Line backers—support grows for linearity Surveys
November 2002 Karen Southwick
The CAP’s linearity Surveys for quality assurance
have proved so effective that a recently proposed NCCLS guideline
is largely modeled on the College’s approach.
NCCLS guideline EP6-P2, "Evaluation of the linearity of quantitative
analytical methods; proposed guideline—second edition," which
was published in December 2001, awaits full approval. "They’re using
essentially the same method as we’re doing for evaluating linearity,"
says Martin Kroll, MD, director of clinical chemistry at the Dallas
VA Medical Center and associate professor at the University of Texas
Southwestern Medical School. Dr. Kroll also co-chairs the CAP Standards
and Instrumentation Resource Committee, which oversees what are
commonly referred to as the LN Surveys.
The CAP offers 16 LN Surveys, and three more will be added for
2003. They help laboratories check the linearity of their instruments.
But they can also be used to determine analytical measurement range
and to verify calibration, thus satisfying CLIA as well as CAPinspection
checklist requirements for accreditation.
Though linearity checks are not required under CLIA, they are
an important part of overall quality assurance, Dr. Kroll says.
"When you’re talking about linearity, you’re talking about the whole
range [of values]," not just a midpoint.
"Linearity evaluations," he adds, "are the only ones that give you information
about what’s occurring near the extremes," the high and low ends of the measurement
range. Measuring those values accurately can be important in diagnosing or monitoring
disease. "This is one of the strongest quality assurance tools you can use in
the lab," Dr. Kroll says. "I don’t see how you can run a lab in good conscience
and not do it."
Skeptics abound
When the CAP first offered the LNSurveys about a decade ago, not
everyone was convinced of the need to check linearity. David A.
Floering, MD, a member and later chair of what was then the CAP
Instrumentation Committee, was skeptical. "A lot of people didn’t
see the purpose of this, including me," he says. "In Ohio, we weren’t
required to do this." Dr. Floering is a pathologist at Middletown
(Ohio) Regional Hospital.
In the 1980s, the state of Florida was asking laboratories to
perform linearity checks, but there was no readily available material
with the right matrix and analyte concentration to perform the linearity
and establish a reportable range. "There was an absolute lack of
material for this purpose," says Tony Vazquez, MT(ASCP), who was
then technical director of laboratories at Cedars Medical Center,
Miami. "You had to go through all sorts of contortions to try to
come up with a product. In some cases a high enough patient sample
had to be used. Checking for linearity and reportable range is an
integral part of method validation."
Because he was a consultant to the CAP committee, Vazquez turned
to the committee members. "It took several years to get the LN Surveys
off the ground," he says. "Some members didn’t see why there would
be a need for this." In fact, the number of subscribers was unexpectedly
high. "No question there was a need," he says.
Dr. Floering agrees. "When I started using the LNSurveys in my
own lab, I found them very helpful in evaluating how a new test
was running," he says. "I evolved from skeptic to champion." The
CAP developed freeze-dried samples for linearity testing and analyzed
the results. "One thing that showed us we were on the right track
was that a cottage industry sprang up to provide linearity material
as a commercial enterprise," Dr. Floering says.
Furthermore, laboratories became more aware of what kind of quality
assurance would satisfy CLIAregulations. LN Surveys are particularly
useful in calibration verification, Dr. Floering notes. "By having
multiple points through the reportable range, you can learn a lot
about your assay. Many times the Survey will pick up problems before
they become noticeable in other ways. It’s like an early warning
system."
Adds Dr. Kroll: "Linearity is a property of the entire system over the entire
range. When your reagents age, when you have problems with instruments, the
first thing to go is your upper or lower limit." Laboratories performing well
on the LN Surveys, he says, tend to score better on proficiency testing.
Improving the Surveys
The LN Surveys have become more popular thanks in part to the
growth of automation and point-of-care testing and the proliferation
of new lab tests. Each time you begin using a new method or new
machine, you want to check for linearity, Dr. Kroll says.
After analyzing the results for an individual laboratory, the
CAP compares the results to those of a peer group culled from the
2,000 Surveys subscribers. "You have a material that many of the
other labs are also running, so you can compare yourself with others,"
says Dr. Floering. The comparison helps pinpoint whether problems
stem from method or machine.
The CAP has made its LN Surveys more user-friendly with regard
to reporting results and handling material. In 1999, the College
moved from freeze-dried to liquid samples, reducing the possibility
of preparation error.
The labs had to reconstitute freeze-dried samples, Dr. Floering
says. This led to matrix effects and possible user error in reconstituting
the sample, yielding poor results. "Any time you have to add something,
there’s a propensity for error," Vazquez says. The liquid-based
samples are ready to use.
Dr. Kroll recalls a persistent case in which a particular dry-slide
technique kept performing poorly on the LN Surveys: "When we moved
from freeze-dried to liquid-based linearity specimens, those problems
disappeared."
The CAP has also made its analysis easier to interpret. "We’ve
been trying to simplify it so you don’t have to be a statistician
to understand it," Vazquez says. "You can have the best Survey in
the world but laboratories won’t use it if they don’t feel comfortable
with it." Although medical technologists typically run the linearity
Surveys, other professionals, such as nurses, can use them.
Generally, more than 90 percent of Surveys participants pass all
their methods, Dr. Kroll says. Problems can range from old reagents
to a defect in the instrument itself. That’s where the CAP’s peer-group
comparison is so valuable, he says. "If everybody is coming up with
non-linear results, the manufacturer probably needs to deal with
it. If 95 percent are linear and you’re not, you have a problem
with your assay system."
Failure commonly results from entering values incorrectly and
imprecise results at the extremes. "Linearity helps you find out
if you have an imprecision problem," Dr. Kroll says. About one to
five percent of participants using any given method experience imprecision
problems, he adds.
Another advantage of repeating LN Surveys is that this reduces
the possibility of false-negatives. One nonlinear result could be
due to chance or user error. "But if you’re nonlinear twice, you
probably have a problem," Dr. Kroll says. For example, the false-negative
rate for one LN Survey is 20 percent; with two Surveys, it drops
to less than four percent, and with three to 0.8 percent.
Karen Southwick is a writer in San Francisco. For more information
on the linearity Surveys, contact Sharon Burr at the CAP at 800-323-4040
ext. 7417 or at sburr@cap.org.
To learn more about the upcoming changes to the CAPcalibration/calibration
verification accreditation requirements, visit www.cap.org.
|