Guide
Changing analyser: a checklist from plan to go-live
Replacing an analyser or a method is routine work with a long tail. Most of the effort goes into verification; most of the surprises come later, from reference intervals, calculated results, patients on serial monitoring and clinicians who were not told. This checklist puts the steps in order, with the reason for each and where to read more.
The checklist at a glance
- Scope the change
- Fix acceptance limits before any data
- Verify the new system on its own
- Compare on patient samples
- Judge the change at the decision limits
- Transfer or verify reference intervals
- Recompute derived results
- Plan for patients being monitored
- Tell clinicians before go-live
- Watch the first weeks
The order matters. Steps one and two happen before a single sample is run, so the decision at step five is made against limits nobody could tune after seeing the data. Steps six to ten are where an otherwise clean verification still causes trouble.
1. Scope the change
List every reported result the change touches, not just the measured analytes: calculated results such as eGFR, ratios and gaps; reference intervals and flags; units; the decision limits each clinical service uses; and any rule in the laboratory information system that fires on a value. Note which results are followed over time in the same patients. This list defines what the rest of the plan has to cover.
For a first idea of direction and size before your own data exist, the free analyser comparability lookup shows what published open-access comparisons found for several common analytes. It is a planning aid, not a substitute for your own comparison.
2. Fix acceptance limits before any data
For each analyte, decide how large a difference is acceptable, and write it down before any data are collected. The European consensus on analytical performance specifications sets out three ways to derive them: from the effect of analytical performance on clinical outcomes, from biological variation, and from the state of the art [Sandberg et al., 2015]. Pick one per analyte and record why.
Do the same for the risks the change brings. A working group of the European federation recommends using risk evaluation to decide the extent of verification at each stage of introducing a new method, so that effort goes where the risk to patients is [Roelofsen-de Beer et al., 2020].
3. Verify the new system on its own
Before comparing old with new, check that the new system meets its own claims in your hands. The CLSI protocol for user verification of precision and estimation of bias can be completed in as few as five days [CLSI EP15]. Check the reportable range at both ends, carry-over where high and low samples run back to back, and the interferences that matter for your population, such as haemolysis, icterus and lipaemia in the samples you actually receive.
4. Compare on patient samples
Run the same patient samples on both systems and estimate the bias between them [CLSI EP09]. Use fresh patient samples rather than control material, spread across the whole reporting range and concentrated around each decision limit. The width of the range matters more than any other design factor, and conventional sample numbers are often too small for narrow-range analytes [Linnet, 1999].
Choose the regression for the data you have, check linearity, scatter and outliers first, and keep the exclusion rules you fixed in advance. The guide to method comparison regression covers the choice between least squares, Deming and Passing-Bablok.
5. Judge the change at the decision limits
Read the bias off the comparison line at each decision limit, with an interval, and set it against the acceptance limit fixed at step two. Then count the patients: how many results land in a different category at each limit, in each direction, and how that compares with the number repeat testing alone would move. The guide to bias at clinical decision limits and the guide to crossings by chance set out both calculations.
The change check runs them on your paired results in the browser: Passing-Bablok line, bias at each decision line you set, and crossings at each line against repeat testing. The file never leaves your computer.
6. Transfer or verify reference intervals
A new method can need new reference intervals even when it agrees well with the old one at the decision limits. The CLSI guideline covers defining, establishing and verifying them [CLSI EP28]. Its standard verification uses a minimum of 20 samples from healthy people in the local population; mining large sets of routine patient results is an alternative, and children and older adults remain harder to cover [Ozarda et al., 2018]. Decide which intervals will change, and from what date, before go-live.
7. Recompute derived results
A bias in a measured analyte passes into every result calculated from it, sometimes amplified. Recompute each derived result from the comparison pairs and check its categories as you did for the measured ones. For creatinine, the guide to eGFR after a creatinine bias shows how a small bias moves eGFR categories, and the free eGFR calculator shows the effect for one patient with either equation.
8. Plan for patients being monitored
Patients whose results are followed over time are the ones a method change can mislead most: the first result on the new system is compared with the last on the old. Identify the tests and services where that happens, state the expected shift at the concentrations they work at, and decide how to handle the crossover: a comment on reports for a defined period, a fresh baseline on the new system, or a repeat on the new system before a change is acted on.
Whether a change between two results is more than imprecision and biological variation explain is what the reference change value tests, and a known method bias can be taken out first. The guide to the reference change value explains how, and the free RCV calculator does it for two results.
9. Tell clinicians before go-live
Tell the people who use the results before the change, not after the first query. Say which tests change, from what date, by how much at the limits they use, which reference intervals change, and what the laboratory is doing about serial results. Keep it short and give numbers at their decision limits rather than a regression slope.
10. Watch the first weeks
Set quality-control targets and limits from data on the new system rather than carrying the old ones over, and move to the right peer group in external quality assessment. In the first weeks, watch patient results as well as controls: patient-based real-time quality control, including moving averages of patient results, uses the laboratory’s own patient data to monitor assay performance [Duan et al., 2024]. A shift in the median of routine results after go-live that the comparison did not predict is worth a look.
Keep the plan, the acceptance limits, the comparison, the decisions and the communications in one record. It is the evidence of verification, and the baseline for the next change.
Sources
- Sandberg S, Fraser CG, Horvath AR, et al. (2015). Defining analytical performance specifications. Clinical Chemistry and Laboratory Medicine. doi.org/10.1515/cclm-2015-0067
- Roelofsen-de Beer R, Wielders J, Boursier G, et al. (2020). Validation and verification of examination procedures in medical laboratories: opinion of the EFLM Working Group Accreditation and ISO/CEN standards (WG-A/ISO) on dealing with ISO 15189:2012 demands for method verification and validation. Clinical Chemistry and Laboratory Medicine. doi.org/10.1515/cclm-2019-1053
- Clinical and Laboratory Standards Institute (2014). EP15: User Verification of Precision and Estimation of Bias. CLSI guideline, third edition. clsi.org/shop/standards/ep15/
- Clinical and Laboratory Standards Institute (2018). EP09: Measurement Procedure Comparison and Bias Estimation Using Patient Samples. CLSI guideline, third edition. clsi.org/standards/products/method-evaluation/documents/ep09/
- Linnet K (1999). Necessary sample size for method comparison studies based on regression analysis. Clinical Chemistry. doi.org/10.1093/clinchem/45.6.882
- Clinical and Laboratory Standards Institute (2010). EP28: Defining, Establishing, and Verifying Reference Intervals in the Clinical Laboratory. CLSI guideline, third edition. clsi.org/shop/standards/ep28/
- Ozarda Y, Higgins V, Adeli K (2018). Verification of reference intervals in routine clinical laboratories: practical challenges and recommendations. Clinical Chemistry and Laboratory Medicine. doi.org/10.1515/cclm-2018-0059
- Duan X, Zhang M, Liu Y, et al. (2024). Next-generation patient-based real-time quality control models. Annals of Laboratory Medicine. doi.org/10.3343/alm.2024.0053