Why 'Normal' Travels Poorly: Local Reference Intervals Before the Next UK Analyzer Replacement

Every analyzer replacement brings a familiar project plan: validation, training, connectivity testing and a cut-over date. Reference intervals are often treated as a configuration item, copied from the vendor's package insert and loaded with the assay settings. That is convenient, but it assumes that 'normal' is a property of the method rather than a property of the population being tested. In UK laboratories serving diverse and ageing populations, that assumption deserves more scrutiny.

The default that quietly becomes a clinical decision

A reference interval is not a neutral label. It determines whether a result is flagged, repeated, referred or ignored. If the interval is too wide, clinically important abnormalities may be missed; if it is too narrow, patients may be investigated unnecessarily. Vendor intervals are usually derived from a manufacturer's reference population, which may differ from the local population in age structure, ethnicity, body mass index, renal function, medication use and clinical case mix.

That does not make vendor intervals wrong. It makes them a starting point. ISO 15189 requires laboratories to establish and periodically review reference intervals appropriate to the population served. UKAS accreditation assesses whether that requirement is met. An analyzer upgrade is a natural moment to ask whether the intervals still fit.

Where UK populations diverge from the vendor sheet

Several analytes are particularly sensitive to population differences. Creatinine and estimated glomerular filtration rate are affected by muscle mass and ethnicity, and the latter has been the subject of important methodological debate. Liver enzymes, haemoglobin, ferritin, vitamin D, thyroid-stimulating hormone and prostate-specific antigen can vary with age, sex, ethnicity, diet and clinical history. Paediatric intervals are a separate challenge because children are not small adults and because local paediatric case mix can differ markedly between hospitals.

There is also a service-level issue. UK pathology networks and integrated care systems increasingly share assays and pathways across sites. If two laboratories use different intervals for the same analyte, clinicians may see inconsistent flags as patients move between settings. Harmonisation is desirable, but it should be harmonisation to a verified local or network standard, not blind copying of a vendor default.

A prioritisation method for busy laboratories

Not every analyte needs a de novo study. A risk-based approach allows laboratories to focus resources where the clinical and financial consequences of misclassification are greatest.

Step 1: Build an analyte risk register

List the analytes affected by the upgrade. For each, record the clinical decision limits, the current interval source, the population served, the availability of local data and the potential harm from false-positive or false-negative classification. Prioritise analytes used in urgent pathways, paediatric care, renal medicine, oncology and endocrine diagnosis.

Step 2: Test transference before commissioning

Before accepting a vendor interval, review the manufacturer's reference population and the method comparison data. Where the method is sufficiently similar and the population is reasonably comparable, a transference study may be appropriate. CLSI EP28-A3c provides a recognised framework, including a minimum of 20 reference individuals for transference validation. If the results fall outside acceptable limits, a larger local study is needed.

Step 3: Commission de novo intervals only where the risk justifies it

A de novo study typically requires at least 120 reference individuals per partition, such as sex or age group. That is a substantial commitment. It should be reserved for analytes where the local population is clearly different, where clinical decision-making is sensitive to small changes, or where existing intervals are known to be contested. Retrospective data can help with planning, but it should not replace a properly constituted reference population.

Step 4: Write the finance case in clinical terms

Finance teams are used to business cases built on throughput and cost per test. Reference interval work is better justified by avoided harm and avoided downstream cost. A misplaced interval can generate repeat tests, unnecessary referrals, delayed diagnoses and clinician distrust. It can also create an accreditation finding. The case should quantify, where possible, the number of results near decision limits, the likely change in flag rates and the clinical pathways affected. Even a modest reduction in repeat testing can offset the cost of a targeted study.

Accreditation and the paper trail

The evidence should be documented before the analyzer goes live. That includes the source of each interval, the verification or study performed, the statistical method, the reference population, the date of review and the approval route. Laboratories should also record how intervals will be reviewed after implementation, particularly if the population or service changes.

Configuration is a clinical act

Loading a reference interval is not an IT task. It is a clinical decision about who counts as normal and who needs further investigation. UK laboratories cannot eliminate every population difference, but they can avoid treating a vendor default as a universal constant. Before the next analyzer upgrade, the most useful question may be simple: whose normal is this, and does it match the patients we serve?