Determining Reference Intervals

I’ve enjoyed an interesting exchange with an engineer who contacted the American Society for Clinical Laboratory Science Consumer Forum. She was concerned that her blood specimen, which had been sent to two laboratories for complete blood counts, had generated the same hemoglobin result from each, however one reported the value as normal and the other as elevated. Applying an engineer’s precision, she assumed one laboratory had made an error. Here is my response.
Hello and thank you for your question.  Each laboratory is required to develop its own reference interval (RI, also called reference range or normal range) for each assay it performs, so you will always see small differences among laboratories. The differences arise partly because each laboratory serves a slightly different population, and partly because technology varies from laboratory to laboratory.

If the RI differences between two laboratories arose only from differences in their normal populations, a person could conceivably be told their hemoglobin is elevated by one laboratory and normal by another. For instance, at laboratory A, the normal hemoglobin range for men is 14 to 18 g/dL, whereas at laboratory B, it is 13.5 to 16.8. Both laboratories employ the same analyzer, so if I sent an identical tube of my blood to both, and the result were, for example, 17.2 from both, laboratory B would call it elevated and laboratory A would call it normal. The difference between the two institutions could probably be traced to the ethnic and age mix of their respective populations. My physician will realize that a slightly elevated hemoglobin, as reported at laboratory B, would have no clinical consequence.

If the RI differences between two laboratories were caused solely by instrument technology, however, you could expect to see variance in the actual test result. In our example, laboratory A uses equipment that employs the cyanmethemoglobin assay, which provides a hemoglobin concentration by lysing (rupturing) red cells using a detergent solution, mixing with a non-lethal stabilizing solution of potassium ferricyanide, and measuring the resultant clear red “cyanmethemoglobin” solution using a fixed photometer and a flow cell.

Laboratory C, by contrast, uses an instrument that does not lyse the red cells, but instead passes them intact single file through a laser field, measuring the hemoglobin concentration of each red cell and computing the final concentration in g/dL of whole blood. Both methods are equally precise and accurate, but the two technologies report different results. The published RI from laboratory C is 13.2 to 16.1 g/dL So, if I sent the same tube of blood to both laboratories, laboratory A might again report 17.2 g/dL, whereas laboratory B might report 16.3 g/dL. Both would categorize the result as slightly elevated. Again, my physician would apply her clinical judgment to these results.

In reality, each laboratories’ RI always reflects both technology and population, so the differences among laboratories are never clear-cut like my theoretical examples. Actually, hemoglobin varies rather little from institution to institution compared to analytes such as cholesterol or blood sugar. Fortunately, physicians understand the variations in RI and are able to interpret laboratory results in the light of their full clinical information with the

assistance of clinical laboratory scientists.

As an engineer, you may feel skeptical about this explanation. Keep in mind that most biological systems lack a weighed-in standard and are instead validated to preserved human blood calibrators whose primary values are generated from “expert laboratories.” Thus some variation is inevitable.