Reference Ranges and Repeatability

What a reference range is

A reference range, sometimes called "normal values", is the span of results seen in a group of healthy people. MedlinePlus says the high and low ends are based on test results from large groups of healthy people, and that different ranges may apply to different groups, such as children and adults. It also warns that the word "normal" can be misleading. [1] A laboratory-science review makes the same point: the ranges on a lab report help a clinician interpret a result against values for healthy populations. [2]

This page explains concepts and does not tell you what any of your values mean.

Three things a reference range does not mean

• Outside the range is not always disease. A result above or below the range may or may not signal a health problem, and it is common for healthy people to sometimes fall outside it. [1]
• Inside the range is not a guarantee. Some people with health problems have results considered normal, so more testing may be needed when there are symptoms. [1]
• It is not a treatment target. A reference range describes a group, whereas a clinical decision limit is set to guide action. The review notes that confusing reference intervals with clinical decision limits remains an issue. [2] For the marketing idea of "optimal" ranges see reference ranges vs "optimal" ranges.

Why you cannot borrow someone else's range

Labs often use different methods and so different ranges, and tests use different units. MedlinePlus advises checking the range printed on your own report, notes that you cannot compare results from different labs, and suggests using the same lab when tracking trends over time. [1] It does not publish normal ranges for most tests for this reason. [1]

Repeatability: why the same test can give different numbers

Even when nothing has changed, results move for three broad reasons.

1. The body. Natural day-to-day (biological) variation. A meta-analysis of 37 studies of creatinine found within-person variation of about 5.2%, with wide differences between studies. Using an assumed measurement variation of 1.1%, it estimated that an estimated kidney filtration rate (eGFR) would need to change by about 16.5%, or by a conservative 12.5% after excluding outliers, before a change is clearly more than random fluctuation. [3] That figure belongs to one test; other tests differ.
2. The sample and preparation. Foods, medicines, supplements, hard exercise, menstrual cycle and whether fasting instructions were followed can affect some results. [1] In a study of 9.8 million results, fasting under 8 hours was linked to higher glucose, triglycerides, gamma-glutamyl transferase and lipase, whereas differences between 8 to 12 and over 12 hours of fasting did not exceed accepted performance limits. [4]
3. The method. Different labs and kits measure slightly differently. See analytical validity.

Using this when you track a number

• A single value outside the range, or a small change between two values, may be ordinary variation. The meta-analysis above exists because clinicians need a way to tell true change from random fluctuation. [3]
• Compare like with like: the same lab, similar timing and preparation. [1]
• A change in a marker is not automatically a change in health. See biomarker tracking and at-home tests.
• Supplement-related tests have their own limits; see deficiency testing and its limits.

Evidence levels on this page

The statements about reference ranges are established (they come from a government health information site and a published review). The variation figures come from single meta-analysis and single-hospital observational data, so they illustrate the idea but are not universal numbers for every test or person.