Plan a bridge when measurement devices change

Define what must remain comparable

A device change can alter calibration, precision, detection limits or the meaning of the recorded value. Start by describing the old and new measurement procedures, including software, specimen handling and units. Decide what agreement is necessary for the intended research use before inspecting the comparison results.

The NIST calibration handbook explains calibration relative to reference standards or a designated measurement process. Calibration evidence and a device-to-device bridge answer related but different questions: two devices can agree with each other while sharing an error relative to the intended reference.

Design the comparison deliberately

Where appropriate, measure the same suitable specimens or subjects using both procedures under controlled conditions. Cover the range relevant to the study, not only convenient middle values. Record order, timing, operators, lots and conditions that could affect comparability. Randomise or balance measurement order when order effects are plausible.

Specify how repeat measurements will be used to assess measurement variation. A bridge based on one reading per device cannot reveal every source of repeatability or reproducibility error. Preserve exclusions and failed measurements rather than removing inconvenient pairs without explanation.

Hypothetical example

Ten fictional paired measurements from a new device are each two units higher than the old device. The correlation can still be perfect when the old readings vary, because a constant offset preserves their linear association. That does not establish interchangeability.

The team compares the two-unit difference with a prespecified acceptable discrepancy and examines whether the offset remains constant across the relevant range. The invented example is a diagnostic illustration, not evidence that subtracting two is a valid correction for any real instrument.

Decide whether pooling is defensible

Inspect paired differences, trends with measurement magnitude and uncertainty. If a conversion is proposed, justify its form and validate it on appropriate evidence. Retain original values and document transformed values, units and the version of the conversion rule.

Report the bridge design, observed limitations and any periods that cannot be combined reliably. Use the collection device-change worksheet and unit-conversion provenance worksheet to record the decision. If comparability remains unresolved, analyse measurement periods separately or explain the limitations rather than silently treating the devices as equivalent.

Source verification: 6 November 2026. Examples are hypothetical.