The disease, in this case, is strep throat, but the proxy, the only thing we can really test, is the presence of the bacteria that can (but doesn't necessarily) cause illness. Forgetting this distinction is a common cause of diagnostic error.

Another test and a different problem: screening for blood pressure. If your pressure is normal when you are sitting in the doctor's office in a thin little paper gown anticipating this year's prostate exam, it is very likely that it is normal when you are sitting at home having a beer. Similarly, if your blood pressure is high, it might just be nerves; measure again in a week or two and your pressure would be normal. Screening for blood pressure picks up almost all who have hypertension, but it also picks up a lot of people who don't, people who are, really, just anxious. It has a high false-positive rate; it is sensitive, but not specific.

With blood pressure, we have an opportunity for diagnostic closure; keep coming back and eventually we will be able to decide what your usual pressures are. A "diagnosis" in this setting has more meaning because it is the result of a slow accumulation of evidence, in a variety of settings, over a period of time. Chronic diseases benefit from this sort of approach, and the diagnostic certainty that we can attach to some of these conditions becomes very good.

But what about acute care? The child with a head injury, or high fever, or severe abdominal pain doesn't have the luxury of waiting a few weeks so that we can make a clear decision. Yet the CT scans, blood tests and x-rays we use to probe these conditions are subject to the same problems as the examples above. The issues of sensitivity, specificity and proxy testing are not just prevalent in medical care, they are fundamental parts of the bedrock.

What can a "diagnosis" be said to mean, in these circumstances? Is having a diagnosis even an advantage? Imagine, for example, that our six-year old above has a different cause of sore throat, like an abscess in his upper airway. This could be rather dangerous; it might expand and compromise his breathing. Is "diagnosing" this child with strep throat helpful? Or will our certainty that we "know" what is going on keep him at home longer, even after he begins to deteriorate? He has a positive throat swab, after all.

Diagnosis as parents and some physicians see it is an answer; a proven cause of a child's illness. Diagnoses have weight and momentum, particularly if provided by an experienced, trusted, smooth-talking doc. Once a diagnosis is written on a chart, or given to a parent, we believe it, sometimes even discarding contrary evidence simply because it doesn't fit. Yet at first glance, a diagnosis seems to be required: surely if we discard the diagnosis, we lose the capacity to treat the patient?

The answer, actually, is that the diagnosis is unnecessary. We don't need to know what we are treating, rather we need to grasp and address the potential hazards. We need to let go of the need to know, and address the issues that we can actually get a handle on. In other words we need to quantify and manage the child's risks, and we can do this without the pesky diagnosis.

How? Well, the high sensitivity of the physical exam and other screening tests used in acute care mean that we are, paradoxically, much better at diagnosing the absence of disease than diagnosing disease itself. If you have a negative throat swab, the chance that you have strep throat is essentially nil. If your blood pressure is normal in the clinic, it is probably normal, period. We might not know what you have, but we can be pretty clear about what you don't have.

And when the throat swab is positive? The implication here is that your child has a heightened risk of strep throat, but that other (differential) diagnoses aren't discarded. We work within a diagnostic space, rather than narrowing ourselves to a single diagnostic entity. This is why, increasingly, you will see pseudo-diagnoses like "fever" or "throat swab positive pharyngitis" or "wheeze" at the bottom of an emergency chart, replacing "viral infection" or "strep throat" or "asthma". Open minds aren't paralyzed; we can still treat these children appropriately. But we signal to our colleagues, students, and parents our willingness to see our diagnoses challenged. This is a safe ideology, much more protective than the arrogance of certainty.

So is this just an intellectual shell game, in which we talk ourselves around the block and end up where we started, doing the throat swab and writing a prescription? Well, no. If you don't depend upon diagnoses, you can send parents home on a "safe pathway", engage them as partners in the care of their child, and have them more likely to return if their child seems to deviate from expectations. You tell them bluntly that you don't "know" what is causing their child's illness, but you can and do comment on their child's safety, produce a short list of possible causes, and you can do this with some evidence to back up your opinion. You don't give an "answer", you provide a discussion. In short, you might not answer the burning question "what is going on?", but you can answer the question "is my child safe?", and, by doing so, make that child safer.

And that, after all, is what parents really want to know.

Brett Taylor is an associate professor of pediatrics and emergency medicine at Dalhousie University. He works as an emergency paediatrician and researcher at the IWK Health Centre in Halifax. He is in the process of obtaining a Masters in Health Informatics, also through Dalhousie. His website for parents is available at www.thevirtualpediatrician.com.

[an error occurred while processing this directive]