Troubleshooting Problems With The Prothrombin Time (Protime/INR), The Activated Partial Thromboplastin Time (aPTT) and The Fibrinogen Assay
Troubleshooting Problems With The Prothrombin Time (Protime/INR), The Activated Partial Thromboplastin Time (aPTT) and The Fibrinogen Assay
Problems with the Prothrombin Time (Protime/INR), the activated Partial Thromboplastin
Time (aPTT) and the Fibrinogen Assay
Problems with an analytic method may be discovered through unacceptable performance on an external quality
assessment (EQA)/proficiency testing (PT) challenge, through the regularly scheduled quality control (QC) or, occasionally,
through changes in patient testing. Regardless of the way that a problem is identified, the troubleshooting begins with a
detailed evaluation of the QC program. When unacceptable results are reported on a PT challenge, the QC must be
evaluated when the unacceptable results are reported and for the earlier time when the testing on the PT specimen was
performed. Following are principles that will be useful in such an evaluation.
There are two general types of analytical errors, systematic and random. Systematic errors are ones that affect the test
system and results in a consistent way such that all results show a bias from the accurate result. An example of this would
be a failing reagent that is used in all reactions for this analyte. In this case, all or more than one QC value is expected to be
out of acceptable limits. Random errors affect some, but not all results, and are more indicative of a failure in precision of
the test system. An example of a random error is bubbles in liquid delivery tubing so that occasionally the wrong volume is
delivered.
As soon as an out of limits QC result is obtained all patient testing must be suspended pending identification and correction
of the problem. When an individual QC test result falls outside limits then either there is a problem with the individual vial
of QC material, the entire batch of QC has deteriorated or there is a problem with the analytical system. It is important to
differentiate among these possibilities by sequential investigations so that appropriate action can be taken. Stating the
obvious, all reagents and controls need to be examined to be certain that they have not passed their expiration dates and
that they have been continuously stored under the conditions recommended by the manufacturer.
In most laboratories Protime/INR, aPTT and fibrinogen methods are in operation on the same analyzer at the same time. In
this case scrutiny of the whole pattern of QC results for all methods may be informative. Table 1 illustrates how different
patterns of QC results can help identify the nature of the analytical problem. It is common practice, when a QC value is out
of limits, to repeat the test with the same control, then with a new vial of control. It is much better to first evaluate all of
the QC data, and pursue the problem that is indicated by all control results.
The most common cause of an out of control test system in hemostasis laboratories is reagent deterioration or
contamination. As mentioned above, all QC results will be out of limits and in this case reagents should be discarded.
Patient testing should only be safely resumed after the problem has been identified and rectified.
When dealing with control problems involving a single method, many laboratories use the Westgard multirule system for
determining acceptance of quality control values. Use of these rules can be a useful tool for guiding a troubleshooting
journey. Rules most commonly used include:
1‐3S – one control value outside of 3 SD is a cause for rejection. This rule detects large imprecision in a test
system.
2‐2S – two controls outside of 2 SD in the same direction, either within a run or 2 consecutive runs. This indicates
a bias in results and accuracy failure. If a laboratory is using a mean +/‐ 2SD acceptable limit for QC ranges, it is
important to realize that based on statistics alone, 1 in 20 or 5% of results will be outside of this range when there
is nothing wrong with the test system. Failing the Westgard 2‐2S rule indicates a high likelihood of error in the
reagent or instrument used. If indicated, analysis of a second vial of the same batch of QC with the same
reagents/method should be the first action. If this is within limits, it means that the out of range result was most
likely random error, and the test system is under control but the QC sample analyzed earlier had deteriorated or
been contaminated and should be discarded. This should be confirmed by performing a third QC test. If this is
within limits then patient testing can be safely resumed. If, on the other hand, a second out of limits result is
obtained on the replacement QC vial of the same batch, this indicates systematic error and the most likely
explanation is that the test system is out of control. In this case analysis of a second lot or batch of QC material
would also give out of limits results. If further vials of the first batch give out of limits results but the second batch
of material gives within limits results, then the entire batch of the first QC material may have deteriorated. This is
normally a rare occurrence. Thus having more than 1 lot of QC available is very useful when investigating the cause
of out of limits QC results.
In some instances there are compelling reasons to investigate a method with subsequent intervention even though recent
QC results are within the target range. For a QC sample with appropriate target limits, analyzed with a well controlled
method, the results obtained should fluctuate around the mean (Fig 1). However, a series of 7 to 10 results on the same
material showing a progressive trend in one direction (i.e. gradually increasing or decreasing over time ; (Fig 1) may be an
indication of drift in the analytical process. A gradual deterioration of a particular reagent over time, or gradual change in
an analyser could be associated with such a pattern of QC results. On other occasions there may be a series of individual
QC results all within limits but consistently lying above or consistently below the mean (Fig 1). This may indicate that there
has been a change in the method since the target range was established. In this case a new target range should be
established and investigation of recent changes in the method as well as an assessment of the accuracy of the test is
warranted (for example through external quality assessment or analysis of reference samples/materials). Where a batch or
lot number of QC is used for more than 3‐4 months a recalculation of mean and SD is useful to confirm that the original
range remains appropriate.
Figure 2 depicts similar information that can be derived from a proficiency testing evaluation reports. In this case, the
relationship of the reported data to the method mean is reported along with the same data for the past 15 reported values.
Data labeled CG1‐C 2006 is close to the mean and scattered on both sides of the mean. In contrast, those data labeled CG1‐
A and CG1‐B 2006 all are lower than the mean for the method and appear to be slightly more negative with time. Even
though all of these specimens were graded acceptable, this information is a very strong indication of a problem with the
method.
When the cause of out of limits QC results has been identified it is necessary to assess which patient results need to be
repeated. If the problem relates to the deterioration of QC material no retesting is required. If the problem relates to
reagents then all patient results obtained since the last within limits QC result should be reviewed. One practical approach
is to retest every 10th sample in reverse order to establish at which point the problem might have developed. Any patient
samples from that point onward require retesting.
Some consideration of the instrument problems is also useful. In most cases, the laboratory pursues the QC problem as
described above and, if there is no resolution; the manufacturer would be contacted for assistance with troubleshooting
possible instrument problem. Potential instrument problems that may contribute to problems are numerous, some of
which a trained technologist will be able to test. Such items may include but are not limited to:
temperature of the chamber storing reagents and specimens;
delivery volume and cleaning of the pipettes; contaminated our outdated diluent for QC material (e.g. water);
temperature of the reaction; timing of any required incubations;
for optical endpoints, wavelength of the nephelometer light source; cleanliness and output of the detector of the
nephelometer;
for mechanical endpoint measurement, the detection of the impedance of motion of the mobile; accuracy of any
instrument calculations;
instrument/computer interface and others.
Some instrument failures will affect all testing being performed while other may selectively affect only some methods.
Protime/INR
In addition to the information above, there are a couple of specific issues to consider regarding the Protime/INR. Table 2
presents the data from a PT challenge. It demonstrates that the data for this method has little variation. CLIA grading
criteria for the PT challenge states that in order to pass, the value must be within 15% of the mean for the group. In this
case, 15% variation actually represents 6.4SD from the mean. In Figure 2, many of the values for CG1‐A and B for 2006 were
Page 2
beyond 3SD of the mean, however none received a failing score. Despite passing this PT challenge, the method is clearly out
of control and corrective action by the laboratory is needed.
The INR presents a number of its own problems. In order for the INR to be accurate, the Protime in seconds must be
accurate. The Protime in seconds is then converted to the INR by the following calculation:
INR = (Patient Protime / Mean Normal Protime) ISI
Several issues are critical in the management of this calculation including:
Has an appropriate reference range for your current reagent/instrument been identified?
Is the geometric mean of this reference range being used in the calculation? Some common software packages
(such as Excel) have the capability of calculating a geometric mean.
Is the correct ISI being used in the calculation? The ISI varies not only with the lot of reagent but also with the
instrument being used. If in doubt, contact the reagent manufacturer.
When performing the calculation, the values for the patient Protime and the Protime reference range mean
include one decimal place (e.g. 12.0) and ISI should be included using two decimal places (e.g. 1.05). In each case,
three significant digits. The INR should be then rounded and reported to one decimal place (e.g. 2.5; two
significant digits).
How is the calculation being performed? Manually? By the instrument? By the LIS? The calculation should only be
performed by one (not more than one) of these methods. Having more than one method in place increases the
maintenance efforts and increases the risk of a miscalculation.
If a manual method of calculation is used, how do you minimize errors?
Is the calculation tested periodically to assure that the correct INR is being produced? In particular, this calculation
must be validated prior to reporting results if any of the components of the formula (ISI or the geometric mean of
the reference range) have changed. Actual patient reports should be periodically reviewed to assure that the
clinician caring for the patient is receiving the correct result.
Are specimens collected into 3.2% citrate ‐ not 3.8%? Published reports have documented that INR results differ
between the two‐citrate concentrations. Some laboratories continue to use 3.8% citrate despite international
efforts to standardize on 3.2%.
Activated Partial Thromboplastin Time (aPTT)
aPTT reagents have become more sensitive in recent years, leading to some difficulty with PT and patient specimens that
are prolonged. Table 3 depicts the situation in which the laboratory’s value has an SDI of 1.4, but still received an
unacceptable score because the CLIA threshold of 15% was less than 2SD. The 15% level SDI was 1.2. In this case, wide
variability of the high aPTT data would cause a frustrating unacceptable outcome.
The aPTT is a two‐stage reaction, requiring the addition of a measured reagent, the phospholipid and the activator,
followed by incubation of 2 ‐ 5 minutes, followed by the addition of calcium to initiate the reaction. The timing of these
additions is specified by the manufacturer of the reagent and needs to be controlled very closely. If the timing is not
correct, the error may not be detected by QC, but when compared to a peer group in an PT mailing, may be the cause of
unacceptable results.
Fibrinogen Assay
Unlike the Protime and aPTT that are measured as a clotting time in seconds, the fibrinogen activity is reported as mg/dL or
g/L and is a calibrated assay. The exception to this is when determined as a derived fibrinogen based on the PT result. The
same actions taken to troubleshoot the PT as described above apply to troubleshooting derived fibrinogen activities.
Fibrinogen activity is most commonly measured using the Clauss method – thrombin is added to diluted citrated plasma
and the time to clot formation is measured. A standard calibration curve is determined by testing several dilutions of
plasma and the clotting times vs. concentrations in g/L or mg/dL are plotted on a graph. Unknown samples’ fibrinogen
activity is interpolated from the standard curve. If the Clauss fibrinogen assay is run at the same time as the Protime/aPTT
assays on an analyzer, it is important to test for reagent carryover and ensure adequate washing of the reagent delivery
Page 3
probe. The thrombin concentration in the fibrinogen activity assay reagent is high, and incomplete washing or the failure of
a wash component on the analyzer can lead to shortening of the PT and PTT.
When a QC or proficiency failure occurs with fibrinogen activity, it is important to first determine if the error is random
(occurring with one level of QC), or systematic and affecting accuracy across the range of the assay. If one level of QC is
unacceptable, and repeats as acceptable, the incorrect result could be a random error (passes Westgard’s 2‐2s rule) and the
test system accurate. If one level of QC repeats as unacceptable, it could be due to a systematic error, effecting one end of
the calibration curve and not over the entire linear range. Potential sources of error to investigate and a logical
troubleshooting scheme are listed in Table 4.
Some analyzer models use a bar‐coded calibration curve specific to each lot of reagent. These systems cannot be
calibrated, and that troubleshooting step is skipped, leading directly to investigating changes in the instrument liquid
delivery system, reaction temperature, or detection system.
If more than one level of QC is unacceptable in the same direction, this indicates a bias in the test system and inaccurate
results. The first action is usually to re‐assay QC with freshly prepared reagent. Reagents may show substandard
performance if left exposed to air for a prolonged period due to a pH change or due to contamination. If this does not
correct the problem, recalibrate the assay since drift in reagent responsiveness or with the instrument over time can shift
the results away from the mean. Ensure that the correct calibrator value is entered into the system for the lot of calibrator
material in use. If the bias is not corrected with these actions, further investigation of the analyzer systems is warranted.
Comparing the new calibration curve with the old can be useful tool (Fig 3). If the calibration curves have not changed, a
problem with the control material is indicated. If the curves are parallel to each other across the measurable range,
showing a parallel bias, an instrument or calibrator material shift is indicated. If the curves intersect or demonstrate an
inconsistent bias over the linear range, this may be due to a deteriorating reagent.
Reference:
Krishnan J. Coagulation testing. In: An Algorithmic Approach to Hemostasis Testing. Kottke‐Marchant K. CAP Press,
Northfield, IL: 2008, pp 63‐92.
Kitchen S, Preston FE, Olson JD. Internal Quality Control in the Hemostasis Laboratory. In: Quality in Laboratory Hemostasis
and Thrombosis. Kitchen S, Olson JD, Preston FE, ed. Wiley‐Blackwell, West Sussex, England: 2009, pp 43‐50.
Chandler WL. Initial Evaluation of Hemostasis: Reagent and Method Selection. In: Quality in Laboratory Hemostasis and
Thrombosis. Kitchen S, Olson JD, Preston FE, ed. Wiley‐Blackwell, West Sussex, England: 2009, pp 63‐71.
Olson JD, Brandt JT, Chandler WL, et al. Laboratory Reporting of the International Normalized Ratio Progress and Problems.
Arch Path Lab Med, 2007;131(11):1641‐1647.
Mackie IJ, Kitchen S, Machin SJ, Lowe GDO. Guidelines on Fibrinogen Assays, on behalf of the Haemostasis and Thrombosis
Task Force of the British Committee for Standards in Haemotology. Brit J Haematology, 2003; 121:396‐404.
Westgard JO, Klee GG. Quality Management, Chapter 16 in Fundamentals of Clinical Chemistry, 4th edition. Burtis C, ed.,
WB Saunders Company, Philadelphia: 1996; pp.211‐223.
Page 4
Appendix
Table 1. Troubleshooting IQC problems: How different patterns of IQC error for 2 levels of IQC material tested for
Protime/INR and aPTT can direct the sequence of investigations required to correct the problem
IQC Troubleshooting
Table 2. Results of CG2‐C 2004 Dade Behring BCS Protime PT Data
CG2‐C 2004 Dade Behring BCS Protime
Specimen CG2 15p
Lab Result 19.1
Mean 19.24
SD 0.45
n 72
SDI ‐0.31
Lower Limit 18.1
Upper Limit 20.2
15% = 2.89 seconds or 6.42 SD
Page 5
Table3. Results of CG2‐C 2004 Trinity Biotech AMAX aPTT PT Data
CG1‐A 2006 Trinity Biotech AMAX aPTT
Specimen CG1 13
Lab Result 106
Mean 90.1
SD 11.2
n 16
SDI 1.4
Lower Limit 76
Upper Limit 106
Graded Unacceptable ‐ 15% limit: 104
Table 4. Troubleshooting guide for fibrinogen activity assays
Source of error Corrective Action
Pipette used to reconstitute QC Check pipette calibration for accuracy and precision
Contaminated diluent Use fresh diluent
Deteriorated reagent Prepare fresh reagent
Shift in calibration curve due to instrument wear or
drift Recalibrate ‐ Troubleshoot the analyzer’s liquid delivery
system, temperatures and detection system
Page 6
Figure 1.
An example of Levey‐Jennings charts recording control values.
This shows the expected distribution and examples where problems have occurred indicating that a method is moving out
of control. In each case the dotted lines show the 2 standard deviation limits of acceptable results.
3 3
2
Standard Deviation
Standard Deviation
1
1
Mean
Mean -1
-1 -2
-2 -3
-3 -4
0 2 4 6 8 10 12 14 16 18 20 22 0 2 4 6 8 10 12 14 16 18 20 22
Control Events Control Events
3 2
Standard Deviation
2 1
1 Mean
Mean
-1
-1
-2 -2
-3 -3
0 2 4 6 8 10 12 14 16 18 20 22 0 2 4 6 8 10 12 14 16 18 20 22
Control Events
Control Events
Page 7
Figure 2.
Longitudinal report of the deviation of results from the mean for the method from the College of American Pathologists PT
Report.
Figure 3.
Example of fibrinogen activity calibration curve comparison. The calibration curve in blue (diamonds) is the accurate curve.
The pink curve (squares) shows a shift in instrument performance with a parallel bias over the measurable range as could
be seen with an instrument change. The green curve (triangles) demonstrates a non‐parallel curve that may indicate
deterioration of reagent.
60
50
clotting time (secs)
40
30
20
10
0
100 200 300 400 500 600 700 800 900 1000
Fibrinogen (mg/dL)
Page 8
Protime/INR, aPTT and Fibrinogen Assay Toolbox
Protime/INR more than Problem with the Check out‐date of the reagent Assure that all reagents are within date
one control out of reagent used prior to use
control range Check the time the reagent was opened
Assure that all reagents are stored and
Check Levy‐Jennings to for evidence of a trend used according to SOP and package insert
or shift
Evaluate the Control Rules (Westgard
Repeat reagent, if within range, repeat again Rules) used to manage the test
If out of range, use a new vial of reagent
Protime/INR, aPTT and Fibrinogen Assay Toolbox
aPTT more than one Problem with the Check out‐date of the reagent Assure that all reagents are within date
control out of control reagent used prior to use
range Check the time the reagent was opened
Assure that all reagents are stored and
Check Levy‐Jennings to for evidence of a trend used according to SOP and package insert
or shift
Evaluate the Control Rules (Westgard
Repeat reagent, if within range, repeat again Rules) used to manage the test
If out of range, use a new vial of reagent
Page 2
Protime/INR, aPTT and Fibrinogen Assay Toolbox
Fibrinogen Assay more Problem with the Check out‐date of the reagent Assure that all reagents are within date
than one control out of reagent used prior to use
control range Check the time the reagent was opened
Assure that all reagents are stored and
Check Levy‐Jennings to for evidence of a trend used according to SOP and package insert
or shift
Evaluate the Control Rules (Westgard
Repeat reagent, if within range, repeat again Rules) used to manage the test
If out of range, use a new vial of reagent
Page 3
Protime/INR, aPTT and Fibrinogen Assay Toolbox
One control out of range Problem with the Check out‐date of the control Assure that all controls are within date
for more than one control prior to use
analyte (e.g. both Check the time the control was opened
Protime/INR and aPTT Assure that all controls are stored and
out of range) Check Levy‐Jennings to for evidence of a trend used according to SOP and package insert
or shift
Evaluate the Control Rules (Westgard
Repeat control, if within range, repeat again Rules) used to manage the test
If out of range, use a new vial of control
Page 4
Protime/INR, aPTT and Fibrinogen Assay Toolbox
Page 5