Calibration Verification/Linearity

Program

Users Guide

cap.org
Table of Contents
How to Make the Most of Your Users Guide............................................................................................... 2

CVL Evaluations and Reports...................................................................................................................... 3

• Understanding Your CVL Surveys Report Package........................................................................ 3
• Peer Group Summary........................................................................................................................ 3
• Peer Performance Bar Charts........................................................................................................... 3
• Executive Summary............................................................................................................................ 3
• Calibration Verification Evaluation................................................................................................... 4
• Standard Linearity Evaluation.......................................................................................................... .6
• Diluted/Extended Linearity Evaluation............................................................................................ 8
• Linearity Troubleshooting Report.................................................................................................... 11

Summary of Regulatory Requirements .................................................................................................... 12

Calibration Verification Troubleshooting Guide...................................................................................... 13

Linearity Troubleshooting Guide................................................................................................................ 15

Glossary of Terms......................................................................................................................................... 19

Frequently Asked Questions....................................................................................................................... 21

Appendix 1: Worksheet for Calibration Verification Self-Evaluation..................................................... 25

Appendix 2: CVL Investigation Checklist for Problematic Results......................................................... 26

Appendix 3: Flow Charts............................................................................................................................ 28

• Selection of Linearity Evaluation Type........................................................................................... 28
• Standard Linearity Evaluation......................................................................................................... 29
• Diluted/Extended Linearity Evaluation When Results Include
Extended Range (ER) Specimens................................................................................................... 30
• Diluted/Extended Linearity Evaluation When Results Include Diluted Specimens................... 31

Appendix 4: Technical Guide to the CAP Linearity Evaluation............................................................. 32

References................................................................................................................................................... 35

© 2014 College of American Pathologists. All rights reserved. The College does not permit reproduction of any substantial
portion of the material in this report without its written authorization. The College hereby authorizes participants in the
program to use the material in this report solely for educational purposes within their own institutions. The College prohibits
use of the material in the report—and any unauthorized use of the College’s name or logo—in connection with promotional
efforts by marketers of laboratory equipment, reagents, materials, or services.
 How to Make the Most of Your Users Guide

How to Make the Most of Your Users Guide

The CVL Surveys Users Guide includes several easy-to-use sections.

• The CVL Report Package (pages 3–11)—Sample report pages and a summary of the
calibration verification and linearity evaluations. This section of the Users Guide highlights key
concepts and terminology.

• Regulatory Requirements (page 12)—Information about regulatory requirements pertaining to

the CVL Surveys.

• Calibration Verification and Linearity Troubleshooting Guides (pages 13–18)—A

collection of examples are provided if you have any problems with your evaluations.
You can identify the example report that is most similar to your results and that
classifies your issue and suggests what actions should be taken.

• Reference Information: Glossary and FAQ (pages 19–24)—A complete glossary of

calibration verification and linearity terms and a listing of frequently asked questions.

• Additional Information: Appendices (pages 25–34)— More detailed explanations of the

evaluation protocols.

n Appendix 1 provides a self-evaluation worksheet for calibration verification.

n Appendix 2 contains a checklist for investigating common calibration verification and
linearity problems.
n Appendix 3 features detailed flow charts outlining the logic and procedural steps in the

linearity evaluations.
n Appendix 4 offers a more advanced discussion of the statistical methods.

Thank you for participating in the CAP CVL program!

2 College of American Pathologists Calibration Verification/Linearity Surveys

 CVL Evaluations and Reports

CVL Evaluations and Reports

Understanding Your CVL Surveys Report Package

Your CVL report package includes the following individual and peer group performance summaries:
• Peer Group Summary
• Peer Performance Bar Charts
• Executive Summary
• Calibration Verification Evaluation
• Linearity Evaluation — Standard or Diluted/Extended
• Linearity Troubleshooting Report (if applicable)

This section provides a detailed description of each component.

Peer Group Summary

The Participant Summary contains a listing of peer means and coefficients of variation (CVs) for
all specimens. This information can be useful for troubleshooting problems with your calibration
verification and linearity results. We exclude outliers when we calculate peer means and CVs.
Elevated CVs can indicate problems in your peer group related to specimen mishandling,
elevated interlaboratory variability, or possible bimodality.

Peer Performance Bar Charts

The Participant Summary contains bar charts summarizing calibration verification and linearity
performance by peer group. The peer performance bar charts can also be useful to identify
possible peer-level problems. Review these charts for information on overall performance as well
as performance within your peer group.

Executive Summary

The Executive Summary lists your calibration verification and linearity results for all analytes in a given
Survey. The Executive Summary also indicates if your verified and/or linear ranges do not
include all of your reported specimen results. If we have omitted results for low and/or high
specimen levels to obtain a verified or linear evaluation, it is important to review your evaluation
report in more detail to determine if excluded specimens reveal possible analytical problems.
Finally, the Executive Summary includes a note if we are unable to complete your calibration
verification or linearity evaluation due to lack of a peer group (No Peer Group) or insufficient data
(Insufficient Data for Evaluation).

800-323-4040 | 847-832-7000 Option 1 | cap.org 3

 CVL Evaluations and Reports

Calibration Verification Evaluation

The calibration verification evaluation compares participant results to method-appropriate target

values. Most commonly, target values equal the peer means calculated from participants in the
current mailing. In accuracy-based surveys, we assign target values by reference or definitive
method.

EVALUATION
LN11-B Serum Ethanol Calibration Verification/Linearity
ORIGINAL Serum Ethanol mg/dL Calibration Verification Evaluation

l
1 Evaluation Result: Verified from 14.95 to 577.65 Allowable Error: 8% or 2 mg/dL,
Peer Method: GAS CHROMATOGRAPHY whichever is greater

Specimen Assay Assay Your Peer 2 Peer l

l 3 Difference l
4 Allowable
1 2 Mean Mean N Error

LN11-08 14.7 15.2 14.95 15.91 12 -0.96 mg/dL ±2 mg/dL

LN11-09 128.5 125.7 127.10 130.14 12 -2.3% ± 8.0%
LN11-10 236.8 238.1 237.45 245.23 12 -3.2% ± 8.0%
LN11-11 289.7 293.0 291.35 304.65 12 -4.4% ± 8.0%
LN11-12 338.1 348.2 343.15 359.63 12 -4.6% ± 8.0%
LN11-13 461.1 468.9 465.00 475.95 12 -2.3% ± 8.0%
LN11-14 579.5 575.8 577.65 587.39 11 -1.7% ± 8.0%

l
5
Calibration Verification Plot: Percent Differences with
Allowable Error Limits

30

20
Percent Difference

10

-10

-20

-30

0 100 200 300 400 500 600

Peer Mean

l
6 Peer Results Summary Table l
7 Peer Group Size: 12

Calibration Verification Linearity Evaluation

Range % Verified % Different % Linear % Nonlinear % Imprecise
LN11-08 - 14 83.3 8.3 91.7 0.0 0.0
LN11-08 - 13 8.3 0.0 8.3 0.0 0.0

4 College of American Pathologists Calibration Verification/Linearity Surveys

 CVL Evaluations and Reports

Your calibration verification evaluation result can be Verified or Different. Your verified range may
not include all reported specimens.

• Verified means all differences between your mean and the peer mean are within the
calibration verification allowable error limits in the range specified.

• Different means the difference between your mean and the peer mean exceeds the
calibration verification allowable error limit for at least one specimen.

If the calibration verification evaluation is Different over the full range of submitted results, we run
subsequent evaluations with results at the high and low ends systematically omitted. At the low end,
we only remove the results for the lowest specimen. We provide a final evaluation for the first range
of specimens that is Verified and includes the required number of consecutive data sets stated in
the kit instructions. If there are no verified ranges, the final evaluation result is Different. Even if your
calibration verification evaluation result is Verified in a range that includes all of your reported
specimens, we recommend reviewing your differences from your peer means and your
calibration verification plot. This information can provide early insight into problems with calibration
and repeatability.

Key to Calibration Verification Report

1 Your evaluation result will be either Verified or Different. If your evaluation result is Verified, we list
your means of the low and high specimens included in the verified range.
2 When we use peer means for target values, we list the number of participants reporting results
for each specimen. These numbers provide additional information about your peer group
performance.
3 The difference is 100 x (Your Mean – Peer Mean) / (Peer Mean). If the alternative error limit in
absolute units is greater than the error percentage, then the difference will be expressed in
absolute units.
4 The allowable error is the larger of the two limits listed above the table for that particular
specimen.
5 Your calibration verification plot provides a visual comparison between your results and peer
means. The appropriate error limits are also presented for comparison. See the Calibration
Verification Troubleshooting Guide section of the Users Guide for assistance in interpreting
common patterns.
6 The peer results summary provides information about the performance of others in your peer
group. Listed are the final evaluation ranges along with the percent of peer group members for
each range. We calculate the percentages using the peer group size as the denominator.
7 The peer group size is the highest count of reported results for the Peer N’s reported by
specimen in the top table. We use the peer group size for the denominator when we calculate
the percentages in the peer results summary table.

800-323-4040 | 847-832-7000 Option 1 | cap.org 5

 CVL Evaluations and Reports

Standard Linearity Evaluation

In the linearity evaluation, we plot the relative concentrations on the horizontal axis and the
measured results on the vertical axis. We fit a straight line through the points and determine
if this line adequately describes the relationship between your reported results and the relative
concentrations.

Your linearity evaluation result can be Linear, Nonlinear, or Imprecise (Poor Repeatability and/or
Fit). Because we may evaluate smaller ranges in your data, your linear range may not include all
reported specimens.

EVALUATION
LN3-A TDM Calibration Verification/Linearity
ORIGINAL Gentamicin µg/mL Linearity Evaluation

l
1 Evaluation Result: Linear from 1.25 to 9.90 l2 Evaluation Type: Standard
Instrument: SIEMENS ADVIA CHEM SYST l
3 Goal for Total Error (TE): 20%
Method: SYVA EMIT 2000 l
4 Mean of Included Results: 5.67 µg/mL

Specimen Assay Assay Your l

5 Best-fit l
6 Relative
1 2 Mean Target Concentration

LN3-01 1.3 1.2 1.25 1.31 0.000

LN3-02 3.2 3.0 3.10 3.05 0.200
LN3-03 4.8 5.0 4.90 4.80 0.400
LN3-04 6.2 6.4 6.30 6.54 0.600
LN3-05 8.5 8.6 8.55 8.28 0.800
LN3-06 9.7 10.1 9.90 10.02 1.000

Linearity Plot 1: Reported Results Linearity Plot 2: Differences with

with Best-fit Line Limits of Acceptable Imprecision l
8
10 TE
1.0

l
7
Difference from Best-fit Line µg/mL

8
0.5
Assay Result µg/mL

TE/4

6
0.0

-TE/4
4
-0.5

2
-1.0

-TE

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Relative Concentration Relative Concentration

l
9 Included in best-fit line
Excluded from best-fit line

6 College of American Pathologists Calibration Verification/Linearity Surveys

 CVL Evaluations and Reports

Your evaluation result can be Linear, Nonlinear of Imprecise. Your linear range may not include all
reported specimens.

• Linear means your reported results meet the criteria for acceptable linearity in the specified
range. While there may be evidence of small deviations from linearity, results are within
acceptable limits for nonlinearity and imprecision.

• Nonlinear means your reported results do not meet the criteria for acceptable linearity in any
of the ranges evaluated within the algorithm. The selection of ranges depends on whether your
results include extended range or diluted specimens.

• Imprecise (Poor Repeatability and/or Fit) means your results display too much variability
to permit a reliable determination of linearity. This evaluation result reflects either large
differences between specimen replicates (poor repeatability) or large differences between
specimen means and the best-fit line or curve (poor fit).

In the standard evaluation, if the evaluation is Nonlinear or Imprecise (Poor Repeatability and/or Fit)
over the full range of submitted results, we run subsequent evaluations with results at the high end
systematically omitted. A final evaluation result is given for the first range of specimens that is linear and
includes the required number of consecutive data sets stated in the kit instructions. If there are
no linear ranges, we base the evaluation on the full range of submitted results, and it will be either
Nonlinear or Imprecise (Poor Repeatability and/or Fit).

If you receive a Nonlinear or Imprecise (Poor Repeatability and/or Fit) evaluation, your report
package will include a linearity troubleshooting report. The troubleshooting report includes an
interpretation and set of suggested actions based on your results.

Key to Standard Linearity Evaluation

1 Your evaluation result will be Linear, Nonlinear, or Imprecise (Poor Repeatability and/or Fit). If your
evaluation result is Linear, we list your means of the low and high specimens included in the linear
range.
2 There are two linearity evaluation types: standard and diluted/extended.
3 We use the goal for total error to specify limits on both acceptable imprecision and nonlinearity.
We estimate both imprecision and nonlinearity using averages taken across all specimen levels.
The limits for both are slightly larger than one-fourth the goal for total error evaluated at the
mean of the assay results included in your best-fit line. The maximum total error goal used for the
linearity evaluation is 25%.
4 We provide the mean of included results to assist with the interpretation of your linearity results.
We calculate the limits of both acceptable imprecision and nonlinearity using the goal for total
error and the mean of included results.
5 The best-fit target is the fitted value from the best-fit line shown in Linearity Plot 1 for each specimen.
6 The relative concentration is the proportional concentration of analyte in the specimen. In most
Surveys, we standardize the proportional concentration to zero for the lowest specimen and one
for the highest specimen.
7 These plots use the units of measure from your test result forms. In some cases, the unit of measure
is a percent.
8 The labels on this axis show the differences on the scale of total error units. One total error unit is
equal to (Goal for Total Error / 100) x (Mean of Included Results).
9 In the standard evaluation, results are excluded from the best-fit line if the evaluation result is not
linear in the full range of reported results.

800-323-4040 | 847-832-7000 Option 1 | cap.org 7

 CVL Evaluations and Reports

Diluted/Extended Linearity Evaluation

You may receive a diluted/extended linearity evaluation if your results include at least one extended
range or diluted specimen. An extended range specimen follows a large gap in the relative
concentrations; the diluted/extended evaluation prevents these results from having a large
influence on the estimated best-fit line. You will receive a diluted/extended evaluation for a diluted
specimen when two conditions are met: 1) You indicate that you diluted both assays for the
specimen, and 2) You have an adequate number of undiluted specimen results that can be
evaluated using the standard linearity evaluation.

In a diluted/extended evaluation, we first evaluate the results excluding the diluted or extended
range specimens using the standard linearity protocol. If this initial evaluation is linear, we then
evaluate the first diluted or extended range specimen for consistency with the fitted line. The
evaluation algorithm calculates the predicted value for the diluted or extended range specimen by
extending the line fit in the first step to the additional specimen. We compare the predicted value
to the mean of your diluted or extended range specimen results. If the difference is smaller than the
allowable error limit, we include the specimen in your linear range.

The diluted/extended evaluation also includes a screen for large replicate differences for the diluted
or extended range results. If the difference between replicates for a diluted or extended range
specimen exceeds the replicate limit, the linear range will not be extended, regardless of the
agreement between your mean and the predicted value.

If the evaluation is Nonlinear or Imprecise (Poor Repeatability and/or Fit) over the full range of
submitted results excluding the diluted or extended range specimens, we run subsequent
evaluations with results at the high end systematically omitted. We determine an intermediate
evaluation result for the first solution range that is linear and includes the required number of
consecutive data sets stated in the kit instructions. If none of the ranges evaluated are linear, the
evaluation given is based on the full range of regular specimens and will be either Nonlinear or
Imprecise (Poor Repeatability and/or Fit). For example, if your seventh specimen is diluted and your
evaluation result is Linear over the first through fifth specimens, we do not evaluate the diluted
seventh specimen because the undiluted sixth specimen is not included in the linear range.

NOTE: See the Diluted/Extended Linearity Evaluation flow chart in Appendix 3 for more information.

8 College of American Pathologists Calibration Verification/Linearity Surveys

 CVL Evaluations and Reports

Key to Differences Between Diluted/Extended and Standard Linearity Evaluations When

Page 7
ResultsCollege
includeof AmericanExtended
Pathologists Range
CAP Specimens
Number: 8034828-01
Institution: Providence Lab Svcs
Kit #: 01 Kit ID: 25174580
Kit Mailed: 07/30/2012
325 Waukegan Road, Northfield, Illinois 60093-2750
800-323-4040 - http://www.cap.org Attention: MRS. Debbie Reed MT(ASCP) Original Evaluation: 08/29/2012
City/State: Portland, OR 97213-1545 Next Mailing Date: 01/28/2013
Advancing Excellence
LN13-B Blood Gas Calibration Verification/Linearity 1 Your linear range may
EVALUATION
ORIGINAL PO2 mm Hg Linearity Evaluation contain the extended
Evaluation Result: Linear from 29.0 to 485.5 Evaluation Type: Diluted/Extended range specimen(s).
Instrument: SIEMENS RAPIDLAB 1260 Goal for Total Error (TE): 16%
Barometric Pressure: > OR = 720 mm Hg Mean of Included Results: 83.1 mm Hg No range is given if
Specimen Assay
1
Assay
2
Your
Mean
Best-fit
Target
Relative
Concentration
your result is Nonlinear or
LN13-06 28 30 29.0 28.6 0.000 Imprecise (Poor
LN13-07 63 63 63.0 63.2 0.077
LN13-08
LN13-09
101
142
99
139
100.0
140.5
100.7
140.0
0.160
0.247
Repeatability and/or Fit).
§ LN13-10 483 488 485.5 480.0 1.000
§ Extended range specimen 2 Extended range
specimens are identified
Linearity Plot 1: Reported Results Linearity Plot 2: Differences with in the Specimen column.
with Best-fit Line Allowable Error Limits
500
3 The best-fit target is
the fitted value from the
50

400
Difference from Best-fit Line mm Hg

best-fit line shown in

Linearity Plot 1. We do
Assay Result mm Hg

300

not include extended

0

200
range specimen results
when finding the best-fit
line. We calculate the
-50

100

best-fit targets for the

extended range
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Relative Concentration Relative Concentration specimens from the line
Best-fit line Included in best-fit line fit to the nonextended
Extended line Excluded from best-fit line
range specimens.

Your Extended Range Specimen(s) Analysis 4 If your nonextended

Specimen
Replicate
Difference 1
range results are linear
Your Mean
Best-fit
Target
Difference
From Target
Allowable
Error 2 Evaluation
LN13-10 5 in the full range, we
485.5 480.0 5.5 ± 38.4 Included in linear range
Limit for replicate difference = 5.3% of Best-fit Target
1
2
Allowable Error = 8% of Best-fit Target evaluate your extended
range specimens for
consistency with the best-fit line. We expand your linear range to include consecutive extended range
results that are within the allowable error limits. If the difference between replicates for the extended
range specimen exceeds the replicate limit, the linear range will not be extended. If your evaluation is
not linear, we will not evaluate your extended range specimen results for linearity and your report will
not include an extended range specimen analysis. See Section 5 of Appendix 4 for information on the
determination of the allowable error limits.

800-323-4040 | 847-832-7000 Option 1 | cap.org 9

 CVL Evaluations and Reports

Key to Differences Between Diluted/Extended and Standard Linearity Evaluations When

Results Include Diluted Specimens

LN3-A TDM Calibration Verification/Linearity

EVALUATION
ORIGINAL Phenobarbital µg/mL Linearity Evaluation
1 Your linear range
Evaluation Result: Linear from 9.05 to 72.70 Evaluation Type: Diluted/Extended
may contain your diluted
specimen(s). We do not
Instrument: SIEMENS ADVIA CHEM SYST Goal for Total Error (TE): 20%
Method: SYVA EMIT 2000 Mean of Included Results: 35.03 µg/mL

Specimen Assay Assay Your Best-fit Relative specify a range if your result is
1 2 Mean Target Concentration

LN3-01 8.9 9.2 9.05 8.98 0.000 Nonlinear or Imprecise (Poor Re-
LN3-02 24.7 21.9 23.30 22.00 0.200
LN3-03
LN3-04
34.5
45.1
32.6
48.6
33.55
46.85
35.03
48.06
0.400
0.600
peatability and/or Fit).
LN3-05 61.1 63.7 62.40 61.08 0.800
* LN3-06 73.3 72.1 72.70 74.10 1.000

2 Diluted specimens are

* Diluted specimen

Linearity Plot 1: Reported Results Linearity Plot 2: Differences with identified by an asterisk in the
Specimen column. Both assays
with Best-fit Line Allowable Error Limits

15
●
●
70
must be diluted for the specimen
to be designated as a diluted
10
Difference from Best-fit Line µg/mL

60

specimen.
5
Assay Result µg/mL

50
0

●
3 The best-fit target is
40
●

the fitted value from the

-5

30

best-fit line shown in Linearity

-10

20

10
Plot 1. We do not include diluted
-15

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
specimen results when finding
Relative Concentration Relative Concentration
the best-fit line. We calculate
the best-fit targets for the diluted
Best-fit line Included in best-fit line ◆ Diluted & Included in best-fit line
Extended line Excluded from best-fit line ● Diluted & Excluded from best-fit line

specimens from the line fit to your

Your Diluted Specimen(s) Analysis undiluted results.
Replicate Best-fit Difference Allowable
Specimen Difference 1
Your Mean Target From Target Error 2 Evaluation
LN3-06 1.2 72.70 74.11 -1.41 ± 7.41 Included in linear range
1 Limit for replicate difference = 6.7% of Best-fit Target
2
Allowable Error = 10% of Best-fit Target

4 If your undiluted results are linear in the full range, we evaluate your diluted specimens for
consistency with the best-fit line. If your diluted specimens are within the allowable error limits, we
extend your linear range to contain consecutive diluted specimens. If the differences between
replicates for any diluted specimen exceeds the replicate limit, the linear range will not be
extended. If your evaluation is not linear for the full range of undiluted specimens, we will not
evaluate your diluted specimens for linearity and your report will not include a diluted specimen
analysis. See Section 5 of Appendix 4 for information on the determination of the allowable
error limits.

10 College of American Pathologists Calibration Verification/Linearity Surveys

 CVL Evaluations and Reports

Linearity Troubleshooting Report

You will receive a linearity troubleshooting report for each analyte with a linearity Page 10 evaluation result
CAP Number: Kit #: Kit ID:
of Nonlinear or Imprecise (Poor Repeatability and/or Fit).
College of American Pathologists
325 Waukegan Road, Northfield, Illinois 60093-2750 Institution: Kit Mailed:
800-323-4040 - http://www.cap.org Attention: Original Evaluation:
City/State: Next Mailing Date:
Advancing Excellence

EVALUATION
LN2-A Chemistry/Lipid/Enzyme Calibration Verification/Linearity
ORIGINAL CO2 mmol/L Linearity Troubleshooting Report

Evaluation Result: Imprecise (Poor Repeatability and/or Fit)

Troubleshooting Report Plot: Differences
Interpretation: with Allowable Error Limits
Your results have large differences between assay replicates. TE

Difference from Best-fit Line mmol/L

4
Suggested Actions:
Review specimen handling procedures. Incomplete thawing,
inadequate mixing, or improper temperatures can cause problems. 2
TE/4
Check for evidence of test system malfunction. Pipetting errors, 0
temperature fluctuations, or optical system problems can contribute
to large analytical errors. -TE/4
-2

If unable to resolve, contact your method manufacturer.

-4

-TE

0.4 0.6 0.8 1.0

Relative Concentration

Document investigation and corrective action here or on the CVL Investigation Checklist:

Reviewed by: Date:

Key to Linearity Troubleshooting Report

1 The interpretation describes the pattern of deviations from linearity in your data. Visual inspection of
your difference plot should confirm the interpretation.

2 Suggested actions are provided based on the interpretation of your deviations from linearity.

3 The troubleshooting report plot shows the deviations from the best-fit line. This plot may not show
your diluted results since the interpretation and suggested actions focus on problems within
your analytical measurement range (AMR). If you have fewer undiluted specimens than required for
a linearity evaluation, your difference plot will include both undiluted and diluted results.

800-323-4040 | 847-832-7000 Option 1 | cap.org 11

 Summary of Regulatory Requirements

Summary of Regulatory Requirements

The CAP Calibration Verification/Linearity Surveys provide specimens and statistical evaluations of
the reported results for verification of your current calibration settings as well as for assessing the
analytical measurement range (AMR) of your laboratory method. The CVL Surveys satisfy the
requirements for scheduled calibration verification and AMR verification as specified in the CAP
Laboratory Accreditation Program (LAP) and Current CLIA Regulations Section 493.1255 for most
analytes.1,2 As defined in the CAP LAP Chemistry and Toxicology Checklist, the AMR is the range of
analyte values that a method can directly measure on the specimen without any manual dilution,
concentration, or other pretreatment that is not part of the usual assay process.

The CAP CVL Surveys materials are intended to cover broad ranges of concentrations or activities
to challenge assays from near the low end to the top of the AMR for most analytes and methods.
To fulfill CAP LAP requirements for AMR verification, you must know your analytical measurement
range for each analyte, and CVL Survey concentrations must fall near the low, middle, and high
values of your AMR. If specimens do not meet these requirements, you may find additional
information on regulatory requirements in the Frequently Asked Questions section of this document.

You may use either your calibration verification or linearity evaluation, or a combination of both,
to fulfill requirements for calibration verification or AMR verification. You must confirm that the
performance limits specified in the evaluation are acceptable to your laboratory. For AMR
verification, you must confirm that the evaluation includes specimens near the low, midpoint, and
high values of the AMR. If your CVL Surveys evaluation does not satisfy your own criteria, you have
the option of performing a self–evaluation with alternate target values and/or with modified limits
of allowable error.

NOTE: See the calibration verification and linearity troubleshooting guides for suggested actions if
you have problematic results. Appendix 1 provides a worksheet for calibration verification
self-evaluation. Appendix 2 provides a detailed investigation checklist.

Coagulation

According to the CAP LAP Hematology and Coagulation Checklist, coagulation tests based on
direct measurement of an analyte require AMR verification at least every six months.8 Specifically,
calibrated tests that directly measure activity or concentration of an analyte by enzyme
immunoassay, immunoturbidity, or chromogenic methods require AMR verification. Clot-based
tests do not require AMR verification.

Hematology and Flow Cytometry

According to the CAP LAP Hematology and Coagulation Checklist, linearity studies are not
required to satisfy calibration verification for complete blood count (CBC) instruments. Similarly,
the Flow Cytometry Checklist does not contain explicit requirements for confirmation of linearity for
enumeration of blood lymphocytes. In an effort to standardize reports and documentation, we use
the CAP LAP Chemistry and Toxicology Checklist terminology for most CVL program documents.
Participants should be aware that specific accreditation and regulatory requirements could vary.

12 College of American Pathologists Calibration Verification/Linearity Surveys

 Calibration Verification Troubleshooting Guide
Page 40
College of American Pathologists CAP Number: 7541477-01 Kit #: 01 Kit ID: 24300354
325 Waukegan Road, Northfield, Illinois 60093-2750 Institution: Pfizer Asia Pacific Pte Ltd c/o Pfi Kit Mailed: 06/13/2011
800-323-4040 - http://www.cap.org Attention: Jessie Chan BSC Original Evaluation: 07/25/2011
City/State: Singapore, 188770 Next Mailing Date: 12/05/2011
Advancing Excellence

Calibration Verification Troubleshooting Guide

EVALUATION
LN2BV-A 2011 Chemistry/Lipid/Enzymes Calibration Verification/Linearity
ORIGINAL Alkaline Phosphatase U/L Calibration Verification Evaluation

Evaluation Result: Different

Peer Instrument: BECKMAN UNICEL DxC SYST Goal for Total Error: 25%
This troubleshooting guide providesMinimum
Peer Reagent: BECKMAN COULTER/37 C
suggested actions if you receive a calibration verification
Detectable Difference: 5 U/L

evaluation
Specimen Assay
1
result
Assay
2
ofMean
Different,
Your Peer
Mean
or ifN your evaluation
Peer Difference Allowable
Error
result is Verified over a range that does not
include32all of your
LNBV-21 34 reported
33.0 30.1results.
353 To use 9.6%this guide,
± 16.6% determine which of the following examples
is most502similar505to your
503.5 calibration verification plot. Refer to the corresponding suggested actions,
LNBV-22 271 270 270.5 234.9 353 15.2% ± 12.5%
LNBV-23 437.1 353 15.2% ± 12.5%
LNBV-24 739 731 735.0 656.7 353 11.9% ± 12.5%
in conjunction
LNBV-25 981 991with the CVL
986.0 880.9Investigation
353 11.9%Checklist± 12.5% for Problematic Results, to investigate possible
LNBV-26 1255 1244 1249.5 1079.9 352 15.7% ± 12.5%
causes and corrective actions.
LNBV-27 1510 1502 1506.0 1294.0 352 16.4% ± 12.5%

Constant Bias
Review the ANALYTICAL section of the investigation
Page 32
College of American Pathologists CAP Number:
Institution:
1084401-04 Kit #: 01
Regina Genl Hosp Main Lab
checklist.
Kit ID: 23469746
Kit Mailed: 06/13/2011
Analytical problems that produce a constant bias
may07/25/2011
be due to a calibration error. Recalibration may be
325 Waukegan Road, Northfield,40Illinois 60093-2750
800-323-4040 - http://www.cap.org Attention: Virginia Marsh MLT Original Evaluation:
City/State: Regina, SK S4P 0W5
needed.
Next Mailing Date: 12/05/2011
Percent Difference

Advancing Excellence

E V A L U A T 20
ION
LN2-A 2011 Chemistry/Lipid/Enzymes Calibration Verification/Linearity
ORIGINAL Review the CLERICAL
Direct Bilirubin mg/dL Calibration Verification Evaluation section of the investigation checklist.
Evaluation Result: Verified from 0.10 to 11.20
Clerical errors that result in a constant bias are likely due to
Goal for units of22%measure or decimal place errors, or incorrect peer
0
Peer Instrument: ABBOTT ARCH C/AEROSET Total Error:
Peer Method: DIAZONIUM SALT/ION w/BL group
Minimum Detectable Difference: assignment. Clerical errors may indicate a need for
0.3 mg/dL

Specimen
-20
Assay Assay Your Peer Peer Difference additional staff training.
Allowable
1 2 Mean Mean N Error

LN-08 0.1 0.1 0.10 0.10 54 0.00 mg/dL ± 0.30 mg/dL

LN-09 3.0 3.1 3.05 3.03 68 0.7% ± 11.0%
LN-10 -40 5.9 6.0 5.95 5.84 68 1.9% ± 11.0%
LN-11 8.7 8.8 8.75 8.52 68 2.7% ± 11.0%
LN-12 11.1 11.3 11.20 11.04 68 1.4% ± 11.0%
LN-13 17.1 17.0 17.05 13.34 67 27.8% ± 11.0%
0 200 400 600 800 1000 1200

Peer Mean

Problems With Low or High Specimens

Peer Results Summary Table Peer Group Size: 353

Calibration Verification Linearity Evaluation

CAP Number: 1245501-01 %Kit #: 01
Review the ANALYTICAL section of the investigation
Page 30
Kit ID: 24282122
College of American Pathologists
40Range % Verified % Different Linear % Nonlinear % Imprecise
325 Waukegan Road, Northfield, Illinois 60093-2750
LNBV-21 - 27
Institution: Maimonides Med Ctr Path & Lab Med
92.4
checklist. Analytical problems that appear at the low or high
Kit Mailed: 06/13/2011
800-323-4040 - http://www.cap.org Attention: Robert4.5 99.4
D. Kalter MD,MS 0.0 Original Evaluation:
0.0 07/25/2011
LNBV-21 - 26 0.8
City/State: 0.0 NY 11219-2916
Brooklyn, 0.3 end may be due to recovery issues near or at the
0.0 Next Mailing 0.0Date: 12/05/2011
Advancing Excellence LNBV-22 - 27 0.8 0.0 0.0 0.0 0.0
analytical measurement range (AMR) limits. If your results
Percent Difference

20
LNBV-21 - 25 0.8 0.0 0.3 0.0 0.0
LN2BV-A 2011 Chemistry/Lipid/Enzymes Calibration Verification/Linearity
E V A L U A LNBV-21
TION
ORIGINAL
- 24 0.6 0.0 0.0 0.0
were diluted, review your laboratory’s dilution protocol.
0.0
Uric Acid mg/dL Calibration Verification Evaluation

Evaluation Result:
0 Different
Peer Instrument: BECKMAN UNICEL DxC SYST
Review the SPECIMEN HANDLING section of the investigation
Goal for Total Error: 17%
Peer Method: URICASE checklist. Problems at the low or high end may also indicate
Minimum Detectable Difference: 0.5 mg/dL

Specimen Assay Assay Your Peer Peer Difference sample degradation.

Allowable
-20
1 2 Mean Mean N Error

LN-01 0.9 0.9 0.90 0.93 356 -3.2% ± 53.8%

LN-02 5.0 5.0 5.00 4.98 357 0.4% ± 10.0%
LN-03 7.1 7.0 7.05 8.94 357 -21.1% ± 8.5%
-40
LN-04 10.4 10.3 10.35 13.36 356 -22.5% ± 8.5%
LN-05 17.3 17.3 17.30 17.31 355 -0.1% ± 8.5%
* LN-06 21.2 21.1 21.15 20.91 324 1.1% ± 8.5%
* LN-07 0
24.6 2
24.9 4 24.75
6
24.83
8 10
296 12
-0.3% ± 8.5%
Peer Mean
* Diluted specimen

Problems With Middle Specimens

Peer Results Summary Table Peer Group Size: 68

Calibration Verification Linearity Evaluation

Review the SPECIMEN HANDLING section of the investigation
Range % Verified % Different % Linear % Nonlinear % Imprecise
checklist. Specimen handling problems may be due to tests
LN-08 - 13 76.5 0.0 76.5 0.0 0.0
40
LN-08 - 12 1.5 1.5 2.9 0.0 performed on an incorrect vial, mixing or reconstitution
0.0
LN-09 - 13 19.1 0.0 20.6 0.0
problems, or improper specimen storage. Check whether
0.0
Percent Difference

LN-09 - 12 1.5 0.0 0.0 0.0 0.0

20 special instructions were performed correctly.

0 ◆ ◆
Review the CLERICAL section of the investigation checklist.
◆
Clerical errors that show inconsistent recoveries for the
middle specimens are likely due to a fax scanning,
transcription, or result entry errors. If you suspect a fax
-20

◆
scanning error, you must contact the CAP Customer Contact
Not Diluted
Diluted
-40
Center. If you suspect a transcription or result entry error,
consider additional staff training.
0 5 10 15 20 25

Peer Mean

Peer Results Summary Table Peer Group Size: 357

Calibration Verification Linearity Evaluation 800-323-4040 | 847-832-7000 Option 1 | cap.org 13

Range % Verified % Different % Linear % Nonlinear % Imprecise
LN-01 - 07 60.5 1.7 60.4 0.0 0.0
 Peer Instrument: VITROS 5,1 FS & 5600 Goal for Total Error: 22%
Peer Method: SPECTROPHOT w/o BLNK/DBIL Minimum Detectable Difference: 0.3 mg/dL

Specimen Assay Assay Your Peer Peer Difference Allowable

1 2 Mean Mean N Error
Calibration Verification Troubleshooting Guide
LN-08 0.0 0.0 0.00 0.00 167 0.0% ± 11.0%
LN-09 3.9 3.2 3.55 3.11 180 14.1% ± 11.0%
LN-10 6.7 6.7 6.70 6.54 180 2.4% ± 11.0%
LN-11 10.4 10.3 10.35 10.12 180 2.3% ± 11.0%
LN-12 13.9 13.6 13.75 13.54 180 1.6% ± 11.0%
LN-13 16.8 16.9 16.85 16.74 180 0.7% ± 11.0%

Large Difference Between Replicates

for a Single Specimen
Review the CLERICAL section of the investigation checklist.
40
Clerical errors Page 58 that may cause a large difference between
College of American Pathologists CAP Number: 1137001-01 Kit #: 01 Kit ID: 23624914
325 Waukegan Road, Northfield, Illinois 60093-2750 Institution: Anna Jaques Hosp Main Lab replicates for a single specimen are likely due to fax scanning,
Kit Mailed: 06/13/2011
800-323-4040 - http://www.cap.org
20
Attention: Carol Gamble MT(ASCP)MS
City/State: Newburyport, MA 01950-3867
transcription,
Original Evaluation: 07/28/2011
Next Mailing Date: 12/05/2011
or result entry errors. If you suspect a fax scanning
Percent Difference

Advancing Excellence
error, you must contact the CAP Customer Contact Center. If
LN2-A 2011 Chemistry/Lipid/Enzyme Calibration Verification/Linearity
EVALUATION
ORIGINAL
you suspect a transcription or result entry error, additional staff
Lipase U/L Calibration Verification Evaluation
0
training may be needed.
Evaluation Result: Verified from 21.5 to 250.5
Peer Instrument: ROCHE COBAS c500/700 SER Goal for Total Error: 35%
Peer Reagent: ROCHE/37 C Review the SPECIMEN HANDLING section of the investigation
Minimum Detectable Difference: 8 U/L

Specimen
-20
Assay Assay Your Peer Peer Difference checklist.
Allowable Specimen handling problems may be due to tests
1 2 Mean Mean N performed
Error on an incorrect vial, mixing or reconstitution
LN-21 22 21 21.5 21.6 150 -0.5% problems,
± 37.0% or improper specimen storage. Check whether
LN-22 -40 75 75 75.0 72.4 150 3.6% ± 17.5%
LN-23 122 122 122.0 119.0 150 2.5% special
± 17.5% instructions were performed correctly.
LN-24 169 170 169.5 163.9 150 3.4% ± 17.5%
LN-25 0 211 2115 211.0 10206.7 150 15 2.1% ± 17.5%
LN-26 251 250 250.5 245.5 150 2.0% ± 17.5%
* LN-27 345 341 Peer
343.0 Mean281.8 150 21.7% ± 17.5%
* Diluted specimen
Peer Results Summary Table Peer Group Size: 180
Problems With
CalibrationDiluted
Verification Specimens
Linearity Evaluation
Range % Verified % Different % Linear % Nonlinear % Imprecise
Review the ANALYTICAL section of the investigation checklist.
LN-08 - 13 83.9 6.7 90.6 0.0 1.1
60
LN-08 - 12 0.0 0.0 1.1 0.0 Check
0.0 whetherPage 6
the dilution protocol was followed (eg, dilution
LN-09
College of American Pathologists
LN-09 - 12
- 13 5.0 2.2
CAP Number: 7197664-01
0.0
6.7
Kit #: 01
0.0 Plough Rsch Inst
0.6
0.0
0.0
factor, diluents
0.0Kit ID: 24282659
0.0
used). Confirm that the autodiluter is functioning
325 Waukegan Road, Northfield, 40 Institution: Schering Kit Mailed: 06/13/2011
LN-10 - 13
Illinois 60093-2750
2.2 0.0 0.0 correctly.
0.0 Original Evaluation:
0.0 If you suspect that the autodiluter is not functioning
Percent Difference

800-323-4040 - http://www.cap.org Attention: Bhavana Bhatt 07/25/2011

Advancing Excellence
20
City/State: Summit, NJ 07901-1200
◆
◆ properly, you must contact the instrument manufacturer.
Next Mailing Date: 12/05/2011

EVALUATION
LN2BV-A 2011 Chemistry/Lipid/Enzymes Calibration Verification/Linearity
ORIGINAL 0 Review the CLERICAL
Chloride mmol/L Calibration Verification Evaluation section of the investigation checklist.
Evaluation Result: Different Clerical errors with diluted specimens are likely due to use of an
-20
Peer Instrument: BECKMAN UNICEL DxC SYST Goalincorrect dilution factor.
for Total Error: 8%
Peer Method: ION SELECT ELECT DIL Minimum Detectable Difference: 2 mmol/L
Not Diluted
Specimen
-40
Assay Assay Your Peer Peer Difference◆ Dilution
Allowable errors resulting from a failure to adhere to protocol may
Diluted
1 2 Mean Mean N Error
-60
indicate a need for additional staff training.
LN-01 61 62 61.5 60.7 374 1.3% ± 4.0%
LN-02 79 79 79.0 80.4 376 -1.7% ± 4.0%
LN-03 97 97 97.0 100.0 376 -3.0% ± 4.0%
LN-04 11350 114100 113.5150 120.0
200 376250 -5.4% ± 4.0%
LN-05 130 131 130.5 139.4 376 -6.4% ± 4.0%
LN-06
LN-07
146
162
146
161
Peer
146.0
161.5
Mean
158.6
177.9
373
368
-7.9%
-9.2%
±
±
4.0%
4.0%
Peer Results Summary Table Peer Group Size: 150

Calibration Verification Linearity Evaluation

Proportional
Range Bias% Different
% Verified % Linear % Nonlinear % Imprecise
LN-21 - 27
LN-21
15 - 26
83.3
16.7
0.0
0.0
86.0
14.0
0.0
0.0
Review
0.0
0.0
the ANALYTICAL section of the investigation checklist.
Analytical problems that demonstrate increasing or decreasing
10 bias may be due to a calibration error. Recalibration may be
needed.
Percent Difference

Review the SPECIMEN HANDLING section of the investigation

0 checklist. Specimen handling problems that result in proportional
bias may be due to mixing or reconstitution problems or improper
-5
storage.

-10

-15

60 80 100 120 140 160 180

Peer Mean
Peer Results Summary Table Peer Group Size: 376

Calibration Verification Linearity Evaluation

Range % Verified % Different % Linear % Nonlinear % Imprecise
LN-01 - 07 91.0 2.7 96.5 0.0 0.3
LN-01 - 06 1.9 0.0 1.3 0.0 0.0
LN-02 - 07 1.3 0.0 0.5 0.0 0.0
LN-01 - 05 1.3 0.0 0.8 0.0 0.0
LN-02 - 06 0.0 0.0 0.3 0.0 0.0
LN-03 - 07 0.5 0.0 0.3 0.0 0.0
LN-01 - 04 0.3 0.0 0.0 0.0 0.0
LN-03 - 06 0.3 0.0 0.0 0.0 0.0
LN-04 - 07 0.8 0.0 0.0 0.0 0.0

14 College of American Pathologists Calibration Verification/Linearity Surveys

 Linearity Troubleshooting Guide

Linearity Troubleshooting Guide

If you receive a linearity evaluation result of Nonlinear or Imprecise (Poor Repeatability and/or Fit),
your report package will include a Linearity Troubleshooting Report. This troubleshooting report
provides an interpretation and set of suggested actions based on the deviations from linearity
displayed in your results. The following examples show diagnostic plots, interpretations, and
suggested actions taken from recent
Page 22 linearity troubleshooting reports.
01 Kit #: 01 Kit ID: 23855980
p Lab Svcs You may receive an evaluation
Kit Mailed: 06/13/2011result of Linear over
a range that does not include all of your
atory Original Evaluation: 11/01/2011
GA 30901-2612 reported results. This Date:
Next Mailing outcome will be noted on your
12/05/2011 Executive Summary, but you will not receive
a troubleshooting report. The final example in this guide shows a linear evaluation result over a
mistry/Lipid/Enzyme Calibration
partial range ofVerification/Linearity
specimens with a set of suggested actions. We recommend that you review
rus mg/dL Linearity Troubleshooting Report
these suggested actions when your linear range does not include all of your reported specimens.

or Fit)
Problems With Diluted Results
Review your dilution protocol, including appropriateness
arge differences
of the diluents.
Line mg/dL

Ensure that the dilution protocol is followed.

Difference

s of the
Rely on calibration verification results for undiluted
From Best-fit

-1 ◆
◆ specimens.
DifferencePercent

◆ If your calibration verification evaluation result is Different,

◆
recalibrate and consider running a new linearity study.
imens. -2
Page 22
01
alibrate Kit #: 01 Kit ID: 24056589
George VAMC Path & Lab Med Kit Mailed: 06/13/2011
ganus MT SBB -3 Original Evaluation: 11/01/2011
◆
NC 28805-2576 Next Mailing
PeerDate:
Mean12/05/2011
0.0 0.2 0.4 0.6 0.8
Relative
mistry/Lipid/Enzyme Calibration Verification/Linearity
Concentration
rus mg/dL Linearity Troubleshooting Report
Note: To highlight differences between your
Indeterminate­
undiluted and diluted—results,
Problems Within AMR
the best-fit
or Fit)
and/or With Diluted Results
line is calculated from the undiluted results.

having problems TE Review your dilution protocol and ensure the protocol is
Difference From Best-fit Line mg/dL

he CVL Investigation Checklist:

s, or both. being followed.

0.5 Rely on calibration verification results for undiluted specimens.

ol is being
◆
TE/4
If your calibration verification evaluation result is Different,
◆
◆ recalibrate and consider rerunning a linearity study.
0.0
imens.
-TE/4
alibrate
-0.5

-TE

0.0 0.2 0.4 0.6 0.8

Peer Mean
Relative Concentration

800-323-4040 | 847-832-7000 Option 1 | cap.org 15

e CVL Investigation Checklist:
Date:
my Cmnty Hosp Dept of Path Kit Mailed: 06/13/2011
atory Supervisor Original Evaluation: 11/01/2011
ing, GA 31905-5647 Next Mailing Date: 12/05/2011
Linearity Troubleshooting Guide
mistry/Lipid/Enzyme Calibration Verification/Linearity
ubin mg/dL Linearity Troubleshooting Report
Large Difference Between Replicates
or Fit) for a Single Specimen

ssays TE Rule out clerical error.

Difference From Best-fit Line mg/dL

Review specimen handling procedures. Incomplete thawing,

1
inadequate mixing, or improper temperatures can cause
TE/4 problems.
0
Review your testing procedure to ensure that the correct sample
wing,
-TE/4 was analyzed and reported.
problems.
Page 6
-1
01 Kit #: 01 Kit ID: 23598507
sample
ab-Federal Dr Lab Kit Mailed: 03/28/2011
llborn Original Evaluation: 11/01/2011 -TE
-2
oro, NC 27410-8149 Next Mailing Date: 10/03/2011

Flow Cytometry Calibration Verification/Linearity

3+/CD4+ % Linearity Troubleshooting
0.0 0.2 0.4 Report
0.6 0.8 1.0
Relative Concentration

or Fit)
Reported Results Not in Sequential Order
east Rule out clerical error, including the possibility that your samples
were reported in the wrong order.
Difference From Best-fit Line mg/dL

Check the AMR. Verify that you are not reporting results beyond
e CVL Investigation Checklist:
5 TE
your upper or lower AMR limit.
• Loss of recovery may be seen when values are close to the
AMR limits.
that you
TE/4 • Check for dilution errors if values outside the AMR were
0
diluted.
MR -TE/4
Page 12 Review specimen handling procedures. Incomplete thawing,
to the AMR
01 Kit #: 01 Kit ID: 24282110
inadequate mixing, or improper temperatures can cause
Roosevelt Hosp Ctr Dept o Kit Mailed: 06/13/2011
-5 -TE problems.
uhan
re diluted. Original Evaluation: 11/01/2011
, NY 10019-1147 Next Mailing Date: 12/05/2011
Review your testing procedure to ensure that the correct sample
wing,
mistry/Lipid/Enzyme Calibration
0.0 0.2 Verification/Linearity
0.4 0.6 0.8 1.0
was analyzed and reported.
e problems.
Relative Concentration
ne mg/dL Linearity Troubleshooting Report
t sample
or Fit)
Poor Fit Between Results and Best-fit
Line or Curve
TE Check if both assays are equal. You may need to update your
Difference From Best-fit Line mg/dL

1.0
testing procedures—both assays from each vial should be tested
e CVL Investigation Checklist: within the same run.

wing, 0.5 Review specimen handling procedures. Incomplete thawing,

problems. Date: TE/4 inadequate mixing, or improper temperatures can cause
problems.
0.0
sample
Review your testing procedure to ensure that the correct sample
-TE/4
was analyzed and reported.
-0.5

If unable to resolve, consider recalibration.

-1.0

-TE

0.0 0.2 0.4 0.6 0.8

Relative Concentration

Note: It is important to resolve problems within

16 College ofAMR
your American Pathologists
before assessing dilution performance. Calibration Verification/Linearity Surveys
Your plot displays only the undiluted results.
d, MA 01742-4166 Next Mailing Date: 12/05/2011

mistry/Lipid/Enzyme Calibration Verification/Linearity Linearity Troubleshooting Guide

mmol/L Linearity Troubleshooting Report

d/or Fit)
Large Differences Between Replicates
licates. TE Check if one set of assays is consistently greater than the other.
Difference From Best-fit Line mg/dL

You may need to update your testing procedures­—both assays

4
from each vial should be tested within the same run.
awing,
se problems. 2
Review specimen handling procedures. Incomplete thawing,
TE/4 inadequate mixing, or improper temperatures can cause
problems.
ng errors, 0
an contribute Check for evidence of test system malfunction. Pipetting errors,
Page 29
-TE/4 temperature fluctuations, or optical system problems can
1 Kit #: 01 -2
Kit ID: 24281977 contribute to large analytical errors.
r.ga VA OP Clin Lab Kit Mailed: 06/13/2011
aboratory Service 113 Original Evaluation: 11/01/2011 If unable to resolve, consider recalibration.
-4
ga, TN 37411-5616 Next Mailing Date: 12/05/2011
-TE
mistry/Lipid/Enzyme Calibration
0.4
Verification/Linearity
0.6 0.8 1.0
ol mg/dL Linearity Troubleshooting Report
Relative Concentration

or Fit) Poor Agreement With the Best-fit Line or Curve

and Large Differences Between Assay Replicates
ine or curve TE Review specimen handling procedures. Incomplete thawing,
15
inadequate mixing, or improper temperatures can cause
problems.
Difference From Best-fit Line mg/dL

the CVL Investigation Checklist:

10

ing, Check for evidence of test system malfunction. Pipetting errors,

problems.
5
TE/4
temperature fluctuations, or optical system problems can
contribute to large analytical errors.
0
errors, Page 38 If unable to resolve, consider recalibration.
01contribute
Kit #: 01 -5
Kit ID: 23502688 -TE/4
ss Med Ctr Lab Kit Mailed: 06/13/2011
zynski Original Evaluation: 11/01/2011
-10
NJ 07109-3550 Next Mailing Date: 12/05/2011

istry/Lipid/Enzyme
-15Calibration Verification/Linearity
-TE
rol mg/dL Linearity Troubleshooting Report
0.0 0.1 0.2 0.3 0.4 0.5 0.6

Relative Concentration

Nonlinear Results
Note: It is important to resolve problems within
your AMR before assessing dilution performance.
arity Review specimen handling procedures. Incomplete thawing,
Your plot displays only the undiluted results.
TE inadequate mixing, or improper temperatures can cause
problems.
Difference From Best-fit Line mg/dL

20

e CVL Investigation Checklist: Check the AMR. Verify that you are not reporting results
wing,
10 beyond your upper or lower AMR limit.
problems. Date: • Loss of recovery may be seen when values are close to
TE/4
the AMR limits.
hat you are 0
• Check for dilution errors if values outside the AMR were
mit.
-TE/4 diluted.
o the AMR
-10
Check your peer group performance to see if there is
were diluted. nonlinear recovery for your peer group.
-20
nlinear Check for evidence of test system malfunction.
-TE
• Aging reagents may cause low or high recovery at any level.
0.2 0.4 0.6 0.8 • Check calibration. Recalibration may be needed.
• Recalibration with a different lot may be needed.
level. Relative Concentration
• After recalibration, consider running a new linearity study to
confirm that the problems have been resolved.
Note: It is important to resolve problems within
your AMR before assessing dilution performance.
dy to Your plot displays only the undiluted results.
800-323-4040 | 847-832-7000 Option 1 | cap.org 17
rews Original Evaluation: 11/08/2011
NB E4L 4A3 Next Mailing Date: 12/15/2011
Linearity Troubleshooting Guide
mistry/Lipid/Enzyme Calibration Verification/Linearity
e mg/dL Linearity Troubleshooting Report

Linear Range Does Not Include All

Reported Specimens
rity TE Review specimen handling procedures. Incomplete
40
thawing, inadequate mixing, or improper temperatures
Difference From Best-fit Line mg/dL

can cause problems.

20

ing, TE/4 Check the AMR. Verify that you are not reporting results beyond
problems. 0
your upper or lower AMR limit.
• Loss of recovery may be seen when values are close to the
hat you are -TE/4 AMR limits.
mit. -20 • Check for dilution errors if values outside the AMR
o the AMR were diluted.
-40
-TE Check your Peer Results Summary Table to see if there are similar
were diluted.
results for others in your peer group.
-60
nlinear
Check for evidence of test system malfunction. Aging reagents
0.0 0.2 0.4 0.6 0.8 1.0
may cause low or high recovery at any level.
Relative Concentration
level.

dy to

e CVL Investigation Checklist:

Date:

18 College of American Pathologists Calibration Verification/Linearity Surveys

 Glossary of Terms

Glossary of Terms
Acceptable Imprecision Limits for Linearity Evaluation—The linearity evaluation includes a check
on the imprecision of the best-fit line or curve. The purpose of this check, formally called a power
calculation, is to reduce misclassification errors from poor fit of the regression model. We calculate
the specific limit within the statistical algorithm; it depends on the number of included results, the
goal for total error, the mean of your included results, and whether your results are fit by a line or
curve.

Acceptable Nonlinearity Limits for Linearity Evaluation—In the linearity evaluation, small deviations
from linearity are acceptable if, overall, the nonlinearity is within a clinically acceptable range.
We calculate the specific limit within the statistical algorithm; it depends on the number of included
results, the goal for total error, the mean of your included results, and the shape of the best-fit curve
(quadratic or cubic) used to detect nonlinearity. (See definition of best-fit curve.)

Allowable Error for Calibration Verification—The allowable error is the larger of the two limits listed on
the evaluation report and in the Participant Summary. Specification of an alternate allowable error
for low values prevents the error limits from becoming too small at low concentrations.

Allowable Error for Diluted/Extended Linearity Evaluations—The allowable error for extended range
specimens and diluted specimens show the acceptable difference between your results and the
value predicted from the best-fit line. If the mean of your results is within the allowable error limit,
we include consecutive extended range or diluted specimens in your linear range.

Analytical Measurement Range (AMR)—The analytical measurement range is the range of analyte
values that a method can directly measure on the specimen without any dilution, concentration,
or other pretreatment not part of the usual assay process.1

Best-fit Curve—The first step in the CAP linearity evaluation evaluates a series of polynomial curves fit
to your data. This is generally known as the polynomial method.6, 7 In this step, we may select a
quadratic or cubic curve as the best descriptor of the relationship between your reported results
and the relative analyte concentration. In a second step, we assess the amount of nonlinearity
against a non-zero threshold. You will receive an evaluation of Nonlinear and the best-fit curve is
plotted in Linearity Plot 1 if the amount of nonlinearity exceeds this non-zero threshold.

Best-fit Line—The best-fit line is the straight line fit to your included results using linear regression.

Best-fit Targets—The best-fit targets are the points on your best-fit line corresponding to each
specimen for which you reported results.

Excluded Results—We exclude assay results from the calculation of your best-fit line for one of two
reasons: 1) Your evaluation is not linear in the full range of reported results; or 2) results are evaluated
with the diluted/extended range protocol.

800-323-4040 | 847-832-7000 Option 1 | cap.org 19

 Glossary of Terms

Extended Range Specimen—If the relative concentrations have a large gap in the admixture ratios
that could permit some specimens to have severely unequal influence on the fitted regression line,
we will identify one or more of the specimens as an extended range specimen and use an alternate
linearity evaluation.

Goal for Total Error (TE)—Total error is the net or combined effect of random and systematic errors in
a test system.3 We use one-quarter of the goal for total error plus sampling variability estimated as
part of the statistical analysis as the limit of acceptable nonlinearity in the linearity evaluation.

Imprecision of Best-fit Line or Curve—Imprecision in the linearity evaluation refers to the differences
between your reported results and the best-fit line or curve. One component of this quantity is an
estimate of your repeatability or within-run precision, and it also includes a contribution from how
well the best-fit line or curve matches your assay means. The imprecision of your best-fit line or curve
could be due to unacceptably large differences between one or more pairs of assays or large
differences between your assay means and the best-fit line or curve.

Included Results—Included results are assay results that are included in the calculation of the best-fit
line or curve in the linearity evaluation.

Insufficient Data for Evaluation—Depending on the Survey, you are required to have four or five
consecutive pairs of results to receive a linearity evaluation. If you have at least three undiluted
consecutive pairs of results, you will receive a calibration verification evaluation even if you have
insufficient data for the linearity evaluation. Your Executive Summary will indicate if you have not
reported data for the minimum required number of specimens.

Peer Mean—The peer mean is the mean calculated from peer values at each solution level after
outliers are removed from the data.

Relative Concentration—The relative concentration values are the rescaled values for a range of
concentration levels where the lowest concentration is set to zero and the highest is set to one.
We have selected this scale to reflect the common use of admixtures to produce specimens. High
and low specimen pools are mixed in varying proportions to produce a set of specimens with
known relationships among the concentration or activity levels. A value of zero means the mixing
proportion is zero from the high specimen pool and, by subtraction, one from the low specimen
pool. A value of 0.25 means the mixing proportion is 0.25 from the high pool and 0.75 from the low
pool. A value of one means a mixing proportion of one from the high specimen pool and zero from
the low specimen pool.

Repeatability—Repeatability is the measurement precision of replicate measurements on the same

or similar objects over a short period of time. Repeatability may also be called within-run precision in
laboratory medicine.

Reportable Range—The reportable range is the span of test result values over which the laboratory
can establish or verify the accuracy of the instrument or test system measurement response.
(Definition provided in Current CLIA Regulations, Subpart A, Section 493.2)

20 College of American Pathologists Calibration Verification/Linearity Surveys

 Frequently Asked Questions

Frequently Asked Questions

Interpretation of the CVL Evaluations

Q: On Linearity Plot 1, it looks like my results are very close to the line. Why is my evaluation result
Imprecise (Poor Repeatability and/or Fit) or Nonlinear?
A: Linearity Plot 2 shows the differences between your results and the best-fit line on a scale that
makes it easier to see differences that are large relative to the error goals for a particular
analyte. If your evaluation is Imprecise (Poor Repeatability and/or Fit) or Nonlinear, it indicates
that the overall measure of differences is larger than the allowable error limit. Large deviations
between assay replicates (poor repeatability) and/or large deviations between replicate means
and the best-fit line (poor fit) contribute to results of Imprecise (Poor Repeatability and/or Fit) or
Nonlinear.

Q: All of my mean values are very close to the peer mean values. How can my linearity evaluation
be Nonlinear?
A: With the exception of the Blood Gas Survey (LN13), we assess linearity using the relative
concentration or activity, not the peer mean values. The mixing ratios used during the specimen
manufacturing process determine the relative concentrations. If your peer group means were
nonlinear in relation to the relative concentrations and your linearity results were similar to others
in your peer group, then your results would also be Nonlinear.

Q: How are the calibration verification allowable errors and goals for total error determined?
A: The Instrumentation Resource Committee (IRC) determines the total error goals for each new
analyte, and they are periodically reviewed by this committee. In this process, the IRC reviews
both published error goals based on biological variation and the peer group performance
statistics from relevant CAP proficiency testing and CVL Surveys.

Q: Why do the calibration verification and linearity evaluations use different error limits?
A: The calibration verification evaluation is intended to have more stringent error limits than
proficiency testing to allow early identification of analytical error. Also, because this evaluation
uses means of replicate measurements, it is appropriate to consider some reduction in the error
limits. The use of one-half the total error goal reflects a reduction in error limits due to the use
of means.

For the linearity assessment, participants take duplicate measurements of multiple samples within
the same run. This precision component is expected to be similar in magnitude to within-run
precision. The linearity imprecision limit has been set as a small fraction, approximately
one-quarter, of the total error goal because many sources of variation are excluded from
this assessment. The use of one-quarter the total error goal aligns with well-established
recommendations to use performance goals that will prevent failure on proficiency
testing challenges.9

Q: On my linearity evaluation report, the best-fit target for my first specimen is “<0.” How can the
best-fit target be a negative number?
A: As the best-fit line is influenced by all data points, this could be due to random measurement
error across the range of specimen values. It could also be an indication of nonlinear recovery
at the low end.

800-323-4040 | 847-832-7000 Option 1 | cap.org 21

 Frequently Asked Questions

Q: Why is there a diluted/extended linearity evaluation for some diluted specimens?

A: The diluted/extended linearity evaluation provides both assessment of linearity over the AMR
with undiluted specimens and validation of dilution protocols. You must have an adequate
number of undiluted specimen results that can be evaluated using the standard linearity
evaluation. If only one of the two assay results is diluted, the specimen is not classified as
a diluted specimen.

Q: I received a diluted/extended linearity evaluation, but it does not include the dilution specimen
analysis and my Linearity Plot 2 does not show the allowable error limits for my diluted specimen.
Why have these elements been omitted from my report?
A: We evaluate undiluted specimen results first, using the standard linearity evaluation. If your
undiluted results are not linear in the full range, we do not evaluate your diluted results. Since
your diluted results are not evaluated, we do not calculate the allowable error limits for the
diluted specimen analysis or include those limits in Linearity Plot 2.

Q: Why is there a limit for the difference between Assay 1 and Assay 2 in the diluted/extended
linearity evaluation?
A: Large differences between replicate measurements may be an indication of poor repeatability.
Problems with measurement precision need to be addressed in order to obtain a meaningful
assessment of linearity.

Q: Why is there a diluted/extended linearity evaluation for extended range specimens?

A: When there are severely nonequidistant relative concentrations, the specimens with higher
concentration have large influence on the fitted regression line and allowable error estimates.
The extended range linearity evaluation reduces the chance of “false linear” misclassifications
from extreme nonequidistant admixture ratios.

Q: Can a specimen that is excluded from the best-fit line also be included in the linear range?
A: Yes, this can occur with the diluted/extended linearity evaluations. In the dilution validation
protocol, we use only undiluted results to identify the best-fit line. We may extend the linear
range, however, to include diluted results that are consistent with this line. Similarly, for extended
range specimens, we use only the nonextended range specimens to identify the best-fit line.
We will expand the linear range to include extended range results that are consistent with the
fitted line.

Q: Why does my Linearity Plot 2 have points outside the grey shaded box, even though my
evaluation result is Linear over the full range of my results?
A: Points can fall outside of the shaded area for two reasons: 1) Since we use an average to
estimate imprecision, many small differences can offset a few larger differences, and 2) clinically
insignificant nonlinearity (curved fit) can contribute to differences between your results and the
best-fit straight line. In the first case, larger differences between replicates may be an early
warning sign of poor repeatability over a subset of your linear range. In the second case,
evidence of a U-shaped or inverted U-shaped trend in the difference plot may be an early
warning sign of nonlinearity.

22 College of American Pathologists Calibration Verification/Linearity Surveys

 Frequently Asked Questions

Method Verification and Regulatory Requirements

Q: How often is AMR verification required?

A: As specified in the CAP LAP Chemistry and Toxicology Checklist, laboratories are required to
verify the AMR at least every six months, or if changes are made to an assay without
recalibration. If your laboratory performs calibrations or calibration verification at least every six
months using at least three levels of material that span the AMR, separate AMR verification is not
required. Coagulation tests based on direct measurement of an analyte require AMR verification
at least every six months according to the LAP Hematology and Coagulation Checklist.

Q: How often is calibration verification required?

A: If your laboratory performs calibrations at least every six months, current CLIA regulations and
the CAP Laboratory Accreditation Program do not require separate calibration verification.
Otherwise, laboratories are required to confirm calibration verification at least every six months,
or if changes are made to an assay without recalibration.

Q: What am I required to do if my linearity evaluation result is Nonlinear and my calibration

verification result is Verified?
A: If your evaluation is Nonlinear, but your results are similar to your peer group, the nonlinearity may
be inherent in the method and it is not specific to your instrument. The same scenario may occur
due to a nonlinear matrix effect in the Survey material. In this case, it may be useful to
investigate the apparent nonlinearity with previously tested patient specimens to identify a
possible matrix effect. If your CAP calibration verification evaluation results include low, mid
point, and high values and the evaluation criteria are acceptable to your laboratory, your AMR
is verified.

Q: What am I required to do if my linearity evaluation result is Linear but my calibration verification

result is Different?
A: Make sure you are reporting in the correct units of measure and have not made a decimal
place error. If your results are in agreement for some specimens, look for evidence of a peer-
level problem. The peer group means may be influenced by laboratories that are nonlinear due
to instrument differences or because of aging reagents. In this case, the coefficients of variation
(CV) for the higher specimens tend to be much larger than the CVs of the midrange specimens.
If your evaluation is Linear and there is evidence that the mean of your peer group includes
nonlinear results, you should note that you suspect a peer-level problem.

Q: What is required by the LAP checklists to demonstrate linearity?

A: The LAP Chemistry and Toxicology Checklist requires both calibration verification and AMR
verification. You can use successful performance in the CVL Surveys to verify your AMR, but you
must confirm that the evaluation includes specimens near the low, midpoint, and high values of
the AMR. The LAP Hematology and Coagulation Checklist states that coagulation tests based
on direct measurement of an analyte require AMR verification. Calibrated tests that directly
measure activity or concentration of an analyte by enzyme immunoassay, immunoturbidity, or
chromogenic methods require AMR verification. Clot-based tests do not require AMR
verification. The checklist also states that linearity studies are not required for calibration
and calibration verification of CBC instruments.

800-323-4040 | 847-832-7000 Option 1 | cap.org 23

 Frequently Asked Questions

Q: Is it more important to be Linear or Verified?

A: If poorly performing laboratories influence peer means, a linear result is a direct confirmation
that your laboratory is performing well across the measuring interval. In this case, you should give
your linearity evaluation priority over calibration verification. However, sometimes there are
nonlinear matrix effects in the Survey material. In this case, if you suspect a peer-level failure of
linearity, you should give more weight to your calibration verification evaluation.

Q: What other options do I have for AMR verification if my analytical measurement range is not
covered by the CAP material?
A: If you are using the CVL Survey to fulfill regulatory requirements, you may need to find an
alternative product that has higher (or lower) specimens, utilize residual patient specimens for
AMR verification, or adjust your AMR to reflect the range that can be successfully verified.

Q: What am I required to do if only three specimens are within my analytical measurement range?
A: You may report three undiluted sets of results and receive a calibration verification evaluation.
You may also report additional diluted results and receive both a linearity and calibration
verification evaluation. See the kit instructions for additional information on reporting results
beyond your AMR.

Data Processing and Reporting

Q: What is an LN Express evaluation and how does it benefit me?

A: For the majority of the CVL Surveys, we offer the LN Express service, which provides your
linearity evaluation via e-LAB Solutions within two business days of receipt of your data. Results
may be submitted by any method, but they must be received by the due date printed on your
result form.

The LN Express evaluation provides the opportunity to verify your submitted results, review your
linearity evaluation, and submit any necessary data corrections prior to the start of the CVL
Survey processing. If data revisions are submitted prior to the survey due date, we will generate
another LN Express evaluation.

Your complete report package, including the calibration verification evaluation, will be
generated after the survey due date when the peer summary results can be calculated.

Q: Why does it take longer to get a calibration verification evaluation and Participant Summary
report?
A: We always use data from the current mailing to calculate peer-based target values and peer
group summary statistics. We strive to include as much data as possible, especially for smaller
peer groups. We believe this provides the most accurate peer group information for evaluations
that utilize peer-based target values. Once we have begun summarizing the peer group data,
it takes several days to generate and review the evaluation reports. We spend additional time
formatting the Participant Summary reports, which may include commentary by Instrumentation
Resource Committee members. Please note that you will have faster access to your full
evaluation report package with e-LAB Solutions.

24 College of American Pathologists Calibration Verification/Linearity Surveys

 Appendix 1: Worksheet for Calibration Verification Self-Evaluation

Appendix 1: Worksheet for Calibration Verification

Self-Evaluation
Use this worksheet to complete a calibration verification self-evaluation for any number of
specimens. If you have results near the low, middle, and high end of your AMR and the allowable
error limits meet your laboratory’s criteria, your self-evaluation may be sufficient to fulfill regulatory
requirements for calibration verification or AMR verification.

Survey and Analyte: ___________________________________________________________________________

Allowable Error: % or
(A) (B)

(1) (2) (3) (4) (5)

Specimen Your Mean Target Value Difference Percent Within Limit
Number Difference (Yes/No)

(6) Result:___________________
(1) Record the mean values of replicate assays for each specimen.
(2) Record the corresponding peer mean, or, if applicable, an accuracy-based assigned value.
(3) Difference = column (1) – column (2). If the absolute value of this result is less than the
Allowable Error (B), record a “Yes” in column (5) and skip column (4).
(4) Percent Difference = column (3) / column (2) × 100. If the absolute value of this result is less than
the Allowable Error (A), record a “Yes” in column (5).
(5) If both the Difference and Percent Difference are greater than the Allowable Error, record a
“No” in column (5). Note that both differences can be positive or negative, so compare the
absolute value of the differences with the allowable error.
(6) Calibration is Verified over the full range of specimens if all results in column (5) are “Yes.” The
calibration evaluation is Different if there are any “No” results in column (5).

Example: LN2 Lipase U/L

Allowable Error: __ 17.5 % or 8 U/L
(A) (B)

(1) (2) (3) (4) (5)

Specimen Your Mean Target Value Difference Percent Within Limit
Number Difference (Yes/No)
LN-01 20.5 22.7 2.2 Yes
LN-02 103.5 99.2 4.3 Yes
LN-03 182.0 171.1 10.9 6.4% Yes

(6) Result:_________________
Verified

800-323-4040 | 847-832-7000 Option 1 | cap.org 25

 Appendix 2: CVL Survey Investigation Checklist

Appendix 2: CVL Survey Investigation Checklist for

Problematic Results
Survey Name: Evaluation Date:

Analyte:

Specimen Handling YES NO NA

Was testing material received in the laboratory within an appropriate

time after shipment?.................................................................................................................................. .c c c
Were Survey specimens reconstituted as indicated in the kit instructions?......................................... c
. c c
Were Survey specimens stored as indicated in the kit instructions?..................................................... .c c c
Were any special instructions provided in the kit instructions performed as indicated?................... .c c c
Were the Survey specimens mixed adequately before sampling?...................................................... .c c c
Were the correct tests performed on the correct vial of testing material?......................................... c
. c c

A response of “No” to any of these questions may indicate a specimen handling error. These types of errors
could indicate a failure to read the instructions provided with the Surveys materials.

Analytical YES NO NA

Was the written procedure followed?...................................................................................................... .c c c

Was instrument maintenance performed on schedule?....................................................................... .c c c
Were quality control results acceptable?................................................................................................ .c c c
Was the most recent calibration acceptable and within established stability
limits at the time testing was performed?................................................................................................ .c c c
Does a review of recent proficiency testing results or past CVL results indicate
evenly distributed data without bias?...................................................................................................... .c c c

Were the reagents prepared according to procedure?....................................................................... .c c c

Were the reagents within their open stability acceptable range?...................................................... c
. c c
Was the intended result within the measuring range for the instrument?........................................... .c c c
Was the dilution protocol followed when diluting samples that are out of range?........................... .c c c
Does a review of records indicate that there were no related instrument/test
problems noted prior to or after the testing was performed?.............................................................. .c c c

A response of “No” to any of these questions may indicate an analytical error. These types of errors could
indicate a failure to follow recommended instrument maintenance and calibration. You may need to review
the instructions provided with the testing material and/or laboratory procedures. If recalibration has not
already occurred, recalibrate the instrument.

26 College of American Pathologists Calibration Verification/Linearity Surveys

 Appendix 2: CVL Survey Investigation Checklist

Clerical YES NO NA

Were the results correctly transcribed from the instrument read-out or report?................................. .c c c
Was the correct instrument/method/reagent code reported on the result form?............................ .c c c
Do the units of measure match between the result form and the instrument results?...................... .c c c
Is the decimal place correct?................................................................................................................... .c c c
Does the submitted result match the result found on the calibration verification
evaluation report?...................................................................................................................................... .c c c
If the result was out of range and a dilution was performed, was the correct
dilution factor used in the calculation of the final result?...................................................................... .c c c

A response of “No” to any of these questions may indicate a clerical error. Although reporting of testing results
is unlike those for patient results, clerical errors may indicate a need for additional staff training, review of kit
instructions, or investigation of the reporting format provided by the testing device. If results reported on the
result form do not match the results found on the evaluation report, please contact the CAP Customer Contact
Center at 800-323-4040.

Corrective Action:

Additional Notes:

Investigated By: Date:

Reviewed By: Date:

800-323-4040 | 847-832-7000 Option 1 | cap.org 27

 Appendix 3: Flow Charts

Appendix 3: Flow Charts

Flow Chart 1: Selection of Linearity Evaluation Type

Are there results for an

No extended range vial or Yes
are conditions met for
dilution validation?

Use Standard Linearity Use Diluted/Extended

Evaluation. Evaluation.

Conditions for extended range:

1. The CAP has identified the specimen as an extended range specimen.
2. Participant submits results for at least four additional specimens.

Conditions for dilution validation:

1. At least four consecutive specimens must be undiluted.
2. Both replicates must be diluted.

28 College of American Pathologists Calibration Verification/Linearity Surveys

 Appendix 3: Flow Charts

Flow Chart 2: Standard Linearity Evaluation

1. Participant data
Plot data against relative
(min. # required)
concentrations. Find
2. Relative
the best-fit straight line.
concentrations
3. Total error goal

No Is there any Yes

evidence of
nonlinearity?

Evaluate Evaluate
imprecision of imprecision of
best-fit line. best-fit curve.

No Is the data Yes No Is the data Yes

too imprecise? too imprecise?

Result Result
Result Imprecise Does the Imprecise
Linear (Poor Repeatability No nonlinearity Yes (Poor Repeatability
and/or Fit) exceed clinical and/or Fit)
threshold?

Result Result
Linear Nonlinear

Note: If the result is Nonlinear or Imprecise (Poor Repeatability and/or Fit), an evaluation may be rerun
on a subset of your data.

800-323-4040 | 847-832-7000 Option 1 | cap.org 29

 Appendix 3: Flow Charts

Flow Chart 3: Diluted/Extended Linearity Evaluation When Results Include

Extended Range (ER) Specimens

1. Participant data
(min. # required) Apply standard
2. Relative evaluation to
concentrations non-ER results.
3. Total error goal

Is the
No evaluation Yes
result Linear?

Result
Evaluate the first
Linear, Nonlinear or
ER vial for linearity.
Imprecise(Poor Repeatability
Fit line to non-ER
and/or Fit)
results.
Final evaluation based
on non-ER results.

Is the mean of
No the ER results Yes
sufficiently close to the
predicted value from
the fitted line?

Result
Linear No Yes
Final evaluation Is there another
linear over a range ER vial?
without ER vial.

Result Return to
Linear comparison step
Final evaluation to evaluate next
linear over a range ER vial.
containing ER vial.

30 College of American Pathologists Calibration Verification/Linearity Surveys

 Appendix 3: Flow Charts

Flow Chart 4: Diluted/Extended Linearity Evaluation When Results Include Diluted Specimens

1. Participant data
(min. # required) Apply standard
2. Relative evaluation to
concentrations undiluted results.
3. Total error goal

Is the
No evaluation Yes
result Linear?

Result Evaluate the first

Linear, Nonlinear diluted vial for
or Imprecise linearity.
(Poor Repeatability and/or Fit) Fit line to undiluted
Final evaluation based on results.
undiluted results.

Is the mean
of the diluted results
No Yes
sufficiently close to
the predicted value
from the fitted
line?

Result
Linear
Final evaluation linear No Is there another
Yes
over a range without diluted vial?
diluted vial.

Result Return to
Linear comparison step
Final evaluation linear to evaluate next
over a range containing diluted vial.
diluted vial.

800-323-4040 | 847-832-7000 Option 1 | cap.org 31

 Appendix 4: Technical Guide to the CAP Linearity Evaluation

Appendix 4: Technical Guide to the CAP Linearity

Evaluation
This appendix is intended for participants who have had training equivalent to a second course
in statistical methods and would like more details about the CAP linearity evaluation. Because this
information is intended for participants with some training in statistical methods, we do not attempt
to provide complete details for carrying out the calculations. You can reference any statistical text
for more information about fitting linear and polynomial regression models and evaluating the
statistical significance of regression coefficients (steps 1 and 2). The calculations noted in steps 3
and 4 require more advanced techniques. While these are fully documented in the manuscript
by Kroll, et al in the September 2000 issue of Archives of Pathology & Laboratory Medicine, the
explanation in this appendix will provide a set of approximate calculations using standard
regression output.

As a preliminary step, determine the type of evaluation that has been or will be used—Standard or
Diluted/Extended. See the linearity flow charts in Appendix 3 of the Users Guide for reference.
If you are trying to duplicate results on your linearity evaluation report, you must match the range
of included specimens. In a Diluted/Extended evaluation, you will use the standard evaluation for
your undiluted/ nonextended range specimens (steps 1 through 4 below).

1. Plot the measured results against the relative concentration and find the best-fit straight
line using linear regression.
Plot your results on the y-axis and the relative concentrations on the x-axis. The linearity evaluation
report lists the relative concentrations for each analyte. You can use any statistical software to fit
the regression line.

2. Evaluate the data for evidence of nonlinearity.

Nonlinearity is estimated using polynomial regression. If there are two results reported for at least
five test specimens, fit a cubic polynomial and assess the cubic term using a standard test of
statistical significance. If there are results for only four test specimens, or if the cubic term is not
statistically significant, fit a quadratic polynomial. Assess the quadratic term using a standard test
of statistical significance. You can use any standard regression software to fit the polynomials
and test the significance of the cubic and quadratic terms. The CAP uses a significance level of
alpha = 0.05 for this evaluation. If there is no evidence of nonlinearity using the polynomial
method, the best-fit curve is the straight line that was found in step 1. Otherwise, the best-fit curve
is either the quadratic or cubic polynomial. To evaluate the imprecision in step 3, you need to
identify the appropriate best-fit line or curve.

Note that in step 2, the hypothesis tests use a threshold of zero to detect deviations from linearity
(ie, the Null Hypothesis in each statistical test is that the nonlinear coefficient equals 0). After
screening for imprecision of the best-fit line or curve, the CAP linearity evaluation assesses the
amount of nonlinearity described by the quadratic or cubic polynomial relative to a non-zero
threshold, described in step 4. For step 4, you will need to calculate the fitted values from both
the best-fit straight line and the best-fit curve from your regression results.

32 College of American Pathologists Calibration Verification/Linearity Surveys

 Appendix 4: Technical Guide to the CAP Linearity Evaluation

3. Evaluate the imprecision of the best-fit line or curve.

We evaluate imprecision based on the differences between the participant results and the
best-fit curve or line. If the differences exceed the limit on imprecision, we conclude that the data
are inadequate to enable a meaningful linearity evaluation, either because of poor repeatability
or poor conformance to the specified linear and polynomial models. The limit on the imprecision
around the best-fit line or curve is determined by the goal for total error for the analyte, the
number of participant results, and a calculated value derived from the formal statistical
evaluation. While the exact calculation depends on several input variables, the limit on the
imprecision is always slightly larger than one-fourth of the goal for total error. If you are performing
a self-evaluation, using a multiple of 0.25 times the goal for total error provides an absolute lower
bound for the imprecision screen. Alternatively, you can use a multiplier from the table below
determined by the number of included results, for a close approximation to the CAP evaluation.

To calculate the estimated imprecision: Find the estimated standard error of the best-fit line or
curve from steps 1 and 2. The software that you use to fit your regression models will calculate the
standard error of the best-fit line or curve.

To calculate the imprecision limit: You need to know the mean of your results included in the
regression model, the number of results included in the model, the analyte’s goal for total error,
and the imprecision multiplier from Table 1. The limit on imprecision is the square root [(number of
included results)/(multiplier from Table 1)] x [goal for total error x .25] x (mean of included results).

Table 1—Imprecision multiplier

Best fit from step 2 Multiplier

Line 6.3
Curve – quadratic polynomial 6.3
Curve – cubic polynomial 6.5

Your evaluation result is Imprecise (Poor Repeatability and/or Fit) if your best-fit line or curve fails
the imprecision screen. If you have no evidence of nonlinearity from step 2 and the results for your
best-fit line pass the imprecision screen, your evaluation result is Linear. If you found evidence of
nonlinearity in step 2 and your results for your best-fit curve pass the imprecision screen, proceed
to step 4.

4. Evaluate nonlinearity relative to a non-zero clinical threshold, if applicable.

We use the difference between the best-fit curve from step 2 and the best-fit straight line from step
1 to evaluate the clinical relevance of nonlinearity. If the summary measure of the difference
exceeds the clinical threshold, the evaluation result is Nonlinear. Otherwise, the evaluation is
Linear.

The summary measure of the difference between the best-fit curve and best-fit line is called the
average deviation from linearity (ADL). We use a multiple of the goal for total error to determine
the threshold of clinical importance. Note that this evaluation of nonlinearity permits small,
clinically unimportant, deviations from linearity. The evaluation in step 2 flagged any apparent
deviation from nonlinearity.

800-323-4040 | 847-832-7000 Option 1 | cap.org 33

 Appendix 4: Technical Guide to the CAP Linearity Evaluation

To calculate the ADL: Calculate a single set of linear fitted values from the best-fit line from step 1.
Note that if you generate fitted values in your regression software, you may have two sets of
identical fitted values. Next, calculate the nonlinear fitted values from the quadratic or cubic curve
that was fit in step 2. Again, use just one fitted value for each specimen level. Calculate
the squared differences = (linear fits – nonlinear fits).2 Calculate the average squared difference =
(sum of the squared differences) / (number of specimen levels). Finally, the ADL = the square root of
the average squared difference.

To calculate the ADL limit: The ADL limit is the 95th percentile of a noncentral Chi-square distribution.
Because we feel this calculation is beyond the level of even a second course in statistics, we
refer readers with advanced training to the September 2000 issue of Archives of Pathology &
Laboratory Medicine for a complete description of how to calculate the exact ADL limit. The ADL limit
is always slightly larger than one-fourth the goal for total error, so we recommend using this as an
approximate limit. Specifically, to evaluate the significance of the nonlinearity estimated in the
polynomial method using a non-zero threshold, compare the calculated ADL to 0.25 x (goal for total
error) x (mean of included results).

5. Evaluate diluted or extended range vials, if applicable.

In the Diluted/Extended evaluation, we evaluate your diluted or extended range specimens only if
the full range of undiluted or non-ER results are linear. When these conditions are met, you can
extend the best-fit straight line from step 1 through the diluted or extended range vials using the
relative concentrations listed on your linearity evaluation report. The difference between the
observed and expected results is equal to the mean of your replicates minus the extrapolated
value from the best-fit line.

Your linearity evaluation page shows the allowable limits for the differences between the observed
and fitted results used in the CAP evaluation. The allowable error limit equals the best-fit target times
one-half the goal for total error. The evaluation also includes a screen for large replicate differences.
If the difference between replicates for the diluted or extended range specimen exceeds the best-fit
target times one-third the goal for total error, the linear range will not be extended, regardless of the
agreement between your mean and the best-fit target.

If you would like to override the CAP evaluation, you may determine your own limits for the allowable
differences. If you have specific questions about the calculation of the limits in the CAP evaluation,
we recommend that you contact the CAP Customer Contact Center for more information.

34 College of American Pathologists Calibration Verification/Linearity Surveys

 References

References
1. College of American Pathologists, Commission on Laboratory Accreditation. Chemistry and
Toxicology Checklist. Northfield, IL: CAP, 2012.
2. Current CLIA Regulations. Centers for Disease Control and Prevention website.
cdc.gov/clia/regs/toc.aspx. Updated January 24, 2004. Accessed September 29, 2012.
3. Linnett K, Boyd JC. Selection and analytical evaluation of methods—with statistical techniques.
In Burtis CA, Ashwood ER, Bruns DE, eds. Tietz Textbook of Clinical Chemistry and Molecular
Diagnostics. St. Louis, MO: Elsevier Saunders; 2006.
4. Clinical and Laboratory Standards Institute/NCCLS. Evaluation of the linearity of quantitative
measurement procedures: A statistical approach; Approved guideline. CLSI/NCCLS Document
EP6-A. Wayne, PA: Clinical and Laboratory Standards Institute; 2003.
5. Kroll MH, Praestgaard J, Michaliszyn E, Styer PE. Evaluation of the extent of nonlinearity in
reportable range studies. Arch Pathol Lab Med. 2000;124(9):1331–1338. http://arpa.allenpress.com.
6. Kroll MH, Emancipator K. A theoretical evaluation of linearity. Clin Chem. 1993;39(3):405–413.
7. Emancipator K, Kroll MH. A quantitative measure of nonlinearity. Clin Chem. 1993;39(5):766–772.
8. College of American Pathologists, Commission on Laboratory Accreditation. Hematology and
Coagulation Checklist. Northfield, IL: CAP, 2012.
9. Burnett RW, Westgard JO. Selection of measurement and control procedures to satisfy the Health
Care Financing Administration requirements and provide cost-effective operation. Arch Pathol Lab
Med. 1992;116(7):777–780.

800-323-4040 | 847-832-7000 Option 1 | cap.org 35

Notes
 325 Waukegan Road
Northfield, IL 60093-2750

cap.org
0562/4100/110/40511-1