The reference range printed beside a result on an Australian pathology report is a statistical construction. A marker is measured across a sample of apparently healthy people, the values are ranked, and the central 95% becomes the reference interval. The outer 2.5% at each end is defined as “abnormal.” Everything in between is reported as “normal.”¹

This is a deliberate, well-governed process — international bodies set the standards for how reference intervals are derived.¹ But it is worth being precise about what the output is. A reference interval is, by definition, drawn from “apparently healthy” individuals, and it is distinct from a clinical decision limit, which is derived from people with known disease.² “Normal,” in this context, means common in the sample — nothing more and nothing less.

It also means the range carries the variability of the sample. Reference intervals stretch to accommodate wide differences between individuals, and a range built to cover a broad population can become so wide that, as one review of the field puts it, it risks losing its usefulness for any single person.²

A reference range is a screening instrument. Its job is to answer a binary question: is this result common enough to be unremarkable, or unusual enough to warrant attention? That is exactly the right question when the task is detecting disease, and the reference range does it well.

An optimal range asks something else: where does this marker sit when the underlying system is performing well? It is a functional lens, not a statistical one, and it is generally narrower — a band inside the reference range rather than a replacement for it.

The two are not in conflict. They are different instruments for different purposes, and the distinction between a reference interval and a decision-relevant limit is well recognised within laboratory medicine itself.² The practical point for anyone reading their own results is simply this: a result clearing the reference range answers the screening question. It does not, on its own, answer the performance question.

The deepest reason a single threshold can mislead is that biology is mostly continuous. For many markers, the relationship with health outcomes is a gradient — risk climbs steadily as a value drifts, long before it crosses a line.

Glucose metabolism is the clearest example. In a large community study of adults without diabetes, HbA1c was associated with cardiovascular disease and future diabetes in a continuous, graded fashion — risk rose across HbA1c bands that all sat below the diabetes cut-off.³ The state described as prediabetes — glucose markers above normal but below the diabetes threshold — already carries measurable links to early kidney, nerve and eye changes and to macrovascular risk; the underlying dysfunction begins well before glucose itself moves.⁴

Other markers behave the same way. A pooled analysis of observational studies found that a homocysteine level around 3 µmol/L lower was associated with modestly lower risk of ischaemic heart disease and stroke — a graded association across the range, not a switch that flips at a cut-off.⁵ High-sensitivity CRP is formally stratified into tiers of cardiovascular risk — broadly under 1, 1 to 3, and above 3 mg/L — values that can all be reported within a laboratory’s “normal.”⁶

A single threshold compresses a gradient into a yes or no. For screening, that compression is useful. For understanding where someone sits on the curve, the gradient is the part that matters.

A few markers show the gap concretely.

Ferritin. In a randomised trial, non-anaemic women with unexplained fatigue and a ferritin at or below 50 µg/L improved with iron supplementation — fatigue fell by 29% against 13% on placebo.⁷ A later systematic review of iron-deficient, non-anaemic adults reached the same direction: supplementation reduced fatigue.⁸ A ferritin in the 20s or 30s clears the lower limit of most reference ranges and is also, on this evidence, a level at which symptoms commonly respond.

Thyroid. The upper limit of the TSH reference range has been examined closely. The reference range printed on a standard Australian pathology report commonly extends to around 4.0 to 4.5 mU/L — a result is not flagged until it passes that mark. Yet evidence reviewed in the endocrine literature indicates that more than 95% of genuinely healthy individuals have a TSH below about 2.5 mU/L, and that older reference samples were partly populated by people with undetected thyroid dysfunction, which pushed the upper limit higher than a truly healthy population would support.⁹ The gap is the point: a TSH can climb from 2.5 toward 4.5 — close to doubling — and never once be marked abnormal.

Vitamin D. Clinical-practice guidance frames 25-hydroxyvitamin D in terms of deficiency, insufficiency and sufficiency thresholds, and notes how common low vitamin D is across the population.¹⁰ A 25(OH)D that clears a laboratory’s lower limit can still sit below the level that guidance treats as sufficient.

None of these is an error in the reference range. They are the predictable result of the reference range answering a different question from the one a person asking “am I at my best?” is actually asking.

Reading a panel through the optimal lens does not mean discarding the reference range. It means adding a second question to it.

When you look at a result, note more than whether it is inside the range. Note where in the range it sits — low, middle or high — and which direction is favourable for that particular marker, because it is not the same for all of them. For ferritin or vitamin D the favourable direction is generally upward within the band; for fasting insulin or hs-CRP it is downward. There is no universal instruction to “aim for the middle.”

Note the trend, too. A single result is one point; the same marker measured again months later shows a direction. Plotted over six months, then a year, then two years and beyond, the path a marker takes reveals what no single reading can — and the longer the series runs, the clearer the signal becomes. A six-month protocol is enough to move the markers; the years that follow are where the trajectory becomes unmistakable.

And remember that the range itself is not a fixed constant. Reference intervals vary with the population sampled, the laboratory, the method, and physiological factors such as age and sex.² Reading results well means holding the number, the direction, and the question being asked all at once. That is the work — and it is the work an Elemental Protocol panel is built to do.