What is a reference interval?

The key points

A reference interval is derived from results in a group of people judged to be healthy.
Region Norrbotten states that reference intervals are in most cases set so that about 95 percent of subjectively healthy people fall within the interval.
That means about one healthy person in twenty falls outside the interval without being ill.
Intervals differ between laboratories and by age and sex, because they are produced with different methods and on different groups.
1177, the Swedish national health guide, states that a single blood test usually gives no certain answer and that the result is weighed together with how you feel and with other tests.

What the number is compared against

A result sheet usually carries two things: your value, and an interval beside it. The interval decides whether the result looks calm or worrying, and most people read it as a line between healthy and ill.

1177 describes what it actually is. The doctor compares the sample against something called a referensintervall (reference interval) when interpreting your result, and the reference interval is arrived at by combining results from a large group of healthy people's samples. 1177 also notes that levels vary between different people, and that they vary between healthy people too.

A reference interval is therefore a description of how healthy people look when you measure them. It is a statistical summary of a group, and it is the starting point for interpretation rather than the conclusion of it.

How the interval is derived

The method is easy to describe. You measure the analyte in a group of people judged to be healthy, sort the values, and trim the outer edges.

Region Norrbotten puts the result this way in its sampling guidance: reference intervals are in most cases set so that about 95 percent of subjectively healthy people fall within the interval. In the scientific literature the limits are given as non-parametric 2.5 and 97.5 percentiles, meaning the values that 2.5 percent of the group sit below and above respectively.

Who those people are matters. Läkartidningen, the journal of the Swedish Medical Association, describes reference limits as calculated from the measurements of a group of reference individuals, preferably 120 or more, who should share the same age, sex, medication use and other biological sources of variation as the patients in question. Region Norrbotten adds that the intervals it publishes apply to adults unless stated otherwise, and that they are based either on its own studies or taken from the literature.

In the Nordic countries, several of the intervals in use came out of NORIP, the Nordic reference interval project, which Equalis describes as common Nordic reference intervals for a number of ordinary clinical chemistry analyses.

Why one healthy person in twenty falls outside

If the interval is built to hold about 95 percent of healthy people, the rest of the arithmetic follows directly. About 5 percent of healthy people land outside it. That is roughly one in twenty.

It follows from where the line is drawn. Choose a wider interval and you miss more sick people. Choose a narrower one and you flag more healthy people. The line has to sit somewhere, and it sits at 95 percent.

In practice that means two things. A value just outside the interval is not in itself a verdict of illness, and the size and direction of the deviation matter more than its existence. A value inside the interval is no guarantee either, since a person can be moving in the wrong direction and still sit within the limits for a while.

1177 states it plainly about individual tests: a single blood test usually gives no certain answer as to whether you have a disease, and the result is therefore weighed together with how you feel and with other tests.

Why intervals differ

Two laboratories can publish different intervals for the same analysis, and that does not have to mean something is wrong.

The first reason is the measurement method. Läkartidningen describes how some immunochemical assays' reference intervals, for thyroid hormones for example, apply primarily to the same analytical platform. Different instruments from different manufacturers produce different results, and the interval travels with the instrument.

The second reason is the group that was measured. Age and sex affect most analyses. For haemoglobin, 1177 gives separate intervals for women, for men, for newborns and for infants in their first month. Läkartidningen describes age variation as a common phenomenon, notes that most analyses change substantially in healthy children during the first weeks after birth, and that some values rise continuously with age.

There is also a different kind of limit that is easy to confuse with the reference interval. Läkartidningen separates the reference interval, which is statistics from healthy people, from the beslutsgräns (decision limit), which is a threshold for medical action. By tradition, clinical chemistry has used the reference limits as decision limits, but the two need not coincide, and for some analyses the limit that governs management is derived another way.

The practical conclusion when you read a result: do not compare a value from one laboratory against an interval from another. Read your value against the interval printed beside it.

Why a series says more than one test

A single value places you somewhere in a distribution. It does not tell you which direction you are moving in.

Repeating a test is standard practice in healthcare. Internetmedicin describes, for instance, how a mildly raised TSH alongside unspecific symptoms should be followed up with a fresh sample after 6 to 12 weeks, with TPO antibodies added, before any conclusion is drawn. A value becomes interpretable once there is something to compare it against.

Your own earlier measurement is often the best comparison. A value that has sat in the same place in the interval for several years means something different from the same value after a clear shift, even though both sit within the limits. The distribution among healthy people tells you what is common in the population. Your own series tells you what is common for you.

That comparison assumes comparable samples: the same analysis, preferably the same laboratory and method, and similar conditions around the sampling.

How to read your own result

Three things are worth taking away.

A value outside the interval is a reason to look closer. About one healthy person in twenty sits outside, and how far outside matters more than the fact that it is.

A value inside the interval means you resemble the majority of healthy people in the group the interval was built on. It does not rule out illness, and 1177 is a reminder that a single test usually gives no certain answer.

The interval belongs to the laboratory and the method that made the measurement. The number printed beside your own value is the only one to compare against. That is also why this article contains no intervals for you to measure yourself against. A threshold without its context is harder to use than no threshold at all.

Common questions

Why does 95 percent come up so often with reference intervals?

Because that is how the limits are set. Region Norrbotten states that reference intervals are in most cases set so about 95 percent of subjectively healthy people fall within them. The literature describes the limits as non-parametric 2.5 and 97.5 percentiles, meaning the outermost 2.5 percent at each end of the group.

I am just outside the interval. Am I ill?

Not automatically. Since the interval holds about 95 percent of healthy people, roughly one healthy person in twenty sits outside it. 1177 states that a single blood test usually gives no certain answer and that the result is weighed together with how you feel and with other tests. A value clearly outside the interval is assessed by a doctor.

Why do different laboratories have different intervals for the same test?

Partly the measurement method, partly the group that was measured. Läkartidningen describes how some immunochemical assays' reference intervals, for thyroid hormones for example, apply primarily to the same analytical platform. Region Norrbotten states that its published intervals are based either on its own studies or taken from the literature.

Is a value inside the interval a guarantee that I am healthy?

No. The interval describes what the majority of healthy people look like, not whether you specifically are healthy. 1177 states that a single blood test usually gives no certain answer as to whether you have a disease, and that the result is weighed together with how you feel and with other tests.

What is the difference between a reference interval and a decision limit?

Läkartidningen separates the two. The reference interval is statistics derived from healthy people. The decision limit is a threshold for medical action. By tradition, clinical chemistry has used the reference limits as decision limits, but for some analyses the limit that governs management is derived another way.