Reference ranges and abnormal flags: getting lab reports right

A reference range is the interval of results expected in a healthy population of the same age and sex, and an abnormal flag is the H or L a report prints when a result falls outside it. Both depend on the method and the kit the lab uses, so a range copied from a textbook or from another lab is not automatically correct. The flag is a prompt for a pathologist, not a diagnosis.

Where do reference ranges come from?

For most routine tests the first source is the kit insert. In India, in vitro diagnostic kits are regulated under the Medical Devices Rules, 2017, administered by CDSCO. The manufacturer's insert states the range the kit was validated against, for a stated population and method. That is where the lab should start, not with a value remembered from college.

The second step is verification. ISO 15189 expects a lab to review its reference intervals and confirm they suit its own patient population, and NABL's specific criteria, NABL 112A, ask labs to use IVD kits where available and to justify their choice of method. In practice this means checking the insert range against results from healthy adults in your own population and documenting what you found. When a lab changes its analyser or kit, the range has to be checked again.

Why do age, sex and method matter so much?

The same number can be normal for one patient and abnormal for another. A few examples every pathologist knows:

• Haemoglobin: the expected range for adult men is higher than for adult women, and children and pregnant women have their own bands.
• Alkaline phosphatase: growing children have much higher values than adults because of bone growth. An adult range applied to a ten-year-old flags a healthy child as abnormal.
• Creatinine and urea: differ by sex and muscle mass, and with age in the elderly.
• Thyroid hormones and hCG: vary through pregnancy, so a single adult range misleads.
• Method: a bilirubin measured by one chemistry and one by another can carry different ranges from their manufacturers. The range belongs to the method, not to the test name.

The lesson for software is that a test does not have one range. It has a set of ranges keyed by sex, age band, sometimes pregnancy or trimester, and the method or analyser that produced the value.

How should the lab software store ranges?

A test master that holds a single low and high per test cannot do this. The structure below is what Pathixio uses and what we suggest checking in any product.

1. One test, many ranges. Each range row has a sex, an age from and to, an optional condition such as pregnancy, and the method or analyser it applies to.
2. Units on the range, not only on the test. A lab that reports glucose in mg/dL must not import a range in mmol/L by mistake.
3. Age calculated at the time of sampling from the date of birth, in days for neonates and months for infants, not in whole years.
4. Critical limits stored separately from the reference range. NABL asks labs to establish and display critical limits for tests that need immediate attention and to document their communication.
5. Effective dates. When a range changes, old reports keep the range they were issued with, and the change is in the audit log with the person who made it.
6. Free-text interpretation for tests where a number alone is not enough, such as urine microscopy or a peripheral smear.

The pre-loaded Indian test catalogue in Pathixio ships with age- and sex-specific ranges as a starting point. They are meant to be verified and edited by the lab, not used blindly.

Why do H and L flags need a pathologist?

The flag is arithmetic: the value fell outside the interval. What it means is not arithmetic. A mildly low haemoglobin in a menstruating woman, a raised CK after a gym session, a high glucose in a sample that sat unseparated for three hours: each is an H or L on the printout and a different story in the clinic. This is why the report is not released when the analyser finishes. It goes into the pathologist's queue with the flags, the patient's previous results and any delta that has moved sharply.

NABL 112A asks that reports carry the name or signature of the person who authorised release. It permits automated release of results under defined rules, but requires such reports to be marked as auto verified and the rules to be traceable to an authorised person. A small lab is better off with a human sign-off on every report and software that makes that sign-off fast: flagged results at the top, one click to view history, a digital signature applied on authorisation.

What goes wrong in practice

• Ranges copied from a previous vendor's database with the method column blank, so a change of analyser silently reports against the old range.
• A single adult range used for all ages because setting up paediatric bands felt like too much work on go-live day.
• Units mismatched between the analyser interface and the test master, so a correct value is flagged wrong.
• Reports edited after authorisation without a record, so the printed range and the stored range disagree.

All four are software configuration problems, and all four show up in the first week if someone checks a paediatric report and a report from each analyser against the kit inserts. That hour of checking is the best investment a new lab makes.

Frequently asked questions

Can the software decide reference ranges for us?

No. Software can store ranges by age, sex and method and apply them consistently, and a pre-loaded catalogue saves typing. The responsibility for choosing and verifying the range stays with the lab and its pathologist, using the kit insert and the lab's own verification data. Treat any pre-loaded range as a draft to be checked.

What is a critical value and how is it different from a flag?

A flag marks a result outside the reference interval. A critical value is a result so far outside that it may need immediate action, such as a very high potassium. NABL expects labs to define critical limits, communicate such results to the treating doctor promptly, and document the communication. The software should alert on them separately from ordinary flags.

How often should reference ranges be reviewed?

Whenever the method, kit or analyser changes, and otherwise at a fixed interval the lab sets in its quality manual. Many labs tie the review to their annual internal audit. The software should show when each range was last edited and by whom, so the review is a report rather than a search through old files.