CentraBytes - Siemens Healthcare

with IT
Integrating Patient Moving
Averages into the QC Process
Nils B. Person, PhD, FACB and Elizabeth Kozak, MBA
Siemens Healthcare Diagnostics, Inc.
While the pivotal role of quality control (QC) in the delivery of high-quality
patient results is well recognized, the opportunity to fine-tune the QC
process in order to reduce turnaround time (TAT) and improve efficiency
is sometimes overlooked. This tutorial reviews the use of patient moving
averages (PMA) as a component of the QC program and how it can be
applied in day-to-day operations to detect problems and proactively
address issues.
What are patient moving averages (PMA)?
PMA are running averages of patient results for a specific assay over
a preset number of data points (batch size), and historically have been
commonly used in the hematology laboratory. With the computing power
made available by data management systems, PMA are increasingly used
in chemistry and immunodiagnostics as an adjunct to the QC process.
Answers for life.
Based on actual patient results,
PMA complement scheduled
analysis of control materials used
in conjunction with Levey-Jennings
charts and QC rules. With PMA,
monitoring is continuous, and
thus shifts and changes may be
detected before errors are detected
using QC results. PMA can detect
systematic changes such as
differences in assay performance
between analyzers, changes
attributable to specific reagent
lots, or issues with specimen
containers. Additionally, there
is no incremental cost associated
with reagents or the cost or time
to perform additional tests.
Audit vs. defined PMA
There are two types of PMA.
Audit or undefined PMA is passive,
continuous monitoring. PMA are
calculated and presented, but
no rules or limits are defined and
no alerts are triggered. It is the
essential first step in using PMA
in order to develop understanding
of how PMA trend. It is also useful
in troubleshooting. For example,
when a QC flag is triggered,
time-stamped, instrument-specific
PMA audit data can be retrieved
to possibly see when the problem
began and which samples may
be affected. Additionally, for some
assays, there may be shifts in QC
results when a new reagent lot
is used. These shifts are often
related to the altered sample
matrix of the QC material and do
not indicate a reagent problem.
However, the way to verify this is
to look at patient results. If PMA
Figure 1. Chloride audit in use, courtesy of University of Michigan Hospital.
show no shift in patient results,
then the shift in QC may be a
matrix issue, the reagent lot is
performing acceptably, and QC
targets need to be updated. If PMA
show a comparable shift in both
patient and QC results, there may
be a reagent problem and the
manufacturer should be contacted.
In defined PMA, QC alerts are
triggered and results held based on
preset limits. Many laboratories use
a combination of audit and defined
PMA, depending on the assay.
Chloride audit in use. In the
Figure 1 example, PMA over a
period of five hours show that
chloride results for all four
instruments are within range of
each other. The QC runs, likewise,
are within one standard deviation.
Calcium varying between shifts.
In the Figure 2 example, the
laboratory used PMA to help
troubleshoot drifts in calcium
due to pH changes caused by
a new water system.
Which tests should
be monitored?
Not all assays benefit from PMA
monitoring. Table 1 summarizes
attributes to consider when
deciding when to use PMA.
Determining batch size
and setting targets
The batch size or number of
patient results used to compute
the average should be large
enough so that it is representative
of the patient population, in order
to avoid overreaction and wasting
time in unnecessary follow-up.
Conversely, an overly large batch
may take too long to accumulate
and result in delayed detection of
issues that arise during the longer
intervals. The batch size differs
by method, and determining the
optimal batch size will require a
process of trial and error. There is
guidance in the literature on how
to determine appropriate batch
size for many assays.1 The same
rationale holds true for defining
the acceptable range.
Table 1. Assays vs. suitability for monitoring
Stable assays: day-to-day
Inherent instability can make it difficult to
for a single patient, over time isolate problems
for a patient population
Reasonable analytical range
Inherent broad range (e.g., CA 125, CK, or ALT) may
challenge data collection
Significant volume
Sufficient data can be gathered in a reasonable
amount of time to detect shifts and trends
Target value and deviation
can be established for a
patient population
Once target value (typically the average of the PMA
values obtained from weeks or months of auditing can
be used) and acceptable deviation are set, the system
can be configured to flag “out of control” batches
Figure 2. Calcium varying between shifts, courtesy of University of Michigan Hospital.
Integrating PMA into your
QC process
The data management system
plays an important role in
integrating PMA seamlessly into
day-to-day laboratory operations.
For example, based on a defined
acceptable range, the CentraLink
Data Management System can be
configured to notify the user when
a possible shift in performance
has occurred. The notification,
including quality severity (QS),
is displayed on the Patient Review
screen. The QC screen can readily
be assessed from the navigation
screen to provide additional
information for troubleshooting.
Probable causes may be a change
in patient population, a calibration
issue, an instrument issue, or a
reagent issue.
Once on the QC screen, the user
has the ability to view the assay
across all instruments on the
network (local or remote) to
further troubleshoot the problem
or determine where the assay can
be diverted to minimize impact on
TAT and avoid running assays on
the instrument with the issue,
saving reagents. When integrated
with a Siemens Aptio™ Automation
System, the CentraLink system
reroutes automatically.
The CentraLink system also
provides a means to hold for
review a set of assays based on
one of the assays in the set failing
QC. Examples are in hematology
(e.g., if RBC fails, all CBC tests
could be configured to be held for
review) and for calculated results
derived from multiple tests.
It should also be noted that
the CentraLink system uses a
moving average method based
on a variation of Bull’s algorithm.
The batch size, N, is defined
for each assay configured on a
specific instrument. The PMA is
an exponentially weighted average
of the previous N-1 patient results.
The previous average contributes
Nils B. Person, PhD, FACB, Senior Clinical
Consultant, Siemens Healthcare Diagnostics,
is a board-certified clinical chemist with over
30 years’ experience in laboratory medicine.
He spent 15 years directing hospital laboratories
prior to joining Siemens and has spent the last
18 years supporting Siemens technical staff and
customers. In his current role, Dr. Person provides
support for clinical issues that cross diagnostic product lines. His
particular areas of expertise are quality control, method evaluation
and verification, and laboratory regulatory compliance. Dr. Person
has also been part of a number of Clinical and Laboratory Standards
Institute (CLSI) Standards development teams and was involved
in the development of EP23 Laboratory Quality Control Based on
Risk Management and EP26 User Evaluation of Between Reagent
Lot Variation. He is currently involved in the revisions of EP21 Total
Analytical Error and C24 Statistical Quality Control for Quantitative
Measurement Procedures.
to the calculation of the current
batch average. As in Bull’s
algorithm, all patient samples are
included, not just the normal ones.
Patient moving averages (PMA)
can be a valuable adjunct to the
QC process, allowing tighter
control of assay performance,
faster and better responses to
issues, and cost savings on QC
material and tech time. A clear
vision of goals and methodical
planning will help guide proper
use of PMA to focus on issues that
need to be addressed and avoid
data overload.
1.Westgard JO, Smith FA, Mountain PJ, Boss
S. Clin Chem 1996; 42:1683-1688.
Elizabeth Kozak, Global Portfolio and Product
Manager, CentraLink Data Management
System, Siemens Healthcare Diagnostics, has
an MBA and a BS in computer science and over
25 years’ experience working in the IVD industry.
She spent the first 10 years of her career in
software development and project management.
In her current role as Siemens CentraLink system
global product manager, Elizabeth works with Siemens customers
around the world to understand their informatics needs and
collaborates with Siemens teams to provide the right product to
the customer. She contributes to defining Siemens IT strategies
and roadmaps and leads a cross-functional team that monitors
and supports CentraLink system products. Elizabeth believes that
in today’s competitive and consolidating clinical laboratory
marketplace, labs must operate with Lean workflows and processes
in order to survive. Informatics is customers’ biggest lever to
streamline workflows, improve processes, meet turnaround time
requirements, and facilitate growth with minimal investment.
According to Elizabeth, today’s clinical laboratories are rich in data,
but products like the CentraLink system enable customers to focus
only on the information that needs immediate action and to
proactively address potential issues.
Siemens Healthcare Diagnostics, a global leader
in clinical diagnostics, provides healthcare
professionals in hospital, reference, and physician
office laboratories and point-of-care settings
with the vital information required to accurately
diagnose, treat, and monitor patients. Our
innovative portfolio of performance-driven
solutions and personalized customer care
combine to streamline workflow, enhance
operational efficiency, and support improved
patient outcomes.
Aptio, CentraLink, and all associated marks are
trademarks of Siemens Healthcare Diagnostics Inc.
All other trademarks and brands are the property
of their respective owners.
Product availability may vary from country to
country and is subject to varying regulatory
requirements. Please contact your local
representative for availability.
Order No. A91DX-CAI-140962-GC1-4A00
08-2014 | All rights reserved
© 2014 Siemens Healthcare Diagnostics Inc.
Global Siemens Headquarters
Siemens AG
Wittelsbacherplatz 2
80333 Muenchen
Global Siemens Healthcare ­
Siemens AG
Healthcare Sector
Henkestrasse 127
91052 Erlangen
Telephone: +49 9131 84-0
Global Division
Siemens Healthcare Diagnostics Inc.
511 Benedict Avenue
Tarrytown, NY 10591-5005