Bias at clinical decision limits
Read bias at each limit off the comparison line and count the patients who cross it, net of repeat testing.
Guides
Guides
Practical references on method comparison and result comparability for laboratories, diagnostics makers and trial teams. Each one cites its sources and works through public data.
How many results cross a limit by chance?
Turn imprecision and distance from the limit into the expected count any change has to beat.
Reagent and calibrator lot changes
Verify a new lot with patient samples at the decision limits, and keep a record across lots.
Comparability across analysers and sites
What ISO 15189:2022 and CLSI EP31 ask, and how to report comparability where clinicians decide.
Point-of-care HbA1c against the laboratory
Bias and category changes at the diagnostic and treatment-goal lines, on public device comparisons.
Jaffe or enzymatic creatinine
Why the chemistries differ, what IDMS traceability fixes, and what a method change does at the lines.
Measurement change in trials and medical AI
Size how a change of laboratory, device or site moves people across a protocol threshold or a model input.
Changing analyser: a checklist
The steps from acceptance limits to go-live, in order, including the ones that cause trouble after verification.
Choosing a method comparison regression
What least squares, Deming and Passing-Bablok assume, the checks that matter more, and reading the line at a limit.
How many samples for a method comparison
What CLSI EP15 and EP09 each ask for, how many patient samples the line needs, and how many pairs a patient count needs.
Reference change value
Whether two results from one patient really differ: the formula, choosing Z, log-normal limits and a method change in between.
eGFR after a creatinine method change
How a creatinine bias passes through the eGFR equation, and why the category lines sit where it is steepest.
CLIA verification for a new analyser
What § 493.1253 asks before you report patient results, what it leaves to you, and one check the usual file leaves out.
CLIA acceptance limits as allowable bias
The current proficiency testing criteria read at decision limits, against public method comparisons.
Does a LOINC code tell you the method?
For routine chemistry, usually not. What published codes show, what the shared creatinine code hides, and what to record beside it.