Why two laboratories can disagree about the same blood

One of the most confusing things about a laboratory report is that the reference range beside a result is not a universal standard. It belongs to that laboratory, for that method, for the population it serves.

Two laboratories can measure the same sample, agree closely on the number, and print different ranges beside it. Neither is wrong, and the difference is not a sign that one of them is careless.

Where a reference range comes from

Most reference ranges are built by measuring a group of people considered representative of the population the laboratory serves, then taking the middle 95 per cent of their results. Everything above and below that band falls outside the printed range.

That construction has two consequences worth understanding before reading any report, and the first is pure If the range covers the middle 95 per cent, then one result in twenty from a perfectly well person falls outside it by design. On a panel of twenty markers, at least one result outside the range is more likely than not. This is why a single marginal value rarely means much on its own, and why clinicians look at patterns rather than individual flags.

The second is that a range describes the group that was sampled. A laboratory serving an elderly population, a paediatric hospital and a sports medicine clinic will each arrive at different intervals for the same marker, because the people they measured differ.

Why the measurement method matters

Many tests do not measure a substance directly. They measure something that stands in for it — an antibody binding, an enzymatic reaction, a change in light scattering — and different manufacturers build those reactions differently.

Two analysers can be equally accurate and still produce numbers on slightly different scales, because they are detecting the target in different ways. The manufacturer supplies a reference range validated for that method, and the laboratory adopts or adjusts it.

This is why comparing a result from one laboratory with a result from another is less straightforward than comparing two numbers. Where a trend over time matters, using the same laboratory removes one substantial source of variation.

Ranges that are not ranges

Not every reference value is an interval. Some are printed as an upper limit, some as a lower limit, and some as an expected word rather than a number.

Cholesterol is often reported against a target rather than a population range, because the value considered desirable depends on overall cardiovascular risk rather than on what is typical in the population. A target is a clinical decision point, not a description of what most people have.

Infectious disease tests are frequently reported as an expected result — negative, or not detected — with no numeric interval at all. Others report a titre or an index where the number is only meaningful relative to a cut-off defined by the assay.

Ranges that shift with the person

Many reference ranges differ by sex, and a great many differ by age. Alkaline phosphatase in a growing adolescent sits far above the adult range, and that is expected rather than remarkable. Creatinine ranges differ by sex largely because they track muscle mass.

Pregnancy changes the expected values for a long list of markers, including thyroid tests, haemoglobin and several liver measurements. A laboratory that knows a sample is from a pregnant patient may print pregnancy-specific ranges; one that does not know will print the general adult range beside a result it does not fit.

What this means when reading a report

Compare each result against the range printed beside it on that report, and not against a range found elsewhere. That is the comparison the laboratory intended, using the method it actually used.

A result marginally outside a range is a prompt for a conversation rather than a conclusion. What it means depends on the person, their history, their medicines and their symptoms — which is exactly the information a laboratory does not have and a clinician does.

It is also worth asking whether a value is being compared against the right range at all. If a report was produced without the laboratory knowing about a pregnancy, or for a child using adult ranges, the number may be accurate while the comparison beside it is not.