Where the Line Falls: A Risk-Based Review of Antibody Assay Thresholds in UK Diagnostics
Photo: Unknown authorUnknown author, Public domain, via Wikimedia Commons
A cut-off can look reassuringly simple in a standard operating procedure: one value, one comparator, one direction of interpretation. In practice, that number is the visible edge of a chain of decisions about intended use, population, specimen handling and clinical consequence. For UK antibody laboratories and in vitro diagnostic developers, the question for 2026 is not merely whether a threshold is statistically derived. It is whether the organisation can explain why that operating point remains appropriate for the use to which it is put.
The threshold is an operating point, not a property of the assay
Receiver operating characteristic analysis is useful because it displays the trade-off between sensitivity and specificity across possible thresholds. It does not, by itself, select a cut-off. A laboratory that adopts the Youden index, the closest point to the upper-left corner, or a manufacturer’s stated value has made a choice, even if that choice was made elsewhere. The choice should reflect the cost of a false positive relative to a false negative in the intended pathway.
That distinction matters in the UK, where antibody tests may be used in symptomatic triage, occupational health, transfusion support, infection surveillance or private wellness testing. A threshold suitable for confirming exposure in a high-prevalence clinical group is not automatically suitable for screening a low-prevalence population. The assay may be the same; the decision context is not.
Prevalence changes the meaning of the same result
Sensitivity and specificity are often treated as stable performance characteristics, but positive and negative predictive values depend on prevalence. A test with 95% sensitivity and 97% specificity will produce a very different proportion of false positives when prevalence is 2% rather than 20%. This is not a statistical curiosity. It affects patient communication, confirmatory testing capacity and the credibility of a diagnostic pathway.
UK laboratories should therefore document the prevalence assumptions behind a cut-off. Where local prevalence is uncertain, a sensitivity analysis can show how predictive values change across plausible ranges. For developers preparing UKCA documentation, the same discipline applies: the clinical performance section should make the intended population and prevalence context legible, rather than presenting a single table of sensitivity and specificity without a use case.
Matrix and sample handling can move the line
Antibody assays are sensitive to specimen type and pre-analytical history. Serum and plasma may not be interchangeable without evidence. Haemolysis, lipaemia, delayed separation, freeze-thaw history and anticoagulant choice can all influence signal. A cut-off derived in one matrix should not be transferred silently to another. Where a laboratory introduces a new sample type, a matrix-bridging study should examine performance around the threshold, not only at extreme high and low concentrations.
The same principle applies to sample quality in routine service. A cut-off validated on freshly separated serum may behave differently on samples posted to a reference laboratory or collected in a community setting. The review should ask whether the specimen journey used in validation resembles the specimen journey used in practice.
Build a cut-off dossier that can be inspected
A defensible threshold needs a decision record. For each antibody assay, the dossier should identify the intended use, the target population, the reference standard used to define true status, the derivation dataset, the statistical method and the uncertainty around the estimate. It should state the chosen operating point, the expected sensitivity and specificity, the predicted predictive values at stated prevalence, and the handling of equivocal or grey-zone results.
This is not paperwork for its own sake. ISO 15189 requires laboratories to verify methods for their intended use, and UKCA requirements for IVD devices place emphasis on clinical evidence and performance evaluation. A concise dossier makes those obligations easier to meet. It also supports change control: if the assay, sample matrix, population or clinical pathway changes, the organisation can see which assumptions have been disturbed.
A 2026 review framework for existing thresholds
A practical review can begin with an inventory. List every antibody assay with a cut-off, including laboratory-developed tests and commercial kits used within the laboratory’s scope. For each, record the intended use, current threshold, derivation source, last review date and owner. Risk-triage the list: assays used for high-consequence clinical decisions, assays with known interference issues and assays whose intended population has changed should come first.
Next, reconstruct the evidence. Confirm that the original derivation population resembles the current population. Recalculate predictive values at local or expected prevalence. Check whether the sample matrix and pre-analytical conditions in validation match routine practice. Examine results near the threshold, not only far from it. Where data are available, review the distribution of equivocal results and the downstream actions they trigger.
Make the grey zone explicit
Many antibody assays produce results that are neither convincingly positive nor convincingly negative. A narrow grey zone may be appropriate when confirmatory testing is readily available; a wider equivocal range may be safer when results feed into a pathway with limited follow-up. The decision should be stated, with a rationale and a described action for clinicians or service users. Silence around the grey zone does not remove the uncertainty; it merely moves it into interpretation.
Change the number only with evidence and communication
A threshold should not be altered simply because a new dataset produces a slightly different optimal point. Change should follow a documented review, with an assessment of clinical impact, verification data and a communication plan for users. Manufacturers must consider regulatory change control; laboratories must consider local validation and the effect on historical comparability. Version control and audit trails are part of the decision, not an afterthought.
Equally, retaining a number is not neutral. Keeping a cut-off because it has always been used is a decision to accept its original assumptions. The review should record that choice explicitly, together with the date and owner of the next scheduled review.
The cut-off is where statistical evidence meets clinical consequence. In UK antibody diagnostics, that line should be defensible, documented and revisited when the context changes. It should not be merely inherited.
