Maintaining consistency of results over time is a challenge in laboratory medicine. Each change in reagent lots can adversely affect the consistency and quality of patient results. Good laboratory practice requires evaluation of each new reagent lot prior to use. Assuring lot-to-lot consistency is particularly important for those analytes that are serially measured over time such as hemoglobin A1c, TSH, INR, tumor markers, and liver enzymes.

Manufacturers try to minimize reagent lot-to-lot variation but multiple factors can affect performance of a new reagent. Examples include minor changes in reagent contents, reagent instability, reagent deterioration during transportation or storage, and inaccurate calibration.

Performance of a new lot of reagents should be compared against the existing lot before it is depleted. Three to five patient samples with low, mid and high analyte concentrations should be run with both the current and new lots of reagents. There are no universally agreed upon criteria to accept or reject new reagent lots. The medical director must determine what is acceptable. Factors to consider include method imprecision and clinically significant difference. For most analytes, a difference between reagent lots of less than 10% is acceptable.

Although patient samples are preferred for checking lot to lot variability, other alternatives are permissible.

1. Reference materials or QC products provided by the method manufacturer with method-specific and reagent-lot–specific target values.
2. Proficiency testing materials with peer-group–established means.
3. QC materials with peer-group established means from at least 10 laboratories.
4. Third-party general purpose reference materials if the material is documented in the package insert or by the method manufacturer to be commutable with patient specimens for the method.
5. QC material used to test the current lot is adequate alone to check a new shipment of the same reagent lot, as there should be no change in potential matrix interactions between the QC material and different shipments of the same lot number of reagents.

Reportable range is the functional range of an assay over which the concentrations of an analyte can be measured with acceptable accuracy and precision. Reportable range should not be confused with reference range. Reportable range includes analytical measurement range (AMR) and clinically reportable range (CRR).

AMR is defined as the range of values an instrument can report directly without dilution or concentration. Accreditation agencies require that AMR be validated at least every six months and after recalibration, changes of reagent lots or major instrument maintenance. Validation can be accomplished using at least three levels of commercial linearity materials, standards or calibrators that span the lower limit, mid-point and upper limit of AMR. AMR validation can also be done by calibration verification (see below).

If material is not available to validate AMRs exact upper limit, it is recommended to use calibrators within 10% to 15% of the ends of the AMR. In this situation, the medical director must write a statement documenting the highest range validated. If patient results are higher than the material used to verify AMR, but less than the stated AMR, they can be reported without dilution. For example, if the AMR is initially determined to be 0 - 1000 ng/dL, but the material available for validation only goes up to 900 ng/dL, patient results between 900 and 1000 ng/dL can be reported. Patient results higher than 1000 ng/dL should be reported with a > sign or diluted before reporting. The medical director must specify and document what the maximum dilutions are for each analyte. Maximum dilutions should be based on medical usefulness.

CRR is the range of values an instrument can report as a quantitative result with dilution or concentration. CRR is typically wider than AMR. Values greater than or less than the CRR are reported as greater than (>) or less than (<). CRR is a clinical decision made by the medical director and does not require periodic revalidation. Dilution or concentration protocols must be documented for each analyte.

The following table demonstrates the relationships of reference range, AMR and CRR for some selected chemistry analytes measured on a Vitros 5600 chemistry analyzer.

CRR should always be wider than the reference range.

Functional sensitivity (Limit of Quantification) is the lower limit of an analyte that can be reported with acceptable accuracy and precision. Functional sensitivity is important for analytes such as troponin. Several international expert panels including the European Society of Cardiology, the American College of Cardiology, and the American Heart Association have recommended that increased troponin be defined as a value above the 99^th percentile concentration of a healthy population as long as total imprecision is <10%.

CLIA 88 requires that reportable range be verified only during method evaluation, but CAP requires that it be verified periodically. This requirement can be met in several ways; by doing a linearity study, performing a correlation study that includes low and high samples spanning the manufacturer’s range, serially diluting a high patient sample, running 20 repetitions of zero standard or a very low patient sample to determine the LLD, and determining acceptable range of recovery for low, medium, and high standards.

Once the method evaluation is completed and reviewed, the laboratory manager and medical director should sign a coversheet documenting that all of the required evaluation parameters that have been completed.  An example is included below.

Assay:

Instrument:

Live Date: __________________________

The following parameters have been completed before the go-live date for this assay:

• Clinical application
• Accuracy
• Precision
• Linearity
• Sensitivity
• Interference
• Method comparison
• Quality control
• Reference range/ unit of measure
• Reportable range
• LIS report format
• Interface
• Cost accounting-direct/indirect cost
• CAP and/or TJC notification – add assay/analyte to activity menu
• Proficiency survey ordered or Alternate Proficiency organized
• CPT coding
• Written procedure/s; approved by Medical Director
• Staff training/competency; paperwork placed in personnel files

Complexity -Non-Waived or Waived

FDA status approved or on-approved

This validation study has been reviewed and the performance of the method is considered acceptable for patient testing.

Manager Signature______________________________________________ Date________________

Medical Director Signature_______________________________________ Date_________________

Intermittent Testing: Assay Name

PT or alternative assessment performed within 30 days prior to restarting patient testing

Method performance specifications verified, as applicable, within 30 days prior to restarting patient testing

Validation of a qualitative test differs from a quantitative test. Precision testing is required for qualitative tests that derive a qualitative result, such as negative or positive, from a quantitative value such as optical density. Precision testing is also required if the manufacturer's package insert provides precision specifications. Within run precision can be accomplished by running at least 20 replicates of negative control and 20 replicates of positive control in a single run. Between run precision involves running negative and positive control at least once per day for 20 days. Precision results are used to calculate the %CV for both the positive and negative controls. Precision is acceptable if the calculated CVs are less than or equal to the manufacturer's stated CV.

Accuracy of a new qualitative method is assessed by comparison to a method already in use in the laboratory or at a reference laboratory. Another possibility is to test samples with known values, such as proficiency test samples or commercial standards. A minimum of 10 samples for each expected result should be tested. For example, if a test gives results of positive or negative, accuracy studies must include 10 known positive and 10 known negative samples.

Accuracy can be assessed by calculating sensitivity and specificity using a contingency table.

1. Calculate the estimated Diagnostic Sensitivity(True positive rate) = 100 x [TP/(TP+FN)]
2. Calculate the estimated Diagnostic Specificity(True negative rate) = 100 x [TN/(FP+TN)]
3. Calculate the percent Positive Agreement (Positive Predictive Value)=100 x TP/(TP+FP)
4. Calculate the percent Negative Agreement (Negative Predictive Value) =100 x TN/(TN+FN)
5. Compare the results calculated above with the manufacturer’s stated claims for Sensitivity, Specificity and Agreement found in the test kit package insert.
6. Results must be equal to, or greater than, the manufacturer’s claims for the method to be considered accurate.

Qualitative tests do not require linearity, AMR or reference range studies. For an FDA approved, unmodified method, the laboratory can accept the manufacturer's interference claims.

Calibration is the foundation of all clinical laboratory testing that insures the accurate reporting of patient results. Calibration is the process that links the analytical signal with the concentration of analyte present in serum, urine or other body fluid.

Calibration uses a series of at least five calibrators containing known concentrations of an analyte. Before beginning calibration a medical laboratory scientist programs the instrument with the concentration of each analyte according to the information provided on the package insert supplied with the calibrator kit. The instrument then measures the calibrator and adjusts the signal to match the given values. Depending on the method, this signal might be potentiometric, photometric, fluorometric, chemiluminescent, nephelometric or turbidimetric. Plotting signal on the Y-axis versus analyte concentration on the X-axis creates a calibration curve. The purpose of a calibration curve is to establish the relationship between the concentration of an analyte and the magnitude of the signal given by the measuring device. The relationship can be linear or nonlinear.

A calibration curve shows the signal rising linearly with increasing concentration of analyte from the limit of detection (LOD) to the limit of linearity (LOL). This is the analytical measurement range (AMR). Beyond LOL, the line is no longer linear and the signal is no longer linearly related to analyte concentration. Once the calibration curve is established, the signal from a patient sample can be compared to the calibration curve to determine the concentration of analyte in the patient sample.

Calibration materials should have the same matrix as patient samples. Serum based calibrators should be used when testing patient plasma or serum, while urine based calibrators should be used for urine chemistry tests. Calibrators should be traceable to standard reference materials to insure comparable and accurate results.

Calibration should be repeated when:

• The instrument or reagent manufacturer’s instructions says it is necessary
• Every time a reagent lot is changed
• Whenever quality control results show a systematic bias
• After major instrument maintenance that can cause shifts in quality control
• When reagents have poor stability

Instruments or reagents that require frequent calibration have a much higher cost per test because of the added labor and expense of calibration kits.

Autoverification is a process for automatically verifying test results based on a predetermined set of rules established by the laboratory. Autoverification improves operational efficiency by eliminating the need for a medical laboratory scientist to approve each test result before they are released to the laboratory information system for reporting. Besides more effective use of personnel, autoverification improves turnaround time and reduces reporting errors.

In autoverification, patient results generated by an instrument interfaced to a laboratory information system are compared by computer software against laboratory-defined acceptance parameters. If results fall within these parameters, they are automatically released for reporting with no additional human intervention. Results that fall outside of these defined parameters are reviewed by a medical laboratory scientist prior to reporting.

Software rules for autoverification may reside in either the laboratory information system or in middleware. Several parameters are included in autoverification rules including instrument flags, serum interference indices, delta check, need for manual dilution, analytical measurement range (AMR), reference range, and critical range.

The following criteria must be met before a result is autoverified:

• quality control is acceptable
• Results fall within the specified autoverification range
• Results pass delta check limits
• No instrument flags are present

Common reasons that a result is not autoverified include specimen error (clot, bubble, short sample), need for manual dilution, instrument error, interference index flag, and value outside the AMR.

One issue that complicates chemistry autoverification is the presence of method interferences. The LIS must be able to capture all instrument error flags and use them to prevent autoverification.

College of American Pathologist’s general lab checklist has several questions regarding autoverification. These concern monitoring quality control, suspension of autoverification, rules-based checking, rules validation, and medical director oversight. CAP requires that the medical director sign a policy approving the use of autoverification procedures. 

Delta check is a process to detect discrepancies in patient test results prior to reporting by comparing current patient values to previous ones. Delta check limits define the allowable difference between consecutive results for a specific analyte on the same patient within a certain time interval. Delta check limits should be set so that true changes in patient test results are not flagged but improbable changes are flagged as delta check failures. Delta check limits should be based upon the total expected variation, which include biological and analytical variation (see Appendix A for biological variation).

Delta check limits can be expressed as the absolute or the percent difference between two consecutive results. Absolute limits are calculated as the difference between the larger and smaller result. Percent delta limits are calculated as the difference between the larger result and the smaller result divided by the smaller result. Time interval is the specimen collection time difference between the current and previous results. Time interval is flexible. Most hospital laboratories choose 24 or 48 hours.

Delta checks are recommended for inpatient testing. Generally, you want to select chemistry analytes that have the lowest biological variation. Below is an example delta limits for some common chemistry analytes.

To detect specimen mix-ups in hematology, delta checks should be applied to parameters that show the least short-term biological variation. MCV and MCHC are extremely stable in a patient over a short interval, such as 24 hours. The diurnal biological coefficient of variation in MCV is only 0.5%. Even in medical situations where other hematological parameters are changing rapidly, such as hemorrhage, MCV and MCHC do not change significantly since the reticulocyte response does not begin for two to three days. MCHC has the added benefit of detecting instrument malfunction because it is calculated from hemoglobin, MCV and RBC count.  These three parameters are directly measured. Suggested delta check limits for MCV and MCHC are +/-5 fl and +/-5.0 g/dL, respectively.

Delta checks are not recommended for other hematology parameters including hemoglobin, hematocrit, RBC count, WBC count or platelet count. Acute changes in these parameters are common in hospital patients and the false positive rate is unacceptably high. Some laboratories include a delta check for platelet count to catch specimens with platelet clumping.

Delta failure is an indicator of possible preanalytical error such as specimen mix-ups, patient mix-ups, or IV contamination of the specimen. When a delta failure occurs, the medical laboratory scientist should carefully review all results prior to releasing. Numerous delta failures on the same analyte may indicate an analytical problem with the instrument and should be investigated immediately.

Clinical correlation plays an important role in interpreting delta failures. For example, creatinine results in patients with renal failure rise and fall depending on their dialysis schedule. Patients receiving contrast media for imaging often have elevated creatinine values postprocedure that return to normal within a few days.