Ronald D. Feld, Ph.D., Marian Schwabbauer, Ph.D., CLDir(NCA)*, John D. Olson, M.D., Ph.D.
1) Describe the requirements of Subpart K of the Feb. 28, 1992, CLIA regulations that cover QC.
2) Develop a QC policy for a physician office lab.
3) Describe calibration verification and document successful verification.
4) Develop a QC program for qualitative and quantitative tests.
5) Develop a remedial action program which addresses QC failures.
6) Recognize developing QC problems of quantitative tests.
7) Logically troubleshoot a QC problem.
Latest Revision (March 11, 2003): Final CLIA rules relating to quality control were published in the January 24, 2003 Federal Register (http://www.phppo.cdc.gov/clia/pdf/CMS-2226-F.pdf). These rules take effect on April 24, 2003 but labs will have one 2-year cycle to come into compliance. Changes include:
Suggested links: |
Quality Control
Quality control (QC) is a process for assessing the accuracy and reliability of a test system. In clinical laboratories this
is usually accomplished by running specimens with known values or
outcomes along with patient unknowns to determine if the analytical
system is functioning within prescribed boundaries. Quality control
is covered in Subpart
K of the February 28, 1992 CLIA regulations published in the Federal Register. The basis for any clinical laboratory
quality control program is a written QC policy. This policy should
include a description of: a) the control materials to be used with
each assay, b) frequency of use, c) required documentation as well as
criteria for the acceptability of patients results. The quality
control policy must also state that when results are unacceptable,
any patient's results that were run with those QC specimens cannot be
reported. It should identify a set of remedial actions that should be
instituted.
Since it is assumed that the majority of physicians' office laboratories will fall into the moderately complex category, the QC discussion will be limited to this level. In addition, QC rules specific to a particular discipline, i.e., chemistry, will be discussed in that particular section.
A. Moderate Complexity Tests Using FDA Approved Methods
When the original (Feb. 28, 1992) regulations were published, it was
thought that the Food and Drug Administration (FDA) would approve or
disapprove kits, instruments or test systems as being in compliance
with CLIA '88 rules after Sept. 1, 1994. Since then the FDA has
withdrawn from the CLIA approval process and left this job to the
Centers for Disease Control (CDC). As of this date, CDC has not
addressed this subject. Because of this delay, laboratories are in
compliance with CLIA '88 if they perform procedures or use
instruments that have received FDA 510K or premarket approval (PMA).
The key concept is to follow the manufacturer's instructions without modification. Modification means any change to the assay
that could affect the performance specifications.
Following the manufacturer's instructions includes:
Certain tests may be categorized in a specialty or subspecialty section, i.e., microbiology. These tests may have additional quality control requirements that must be followed. These additional requirements will be covered under the specific laboratory specialties.
As long as the laboratory is performing only unmodified moderately complex tests, this is the extent of the quality control requirements. Laboratories performing modified moderately complex or highly complex tests must conform to a much more extensive set of quality control requirements. Since less than 5% of physicians' office laboratories are classified as highly complex, these rules will not be discussed at this time. Because of the additional complexity of these rules, it is strongly advised that moderately complex laboratories not modify any manufacturers' instructions.
B. Quality Control Rules
The rules for accepting or rejecting a run should be clearly stated
in the quality control policy. For qualitative tests, the quality
control rules are simple. The positive
control must be positive and the negative control must be
negative for the patient's results to be valid.
For quantitative tests, CLIA does not set quality control limits. These limits are set by the laboratory director. They should be clearly stated in the QC policy and in the procedure in the procedure manual.
There are several methods for evaluating quantitative QC data. The
simplest and most widely used is the Levey-Jennings approach. In this
system, the control mean is determined by analyzing the material many
times over a sufficient time period in the analytical system. (The
company may have followed this procedure and published these data
points for a given lot number of control material.) These values
should be graphed to establish that they do conform to a commercially
available gaussian distribution. It is important
that these values are collected over a time period that will
incorporate all of the variances that could affect the precision of the assay. These include but are not limited to
personnel, instrument calibration and maintenance, reagent changes,
and bottle-to-bottle variation of control material. The data are then
represented on a 
Levey-Jennings chart which represents quality control values on the y-axis and time
on the x-axis. The chart has horizontal lines at the mean value and at the control limits, usually 2 or 3 standard
deviations.
Many laboratories choose 2 standard deviations on either side of the mean for their QC limits. Since 2 SD on either side of the mean in a normal distribution encompasses 95% of the values, the control will be out of limits 1 out of 20 determinations by chance. In setting quality control limits, the medical necessity and precision of the assay should also be considered, rather than choosing a single limit for all assays. Each assay should be evaluated with regard to how it is being used in the clinical situation.
Another method for QC data evaluation was developed by Dr. James Westgard of the University of Wisconsin. This system, also known as a multirule approach, does not rely on a single QC limit to determine if a run is valid but rather uses a series of rules or algorithms to determine the validity of a run.
QC charts of a system are useless unless they are reviewed frequently by an experienced laboratorian. Review of QC data will help spot problems before the assay is adversely affected by exceeding QC limits.
Trends and 
shifts often require intervention before QC limits are breached. Outliers
indicate that QC limits have been compromised and remedial action is
necessary. Troubleshooting QC problems is a lot like establishing a
diagnosis in a patient. A good history is always the starting point.
The more information one has, the easier it is to identify the cause
of a QC problem. Data logs such as the dates and extent of instrument
calibration, reagent lot changes, and other factors that might affect
the assay must be maintained. Then to troubleshoot, find the date
when the problem first appeared from the QC charts and try to match
it to something from the procedure logs. For example, if a shift
happened following calibration, recalibrate the assay to see if the
original mean is regained. It is vitally important that all pertinent
information is maintained in a data log and that QC charts are
actively reviewed. Early intervention can forestall more serious
problems later. C. Remedial Action
The Health Care Financing Administration (HCFA) and inspectors are
very insistent on remedial action. The laboratory must have a written
plan to follow when equipment malfunctions or a QC problem is
identified, precluding the reporting of patients' results. As with
much of CLIA, documentation is the key. If it is not written down, it
did not happen.
If QC limits established by the director are compromised, patient test results in the unacceptable run or since the last acceptable test run must be evaluated and remedial action must be taken to ensure the reporting of only accurate and reliable results. Although CLIA requires that controls be run only once a day, controls should be run more frequently for high volume tests. This ensures that large numbers of patient results will not need to be held and repeated if QC or instrumentation fails. Remedial action involves troubleshooting the assay system using appropriate QC data, fixing the problem, and then repeating the patient results during an acceptable analytical run. The easiest first step is to simply repeat the analytical run. If the QC is now within acceptable limits, patient results may be reported. If QC is still unacceptable, the next steps may include instrument recalibration, new controls, new reagents, instrument maintenance, and/or changing pipettes. Whatever it takes, the assay must be brought back into control before patient results can be reported.
It is extremely important that all remedial action steps be clearly documented and kept so an inspector can determine what steps were taken to resolve the problem. All instrument maintenance recommended by the manufacturer must be performed and documented on the recommended schedule with the person performing the maintenance identified.
If patient results are delayed due to an assay problem, the laboratory must notify the appropriate individual based on the urgency of the patient tests requested. For example, a glucose result in a diabetic experiencing acidosis with coma would warrant immediate notification of a delay while a creatinine in an ambulatory patient with no history or signs of renal failure might require no notification of delayed testing.
If patient results have been released and then a problem is identified, the person ordering or utilizing the tests must be promptly notified. This must be scrupulously documented. A corrected report must be promptly issued as soon as the problem is rectified and the patient result is repeated. Exact copies of the original and corrected report must be maintained for two years.
Glossary
Accuracy - Agreement between the analyte value
and its true value. The true value of an analyte may be obtained by
weighing pure analyte, by determination using the best analytic
method available, or by consensus.
Calibration and calibration verification - Calibration is defined as a procedure which uses known quantities of analytes to adjust the analytical system to ensure that results will be accurate. There are two types of calibrators, primary or secondary. Primary standards are usually aqueous-based and consist of weighed-in amounts of pure materials. Secondary calibrators are usually serum-based and the analyte concentrations less precisely determined by running many times against a former secondary calibrator or a primary calibrator. Some determinations such as enzymes may not use calibrators but the molar absorptivity of the product such as NADH or PNP. These values are usually entered when the instrument is installed.
Calibration verification grew out of the original regulation on linearity testing. Calibration verification consists of running at least three standards every six months or more frequently if the manufacturer's instructions so state.
This is especially important in those assays that utilize a single point standard or are calibrated using the extinction coefficient of a product or substrate. There are numerous commercial sources for calibration verification materials and it is best to check availability with reagent or instrument vendors.
The data are best presented in a graphical form with expected values on the x-axis and recovered values on the y-axis. The regulations state that values can only be reported that are no higher or no lower than the highest and lowest verification standard. The limits of acceptability of this data are not stated in the regulations. It is the prerogative of the laboratory director to state the acceptable limits in the QC manual. An example might be to accept a deviation of 5% of the observed value from the expected value.
Gaussian distribution - A symmetrical bell-shaped curve whose shape conforms to a mathematical or normal equation. In this distribution, approximately 95% of the values generated are included in the area bounded by two standard deviations on either side of the mean.
Mean - The arithmetic average of a group of values. This is determined by summing the values and dividing by the number of values.
Medical necessity - The precision required of an assay in order to produce medically useful values. This is different for every analyte and depends on the clinical situation in which that analyte is being used. Biological variation must be considered. For example, an analyte which has a large biological day-to-day variation can have larger QC limits with a precise assay since most of the variance is in the patient rather than the laboratory.
Precision - The agreement of replicate values.
Shift - The control value runs consistently above or below the mean, indicating a shift in the distribution of control values with a new mean.
Standard deviation - A statistic which describes the dispersion about the mean. The standard deviation is related to the width of a normal curve.
Trend - A steady increase or decrease in a control value.
If no action is taken, the QC limit may be breached. Ronald D. Feld, Ph.D., Marian Schwabbauer, Ph.D., CLDir(NCA), John D. Olson, M.D., Ph.D.
