A shared record of what analyser changes do to patient results.

DesignedDesigned and specified. It has not started and holds no rows from any laboratory.

For each change a laboratory makes, the Registry keeps two things: the paired results the laboratory already produced for its own verification (the same sample measured on the old and the new system), and a short change record: old and new analyser, reagent and calibrator lot where known, software version, the month the new system went live, and the laboratory’s own decision limits.

From many such records it answers two questions no single laboratory can:

• Before a change: what did this same pair of systems do to results at the decision limits in other laboratories?
• After a change: how does our bias, and our count of results crossing each limit, compare with every laboratory that made the same or a similar change?

Holders of data from devices used in trials can contribute device changes in the same way, as a change record and aggregates computed on their own systems. Per-person data never leave the holder.

Contributing is optional and separate from the report. The analyser-change report runs inside your laboratory and only aggregate files leave it (see security and data handling). Contributing to the Registry is a further step a laboratory can choose, by opting in on its order form or by signing a one-page accession form. It is never a condition of the report.

A contributing laboratory chooses one of two routes for each change:

• Coded pairs. Your own staff remove every identifier and replace each sample number with a random code of which no key is kept. Each row then holds only the analyte, its unit, the old and the new value, the sample type and, if you choose, the care setting.
• Aggregates only. The totals the report’s program already produces in your laboratory, with small counts suppressed, and the change record.

Never contributed: names, identity numbers, dates of birth, age, sex, the date or time of a draw, ward, requester, diagnosis, free text, or routine results outside the comparison.

Your data protection officer decides. If any row or field would be personal data in our hands, it is not contributed, and we agree a coarser version, such as aggregates only.

• A peer benchmark, refreshed at least yearly. Its own bias and crossings at each of its decision limits, set against the pooled results for the same change. A peer cell is shown only when enough laboratories share it that none can be singled out.
• A forecast before its next change. On request, what other laboratories saw for the pair of systems it is considering, for a tender or an installation plan.
• A credit on its own orders, stated on the order form. No laboratory is paid to contribute, and none pays to contribute.
• Anonymity. Laboratories appear in outputs only as a code and a country, and are never named without their written consent.
• Co-authorship offered to named staff on any scientific publication that uses their rows, under the journal’s authorship rules, with time to review each manuscript first.
• The right to stop. A laboratory can stop contributing at any time, and its rows are then removed from the live tables and from any benchmark, publication or model work that starts afterwards.

OfferAn offer to the first laboratories to join. No founding laboratory has joined yet.

The first laboratories to join become founding laboratories. Each holds a seat on the Registry’s steering group, which meets at least twice a year and decides, by majority, the minimum cell sizes (which it can raise but never lower), what a registry row contains, the publication plan, and requests from academic researchers. Founding laboratories are named as such only if they agree to it.

Ask for the founding terms

The Registry is held by CareerLoop Inc., the company that operates NakedSignal, and NakedSignal chairs the steering group. We use the pooled record to build and improve methods, software and models that size analyser changes, and we may license those to others. We never sell or license registry rows as rows, never try to identify a person, and never combine rows with other data for that purpose.

Rows contributed from the EU or EEA are stored and processed there; rows from the UK are stored in the UK or the EU or EEA. Every row is encrypted at rest, and access is limited to named staff and logged. The full charter is sent on request.