One way to characterize the amount of patient result error that can be tolerated is the use of total allowable error (TEa) limits. If the error in a patient's result is less than the specified TEa, the result is considered acceptable for its intended use. Conversely, if the error in a patient's result is greater than TEa the result is considered unreliable.2
If TEa limits have not been specified it is hard to determine whether a given set of patient results are reliable. Test method performance (imprecision and bias) coupled with TEa limits dictate the capability of a test system to tolerate error. The ratio of TEa to analytical imprecision (or when bias is considered, (TEa - bias ) / imprecision) has been referred to as the sigma metric of the process.3 Test methods with low sigma metrics have lower error tolerance than high sigma test methods.
What to Do With Unreliable Results
Unreliable patient results produced during the existence of an out-of-control condition can be divided into two intervals-those generated before the last acceptable QC event (pre good QC) and those generated between the last acceptable QC event and the QC rejection (post good QC). Based on the TEa specification, test method performance, power of the QC rules employed and the frequency of QC testing, the expected number of unreliable patient results pre and post good QC can be estimated for any possible size out-of-control condition.
Fig. 1 illustrates an analytical process where the number of unreliable results generated prior to the last good QC is about twice as large as the number of unreliable results generated after the last good QC.
The vertical strokes represent patient specimens being examined, the green diamonds represent acceptable QC events and the red diamond is a QC rejection. The asterisk (*) represents an unreliable patient result. The vertical shift in the analytical process represents a persistent error condition that is resolved after it was detected by the QC rejection.
Note that in Fig. 1, it takes three QC events to finally reject and identify the error. While unreliable patient results were produced during the out-of-control condition prior to the last accepted QC event, a significant amount of time has likely passed, making them difficult to identify and correct. For all practical purposes, patient results produced prior to the last accepted QC event are final. However, the unreliable patient results produced between the last accepted QC event and the QC rejection are recent and should be regarded as correctable. Patient results produced after the last accepted QC need to be evaluated, possibly re-examined, and corrected if necessary. These are also the results addressed in CLIA §493.1282(b)(2). Limiting the number of correctable unreliable patient results is critical to recovery.
It is possible to predict the expected number of unreliable results produced for any magnitude of out-of-control condition.4 An example is plotted in Fig. 2.
Small out-of-control conditions are difficult to detect; they may persist for a long time before they are finally identified and corrected. As a consequence, most of the unreliable patient results produced during the existence of a small out-of-control condition will have been reported prior to the last acceptable QC. Fortunately, because the out-of-control condition is small, the number of unreliable patient results produced is relatively small.
Large out-of-control conditions can produce a large number of unreliable patient results. However, large out-of-control conditions are relatively easy to detect. Therefore, for large out-of-control conditions, virtually all of the unreliable patient results are correctable because they were produced between the last acceptable QC and the QC rejection. Fig. 3 is an example of an analytical process during a malfunction that creates a large out-of-control condition.
When a very large out-of-control condition occurs, half the number of patient results tested between QC events can be expected to be unreliable.5 It is prudent to make sure that your laboratory would not be overwhelmed by such an occurrence.
Dr. Parvin is manager of Advanced Statistical Research; John Yundt-Pacheco is Scientific Fellow; and Max Williams is Global Scientific and Professional Affairs Manager, Bio-Rad.
- US Centers for Medicare & Medicaid Services (CMS). Medicare, Medicaid, and CLIA programs: laboratory requirements relating to quality systems and certain personnel qualifications. Final Rule. Fed Regist Jan 24, 2003;16:3640-714.
- Parvin CA, Yundt-Pacheco J, Williams M. The focus of laboratory quality control: Why QC strategies should be designed around the patient, not the instrument. ADVANCE for Administrators of the Laboratory 2011;20(3):48-9.
- Westgard JO. Six sigma quality design & control, 2nd ed. Madison WI: Westgard QC Inc., 2006.
- Parvin CA. Assessing the impact of the frequency of quality control testing on the quality of reported patient results. Clin Chem 2008;54:2049-54.
- Parvin CA, Yundt-Pacheco J, Williams M. Designing a quality control strategy: In the modern laboratory three questions must be answered. ADVANCE for Administrators of the Laboratory 2011;20(5):53-4.