What is Normal?
Part two: Lab work
In medicine, people often put great faith in testing and instruments. For example, studies have shown that for asthma, using an inhaler with a spacer is as effective as using a nebulizer. The former has the advantages of being easier to use, portability, and requiring less time. Nevertheless, many people, including many doctors, believe the latter is better, not because of any data, but simply because it uses a machine. So what about lab testing? This can give objective findings, and one assumes it will easy to determine normal from abnormal.
However, this is not always so. As with growth curves, abnormal is defined as a certain percentage of the population, regardless of health status. Furthermore, if you run large numbers of tests on a healthy person, the odds increase that at least one of them will fall outside the ‘normal’ range. What we are often looking for is the degree of abnormality. Suppose I am concerned that your child may have a liver disorder, and his liver enzyme test comes back at 32, with 30 being the upper limit of normal. Is that a concern? Almost certainly not, whereas a finding of 150 would be.
Blood testing for food allergies is another common area where testing is not as definitive as you might expect. If a random allergy panel is ordered, many children will have abnormal results, yet the child can tolerate those foods without difficulty. Without clinical correlation of symptoms, the tests are often worthless, and do not warrant avoiding those foods.
A similar area, undergoing further revision at present, is with vitamin D levels. Until recently, levels below 30 were felt to need treatment, but newer studies have called this into question. Thus, if there is any concern about vitamin D, it makes more sense to make sure your child is taking in enough, usually by milk (which has vitamin D added to it), or a vitamin supplement (non-milk dairy products are good sources of calcium, but usually do not have added vitamin D), rather than measuring a level.
When I first started doing pediatrics, we used to do many tests routinely, such as a yearly urine. We no longer do this, because it has been shown to be of little value. Thus, we would often find protein in the urine, a marker for a condition called nephrotic syndrome. However, protein can be found normally in children if they have been up and about during the day. If we found protein in the urine on a random sample, we would have the parents collect a urine first thing in the morning. This would invariably be negative, and then we would not worry about it. I have seen many patients with nephrotic syndrome over the years, but never one found by a ‘routine’ urine specimen; they always had clinical findings which directed me to look for it.
Compounding the situation further, no test is perfect, and results can be misleading. This comes up particularly when we consider Bayes’ theorem. You can find the mathematics behind this on-line, but basically, if your child has almost zero probability of having a given disease, a positive test changes that to just being very improbable. Results are similar if a child is very likely to have a condition, but the test is negative. Testing is most valuable when the physician is uncertain.
We can look at Lyme disease as a specific example. If your child comes to the office in the summer with the classic rash (Erythema Chronicum Marginatum), I will treat for Lyme disease, regardless of what a test showed, so there is no need to test. Conversely, testing for Lyme disease in someone with vague, non-Lyme complaints is not a good idea; a positive test would be a false positive, not indicative of Lyme disease. I remember one patient in particular, who I felt was depressed. The parents did not agree with the diagnosis, and went to see a self-proclaimed ‘chronic Lyme expert.’ The child’s Lyme test was negative, but he tested positive for a tick disease seen only on the west coast, a place he had never visited. The child wound up leaving the practice to be treated with long-term antibiotics by the ‘expert’ for a disease he almost certainly did not have; I doubt it helped.
Even when a test is correct, it may still not mean anything. For example, some children are strep carriers. They always have strep in their throat, living with it just fine, not contagious, and not needing treatment. However, if I were to randomly culture them, I would find strep. We don’t do such random cultures, of course. Now suppose that same child came to the office with a runny nose, cough, no fever, no vomiting, no enlarged lymph glands, and a scratchy throat. They do not have a ‘strep throat’ either, but if I were to culture them, ‘just to be sure,’ the test would be positive. This would then mean the child needed to be quarantined for a bit, and take ten days of antibiotics (which will usually not eradicate the strep in a carrier), all for nothing.
Is there a role for any routine testing nowadays in pediatrics? Yes. Many infants are anemic, so we test their blood at 9 months. We are seeing more cholesterol problems, and so we recommend testing for this somewhere between the 9–11-year visit. Lastly, I harken back to elevated BMI (see part one of this series). With the increase we are seeing in type 2 diabetes, even in children, current thinking is to look for this. The test is a hemoglobin A1C, and I order it in older children who are above the 95th percentile in BMI, or the 85th percentile with other factors (e.g., a positive family history).
Next up: medical disorders