Friday, January 15, 2010

Normal

Where do normal values come from? Any time we have numerical data such as a blood pressure or lab value, we compare it to a normal range. But how are normal ranges determined? For example, normal heart rates are between 60-100, but certainly well-trained athletes have a physiologic bradycardia (pulse less than 60) and exercise-induced tachycardia is normal. Who decides these numbers? Weirdly enough, people don't think about this question even though every time we present a patient, we rattle off "chemistries were normal" and "liver function tests are elevated."

One method of determining normal values seems reasonable: test a lot of "normal" people in the population and graph the distribution. If it falls into a normal distribution, a reference range can be set based on standard deviations. Indeed, blood tests like "calcium" are set according to this methodology; each lab tests "normal" people in the community and calibrates its reference range based on the statistical outcomes. That seems reasonable, but in some cases, it doesn't work very well. For example, pediatric growth charts were initially developed in 1978 by the CDC in this manner. However, the "normal" population used was from Fels research institute data which studied white formula-fed middle-class infants living in southwestern Ohio between 1929 and 1975. Although it was the best data at the time, it's hardly representative of the U.S. population. Nevertheless, growth charts became widely used, and nobody questioned this "normal range" until 2000 when the CDC updated growth charts to be better representative of the population.

Another method of determining normal values is using clinical outcomes. For example, clinical studies have found that fasting glucoses of 126 or greater have associated medical conditions like retinopathy. As a result, the cutoff for diabetes is set at fasting glucose >=126. In the same way, hypertension is defined not based on a normal distribution of blood pressures, but rather when we start seeing the clinical effects of high blood pressure.

Although on a day-to-day basis, it may not be crucial to know where normal values come from, I think having a sense of how they are determined is an important part of being a well-rounded physician.

Image of a female growth chart shown above is in the public domain, from Wikipedia.

No comments: