Study reference
Every study type has a reference page, linked from the table below, with its settings, columns, acceptance limits, results, an example and its calculations. Each study type also has a Method ID such as LMS-CMP-1.26, which appears in the report's C Audit trail (with Include technical details) and in the Calculation file (JSON). Whenever a change to a method could alter a value or a decision you see, the method gets a new ID, and earlier calculations and reports keep the ID they were calculated with. See About records and history. The checks behind each method are in Calculation checks.
Among them is a separate recalculation of every study type, published by Berberian Labs LLC and described under Calculation checks. With the evaluation package, you can compare the application with the same expected values in your own workspace. Download the evaluation package (ZIP, 11.4 MB, source a5759c64c412; ZIP SHA-256 4b0d288e0f5f4dceb133629a912590de0da79e3c68411ca88dcae07a7a9538e6). Inside is one study of each type with its input data, settings, limits, expected values and rendered report, plus a manifest and instructions for the evaluator.
Methods by study type
Section titled “Methods by study type”| Study type | Method ID | Main estimators | Main sources | Supported design | Scope |
|---|---|---|---|---|---|
| Descriptive summary · example · checks | LMS-DESC-1.14 |
Mean, sample SD (n − 1), CV, type-7 quartiles, Student t CI for the mean | NIST/SEMATECH e-Handbook §1.3.5.6 and §1.3.5.2 | Any number of results per level: one result gives a summary, 2 give an SD, and a criterion needs 2 specimen IDs. Levels keep unlike populations apart. | Summarizes a set of results as mean, median, SD, CV and quartiles, checked against any limits you set. |
| Repeatability · example · checks | LMS-RPT-1.15 |
As descriptive, for one material, day and run per level | NIST/SEMATECH e-Handbook §1.3.5.6 and §1.3.5.2 | One material, one day and one run per level, with all replicates in that run; 2 replicates decide a criterion, and 20 is common. | Checks whether the within-run SD or CV meets your limits and is consistent with the manufacturer's repeatability claim. |
| Multi-day precision · example · checks | LMS-PREC-1.17 |
Nested variance components for days, runs and replicates, balanced or unbalanced. Repeatability and within-laboratory SD and CV | NIST/SEMATECH nested designs; Chakravarthy et al. 2019 example | Days × runs × replicates per level, balanced or unbalanced; at least 2 days and one run with 2 replicates per level, one material per level. | Checks whether repeatability and within-laboratory imprecision meet your limits and are consistent with the manufacturer's claims. |
| Bias against an assigned value · example · checks | LMS-ACC-1.19 |
Mean minus assigned value, percent bias, recovery, Student t bias CI, combined uncertainty | NIST/SEMATECH e-Handbook §1.3.5.2 | Repeated results of a material with an assigned value, one or more levels; 2 results per level decide a criterion, replicates in one run give a CI. | Checks whether the mean bias from the assigned value, as an amount, a percentage or a recovery, is within your limit. |
| Method comparison · example · checks | LMS-CMP-1.26 |
Specimen differences against the allowable difference, with at least 95% of at least 5 specimens within. Ordinary least squares, Deming, weighted Deming and Passing-Bablok on specimen means. Mean bias and limits of agreement. Bias at decision levels | Linnet 1990 and 1993; Passing and Bablok 1983; Bilić-Zulle 2011; Giavarina 2015 | Paired specimens on a test and a comparative method, single or replicate measurements; 2 specimens to calculate, 3 for regression intervals, 5 for the specimen rule. | Checks the paired differences, the regression bias at your decision levels and the mean bias against your allowances, using the regression model you choose. |
| Instrument comparison · example · checks | LMS-ICOMP-1.14 |
Specimen differences from the reference analyzer against the allowable difference, with at least 95% of at least 5 specimens within. Mean differences and limits of agreement on cell means | NIST/SEMATECH e-Handbook §1.3.5.6 and §1.3.5.2; Giavarina 2015 | The same specimens on a reference analyzer and one or more other analyzers, single or repeat readings; 2 specimens to calculate, 5 for each analyzer's specimen rule. | Checks, analyzer by analyzer, whether the specimen differences from the reference analyzer are within your allowance. |
| Dilution verification · example · checks | LMS-AMI-1.10 |
Dilution factors, corrected results and recovery. Verified range from accepted recoveries | Dilution arithmetic | High pools diluted by stated or volumetric factors, one or more dilutions each; 2 recoveries per level decide criteria. | Checks whether corrected results recover their expected concentration within your limit, and which dilution factors are verified. |
| Reagent lot comparison · example · checks | LMS-LOT-1.20 |
New-minus-previous difference for each patient specimen against an absolute allowance, a percentage of the previous-lot result, or both. At least 95% of specimens, at least 2, must be within and the plan complete; an optional mean-difference limit must also be met. Supporting mean differences, Student t intervals, imprecision and a caution when the allowance is below the spread of imprecision alone | NIST/SEMATECH e-Handbook §1.3.5.2 | Patient specimens measured on the lot in use and the new lot, in one or more concentration groups, with optional QC rows; 2 paired specimens to calculate. | Checks the specimen rule, the lot plan and any mean-difference limit you enter against your allowances. |
| Linearity · example · checks | LMS-LIN-1.22 |
Least-squares line through level means. Deviation from the line and recovery at each level. Optional reportable-range check: a level within the proximity allowance of each limit, or a blank or minimal level at or below the lower limit. Optional Calculate theoretical values: expected values in equal steps from two levels held true | NIST/SEMATECH e-Handbook §1.3.5.2 and §1.3.5.6; Xiao and Chambliss 2023 (endpoint proximity) | Levels with expected concentrations, 2 or more replicates each; 3 levels fit a line, with optional equal-spacing theoretical values and reportable-range endpoints. | Checks each level's deviation from the fitted line, its accuracy against the assigned value and the range coverage against your limits. |
| Reference interval verification · example · checks | LMS-RIV-1.15 |
Counts of subjects below, within and above the claimed interval, compared with your staged rule; each stage counts all subjects so far or only its own new subjects, and the result gives every stage's outcome | Counts against your decision rule; the filled-in 2-of-20 criterion follows CLSI EP28-A3c (2010) §11.2 | Eligible subjects per partition, one result each, repeat draws averaged; 1 subject gives a result, 2 decide a criterion, with an optional staged rule. | Checks whether the claimed interval holds for your population by counting subjects below, within and above it against your rule. |
| Reference interval establishment · example · checks | LMS-RIE-1.17 |
Nonparametric percentiles (type 6) with exact rank-based CIs, or parametric or log-normal limits with manual or automatic selection (no model selected automatically below 20 subjects) | Horowitz 2008; Bland, An Introduction to Medical Statistics; normal-approximation limits on the log10 scale (log-normal) | Eligible subjects per partition, one result each. At 95% two-sided coverage, 39 subjects give nonparametric limits; endpoint confidence intervals need 119 at 90% confidence or 146 at 95% confidence. Automatic distribution selection starts at 20 subjects; a manual parametric or log-normal choice needs 2. | Estimates reference limits and their confidence intervals by the nonparametric, parametric or log-normal method, chosen by you or automatically. |
| Qualitative agreement · example · checks | LMS-QAL-1.10 |
PPA, NPA and overall agreement with Wilson CIs. Kappa and McNemar as supplements | FDA 2007 guidance on diagnostic test studies; Fleiss, Cohen and Everitt 1969; Newcombe 1998 | Paired categorical results on the same specimens, binary or ordered, with nonreportable categories; 1 pair gives a result, 2 per denominator decide a criterion. | Checks positive, negative and overall agreement against your limits, with kappa and McNemar as supplements. |
| Near-cutoff precision · example · checks | LMS-QCO-1.10 |
Hit rates with Wilson CIs. Probit fit. C5, C50 and C95 with Fieller limits | Fieller 1954 (inverse confidence limits); Wilson score interval | Replicate qualitative results at several concentrations around the cutoff; 3 concentrations with mixed results fit a curve, fewer still show hit rates. | Estimates C5, C50 and C95 with their limits and compares the hit rates with your expectations. |
| Detection limits: LoB, LoD, LoQ · example · checks | LMS-LOB-1.10, LMS-LOD-1.11, LMS-LOQ-1.12 |
Rank-position percentile or parametric LoB. Parametric LoD from LoB and the pooled low-level SD with the small-sample multiplier c_β. Goal-based LoQ | Linnet and Kondratovich 2004; Armbruster and Pry 2008; NIST/SEMATECH §1.3.5.8 and §7.2.4.1 | Blank and low materials with replicates across days; 10 blanks for the percentile LoB, 2 low materials with 2 results each for LoD and LoQ, or 2 results per claim to verify one. | Establishes LoB, LoD and LoQ, or verifies claimed values against your goals. |
| Sample carryover, 21-position sequence · example · checks | LMS-CO-2.5 |
Mean low-after-high minus mean low-after-low in each run, against 3 SD, an absolute limit or a percentage of the LL mean | pSMILE carryover guideline VAL 2015 | One or more uninterrupted 21-position runs in the fixed low and high order; one complete run can be evaluated. | Checks whether low-after-high minus low-after-low stays below the error limit you select (3 SD, an amount or a percentage of the low-after-low mean), using the 21-position sequence of the pSMILE guideline. |
| Interference · example · checks | LMS-INT-1.16 |
Test minus vehicle-matched control, Welch–Satterthwaite intervals | NIST/SEMATECH e-Handbook §1.3.5.3 | Control and test pools at each analyte level and interferent concentration, 2 or more replicates each; 2 independent pools per combination decide a criterion. | Checks whether the interferent shifts results by more than your allowable effect at each analyte level and interferent concentration. |
| Stability · example · checks | LMS-STB-1.14 |
Change from baseline at each condition and time, paired or independent design; each claimed duration decided as supported or not | NIST/SEMATECH e-Handbook §1.3.5.2 and §1.3.5.3 | Baseline and later times under each condition, paired or independent specimens; 2 materials at baseline and at a time give an interval. | Finds how long results stay within your allowable change under each storage condition. |
| Sigma metric · example · checks | LMS-SIG-1.6 |
(TEa% − |bias%|) / CV% at a stated concentration, with separate grades, QC-rule guidance and an optional direct laboratory minimum |
Westgard, Bayat and Westgard 2019; Coskun et al. 2019 (counter-position) | One entry per concentration with allowable total error, bias and imprecision, typed or linked from saved studies; one entry gives a sigma. | Calculates sigma at each concentration from the TEa source you choose, with its grade and the QC rules suggested for that sigma. |
| QC targets (QC export) · example · checks | LMS-QC-1.4 |
Candidate target mean, sample SD (n − 1) and CV by instrument, lot, analyte and level. Optional mean ± 1, 2 and 3 SD ranges and combined performance across chosen instruments. Descriptive | Sample SD and CV definitions | QC export files with instrument, lot, analyte, level, date and result; a group needs at least 2 numeric results for a target. | Proposes a target mean, SD, CV and SD ranges for each instrument, lot, analyte and level. |
| Calibration verification · example · checks | LMS-CAL-1.3 |
Independent assigned values/ranges, mean and replicate error decisions, range coverage, with a zero or minimal material accepted at the lower end | 42 CFR 493.1255; CAP; NIST t and chi-square intervals | Low, middle and high materials with assigned values or target ranges, 1 or more results each; 2 replicates per level give its SD and precision index. | Checks whether each material's mean and each replicate are within your allowance, and whether the accepted materials cover the claimed range. |
| PT and INR · example · checks | LMS-PTINR-1.4 |
Geometric mean normal PT, INR equation, entered-resolution programming and lot comparisons | WHO TRS 979 Annex 6; laboratory acceptance rules | Normal specimens for the geometric mean PT, PT and reported INR pairs for programming, or paired PT results for a lot comparison; 20 normal specimens for the mean. | Checks the mean normal PT, the INR programming and the lot-to-lot PT differences against your rules. |
| Glucose error grids · example · checks | LMS-GLUCOSE-1.4 |
Clarke and Parkes zones, differences and optional zone limits | Clarke 1987; Pfützner 2013 | Paired laboratory and meter glucose results, one or more meters; a single pair can be placed on the grids. | Places each result pair in the Clarke and Parkes zones of the published grids and compares the zone shares with any zone limits you set. |
| Proficiency tracking · example · checks | LMS-PROF-1.3 |
Peer/all-method bias, SDI and local allowance decisions | Eurachem PT guide, third edition | One row per event, material and level with the laboratory result and the peer or all-method group statistics. | Reports bias and SDI against the chosen group and checks each event against your own allowance. |
| Measurement uncertainty · example · checks | LMS-MU-1.2 |
Within-laboratory variation plus assigned-material bias evidence | Nordtest TR 537, fourth edition | Within-laboratory precision plus assigned-value evidence from 2 or more materials or events, or 2 run means of one material. | Estimates combined and expanded uncertainty from your precision and bias evidence. |
Supporting methods:
| Method ID | Used for |
|---|---|
LMS-CRITERIA-1.1 |
Comparing estimates with acceptance criteria |
LMS-DIAG-1.0 |
Normal Q–Q plot and box plot in descriptive, repeatability, multi-day precision and reference interval studies |
LMS-RI-REVIEW-1.0 |
Flags values outside the box-plot whiskers in reference interval studies. Flagged values stay in the result |
Calculation details
Section titled “Calculation details”The Results stage lists the method statements and sources under Methods and sources. So does the report's B Methods appendix when Include technical details is on. Each study's reference page gives the formulas under Calculations and sources.