Glucose · lot comparison · Chemistry analyzer A

B Methods

B.1 Additional study details

B.2 Statistical methods applied

1. Each specimen’s replicates are averaged on each lot. Difference = new − previous, in the study unit. A positive difference means the new lot reads higher. Percentage difference = 100 × (new − previous) / mean of the two lot results. It needs a ratio scale, results of 0 or more and a mean above the near-zero floor.
2. A specimen measured on one lot only has no difference but counts in the acceptance total.
3. The mean shift has a 95% Student t CI.
4. The lot passes when at least 95% of patient specimens (rounded up to a whole specimen, so every one of fewer than 20) meet their allowable difference, there are at least 2 specimens and the lot plan is complete. A percentage limit is ± percentage × previous-lot result / 100. It needs a ratio scale and a positive previous-lot result. A combined absolute-and-percentage limit uses the larger allowance, or both must be met, as the limit says. Concentration-dependent limits take each specimen’s segment from its previous-lot result.
5. An optional mean-difference limit judges the mean of the matched differences, in the study unit or as the mean percentage difference, against the entered bounds on the exact entered values; it needs at least 2 matched pairs and fails the lot when the mean is beyond it.
6. A specimen allowance below 1.96 × the SD of a matched difference from imprecision alone (the pooled within-specimen SD of a level’s replicated specimens on both lots, times √(1/k₁ + 1/k₂) for the replicates each pair averages) is cautioned with the share of specimens imprecision alone would put outside it. The caution changes no decision.
7. The lot plan sets the pairs needed at each concentration level. Without one, each level in the data needs the calculation minimum of 2 pairs, and rows without a level form one group. A level is complete when enough pairs fall inside its planned range, if one is set. Missing pairs or unplanned levels make the decision inconclusive.
8. Specimen IDs starting with QC are QC material. They are summarized by material and lot and kept out of acceptance. Pass/fail uses the exact entered values.
9. Reportable-range coverage is the width of the overlap between the observed reference-axis range and the claimed reportable range, divided by the claimed range width, times 100. It describes the tested span; accuracy and verification of the reportable range are separate questions. Replicate results are averaged by specimen before determining the observed range.

B.3 Decision rules and display precision

1. Each acceptance criterion compares the unrounded estimate with its limit. Limits include their end values unless marked exclusive. A criterion with a missing estimate, too little data or a failed fit is inconclusive and does not count as met.
2. Entered results, claims, targets and acceptance limits are shown as entered. Calculated concentrations, SDs and fitted parameters are shown to 1 decimal place. Very small or large values use exponent notation at the same precision. An SD or limit that would round to zero keeps two significant digits.