Optimal Ranges vs Reference Ranges: What Lab Numbers Can and Cannot Tell You

“Your results are normal.” For many patients with persistent symptoms, few sentences are more frustrating. For many practitioners, few are more misunderstood. A result inside the reference range does not mean the result is ideal, and a result outside it does not necessarily mean something is wrong. Functional medicine has responded to this ambiguity with the idea of optimal ranges, which can be useful and can also mislead. Using either kind of range well starts with understanding how ranges are made.

How reference ranges are constructed

A reference interval is typically derived by measuring a marker in a group of people selected as healthy, then defining the range that contains the central 95 percent of their results. Internationally recognised guidance recommends that laboratories establish or verify their intervals using a sufficiently large reference population and appropriate statistical methods, and partition them where needed by age, sex or other factors.

Several consequences follow directly from this method.

• By definition, 5 percent of healthy people fall outside the range. A result just outside the interval is often normal for that individual.
• The more tests performed, the more likely an abnormal result by chance. A healthy person tested for many independent markers has a substantial probability of at least one result outside its interval.
• The range reflects the reference population. If that population includes many people with undiagnosed conditions, such as insulin resistance or iron deficiency, the range may be wider than is ideal for health.
• Ranges differ between laboratories and methods. Comparing results across laboratories requires caution.

3 different kinds of threshold

Much confusion arises because 3 different kinds of threshold are often discussed as if they were the same.

Clinical decision limits are often already different from reference intervals. For lipids and glucose, for instance, the thresholds that guide treatment are based on outcome evidence rather than population distribution, which is why laboratories often report them alongside or instead of a simple range. The question for any optimal range is whether it has a similar evidential basis.

When narrower targets are well supported

Some narrower interpretations are well grounded.

Ferritin and iron deficiency. Ferritin reflects iron stores. Many laboratory reference intervals have lower limits well below the levels at which iron deficiency is increasingly likely, and several guidelines use a threshold around 30 micrograms per litre, rather than the lowest laboratory limit, to support a diagnosis of iron deficiency in appropriate contexts. Ferritin also rises with inflammation, so it must be interpreted alongside inflammatory markers.

Cardiovascular risk markers. For LDL cholesterol, lower is generally better for people at elevated risk, and treatment targets fall well below the upper limit of any population range.

Glucose trajectory. Risk of type 2 diabetes rises continuously across the normal range of fasting glucose and HbA1c, so values in the upper part of the range carry more risk than those in the lower part, even though neither is diagnostic.

When the evidence is contested or thin

TSH. Some practitioners advocate a narrower TSH range and treat symptomatic patients with values in the upper part of the reference interval. Population TSH rises with age, and trials of thyroid hormone treatment in older adults with mild subclinical hypothyroidism did not show improvement in symptoms or quality of life. A narrower optimal range for TSH is therefore not well supported as a basis for treatment, although trends and antibody status can reasonably prompt monitoring.

Vitamin D. Different expert bodies have recommended different thresholds for sufficiency of 25 hydroxyvitamin D, with some considering 50 nanomoles per litre adequate for most people and others proposing higher targets. Large trials of vitamin D supplementation in people not selected for deficiency have generally not shown reductions in major outcomes such as cancer or cardiovascular disease. The appropriate target remains debated.

Fasting insulin. Fasting insulin can help identify insulin resistance, but assays are not well standardised across laboratories, so universal optimal cut offs are difficult to justify.

Many other markers. Optimal ranges circulate for a large number of markers without any published derivation. Where the source of a range cannot be identified, it should be treated as opinion.

Biological variation and meaningful change

Every marker varies within an individual from day to day, and every measurement has analytical variation. Laboratory medicine uses the concept of the reference change value, the smallest difference between 2 consecutive results that is likely to reflect true change rather than variation. For some markers this is small; for others it is surprisingly large. A change in a patient’s result may be noise even when it looks meaningful, and a stable result within range may still be a meaningful departure from that person’s usual level. Personal baselines, when available, are often more informative than population ranges.

Worked examples

A ferritin of 18 in a fatigued woman

A 32 year old woman with fatigue and heavy periods has a ferritin of 18 micrograms per litre, within her laboratory’s reference interval, and a normal haemoglobin. Read against the reference interval alone, the result is normal. Read in clinical context, with her symptoms and menstrual history, iron deficiency without anaemia is likely, and her doctor can confirm and treat it and investigate the heavy bleeding. Here a narrower interpretation is well supported.

A TSH of 3.8 in a tired man

A 64 year old man with fatigue has a TSH of 3.8, normal free T4 and negative thyroid antibodies. An optimal range approach might label this suboptimal. Evidence does not support thyroid hormone treatment here, and his fatigue is more likely explained by other factors on his timeline, such as sleep and activity. His TSH can be rechecked in the future, but it should not distract from the more likely causes.

A vitamin D of 55

A healthy 40 year old has a 25 hydroxyvitamin D of 55 nanomoles per litre. Some bodies would consider this sufficient; others would suggest supplementation. The pragmatic approach considers her risk factors, such as limited sun exposure, darker skin or osteoporosis risk, explains the uncertainty honestly and agrees a reasonable plan.

Talking about results with patients

• Avoid describing results simply as normal or abnormal. Explain where the result sits, what it means for this person and how certain that interpretation is.
• When using a narrower target, say where it comes from and how strong the evidence is.
• Emphasise trends and patterns over single values.
• Avoid creating anxiety about results that are within range and consistent with the clinical picture.
• Link results to the plan: what will change, what will be monitored and when.

A balanced position

Reference ranges are statistical descriptions, not definitions of health, and treating them as the final word can miss real problems. Optimal ranges are sometimes well founded and sometimes not, and treating them as universal truths can create problems where none exist. The skilled practitioner knows which kind of threshold applies to each marker, understands the evidence behind it, and reads every number as one piece of a clinical story.