[ Analysis and Control of Variation.
John McConnell, Brian Nunnally, and Bernard McGarvey
Quality Risk Management
and Variability Reduction
John McConnell, Brian Nunnally, and Bernard McGarvey
“Analysis and Control of Variation” is dedicated
to revealing weaknesses in existing approaches to
understanding, reducing, and controlling variation
and to recommend science-based alternatives that
demonstrably work. The objective of the column is to
combine sound science with proven practical advice.
Reader comments, questions, and suggestions will
help us fulfill our objective for this column. Please
send your comments and suggestions to column
coordinator John McConnell at john@wysowl.com.
au or managing editor Susan Haigney at shaigney@
advanstar.com.
KEY POINTS DISCUSSED
The following key points are discussed:
• Quality risk management (QRM) is a key part of
quality by design.
• Guidance for QRM is provided by ICH Q9.
• QRM is intimately linked with the pharmaceutical
control strategy for a product.
• Risks represent the potential for variability in the
process or product.
• QRM can help reduce variability in the following
two ways:
• By ensuring the initial control strategy does not
lead to high levels of variability.
• By helping the quality systems work better when
the process is in operation and product is being
delivered to the patient.
• Just as there will always be some level of variability
in a process or product, there will always be some
risks associated with the process or product.
[
ABOUT THE AUTHORS
For more Author John McConnell is owner and director at Wysowl Pty Ltd in Queensland, Australia. He may be
information, reached at john@wysowl.com.au. Brian K. Nunnally, Ph.D., is in charge of process validation at
go to Pfizer in Sanford, NC. He may be reached at brian.nunnally@pfizer.com. Bernard McGarvey, Ph.D.,
gxpandjvt.com/bios is process modelling group leader, Process Engineering Center at Eli Lilly and Company in Indianapolis,
IN. He may be reached at mcgarvey_bernard@lilly.com.
12 Journal of Validation T echnology [Summer 2011]
WHAT IS QUALITY RISK MANAGEMENT?
Quality by design (QbD) is a significant initiative within
the pharmaceutical industry at the present time. It is
based on the idea that most of the issues that are seen dur-
ing the manufacturing, transport, and delivery processes
for a product are built in to these processes during the
design phase (1). Quality risk management (QRM) is a
key part of this QbD approach. QRM is a process for iden-
tifying risks within a process, evaluating the significance
of these risks, and taking appropriate actions to mitigate
such risks as are deemed unacceptable. QRM recognizes
that there will always be some level of risk associated
with any process and that it is up to the pharmaceutical
manufacturer to understand the risks inherent in their
products and processes. The regulatory expectations for
QRM are outlined in a guidance document from the
International Conference on Harmonisation (2). Figure
1 shows the high level approach to QRM that is outlined
in Reference 2.
Over several years of being involved with QRM, we
have found that a critical aspect of QRM is the risk
question. The risk question defines the ultimate goal
of the risk analysis. For QRM the risk question being
addressed is clear: What are the risks to the patient
when they use a particular product being manufac-
tured, stored, shipped, delivered, etc. using a particular
set of processes?
When answering this question, there are two guiding
principles within QRM (2), as follows:
• The evaluation of the risk to quality should be based
on scientific knowledge and ultimately link to the
protection of the patient.
iv thome.com
 John McConnell, Brian Nunnally, and Bernard McGarvey.
• The level of effort, formality, and documentation
of the quality risk management process should be
commensurate with the level of risk.
QRM is a proactive approach because we try to
anticipate and manage the risk of issues before they
actually occur.
CONTROLLING
VARIABILITY IN A PROCESS
In any manufacturing system, variability is managed by
proactively defining the controls that must be in place
to manage the variability in the process to acceptable
levels. In many industries, this is referred to as a control
plan. In the pharmaceutical industry, it is referred to as
the control strategy for the product or process. Within
QbD, the formal definition of a control strategy is, “The
planned set of controls, derived from current product and
process understanding that assures process performance
and product quality. The controls can include:
• Parameters and attributes related to drug substance
and drug product materials and components
• Facility and equipment operating conditions
• In-process controls
• Finished product specifications
• And the associated methods and frequency of moni-
toring and control” (3).
Storage and Shipping Considerations
Some pharmaceutical companies add “storage and ship-
ping conditions” to the list of controls in the above defi-
nition. This is a good idea because in reality storage and
shipping are part of the overall process for delivering the
product to the patient. Unless properly controlled, storage
and shipping can introduce significant variability to the
quality of the product and hence have the potential to
increase patient safety risk. Examples of elements of the
control strategy might include the following:
• Temperature of a reaction must be controlled within
the range 75-80 °C.
• Product must be stored in a container with a relative
humidity in the range 10%-25% during shipping
• The particle size distribution (PSD) must be con-
trolled to ensure that the required product dissolu-
tion profile is met
• Operators must pass a proficiency test before they
are allowed to operate the process equipment.
CONTROL STRATEGY PROBLEMS
The control strategy is designed to manage the variability
of the product that is taken by the patient. However,
problems may still occur even when a control strategy has
gxpandjv t.com
Figure 1: QRM process outlined in ICH Q9.
Initiate Risk
Management Activity
Risk Assessment
Risk Mitigation
Ongoing Risk Review
been developed. Two problems are possible: flaws in the
control strategy or incorrectly executed control strategy.
Flaws in the Control Strategy
The control strategy may not have been correctly
designed. The following are examples:
• The control strategy states that the rotational speed of
the agitator in a reaction tank should be in the range
120-130 RPM. However, this range was defined for
a 2000-gallon reactor and the process is now being
manufactured in a 4000-gallon reactor. This caused
the reaction to proceed differently, and at the end
of the reaction an impurity level was higher than
originally seen.
• The control strategy does not specify any control of
the color of glass used to contain a liquid product.
It turns out that the product is light sensitive and
the bottle should have been specified to consist of
darkened glass.
Good Control Strategy—Not Executed
Correctly
The control strategy may have been well designed
and developed, but was not executed properly. The
following are an example:
• The control strategy states that in order to protect
the product from cross contamination, certain
valves must be used to maintain a tight seal. How-
ever, over time the valves are not maintained cor-
rectly and they begin to leak, thus introducing a
contaminant into the product.
• The pH meter used to verify the pH of a buffer is
not maintained frequently enough and so the mea-
Journal of Validation T echnology [Summer 2011] 13
Analysis and Control of Variation.
surement error of the meter increases over time and
subsequently impacts the quality of the product.
Not only is it important that the right control strategy
is in place, but also that the control strategy is executed
in a reliable fashion.
RELATIONSHIP BETWEEN
CONTROL STRATEGY AND QRM
The control strategy is concerned with specifying what
controls must be in place to manage variability. QRM
is concerned with identifying the risks to patient safety
when a product is manufactured using a given process.
These observations provide the linkage between QRM
and variability reduction. In fact, we can say that QRM
assesses and mitigates the risks associated with a con-
trol strategy where such risks, if not mitigated, would
lead to products and processes with too much variabil-
ity. Another way to say this is that risks represent the
potential for variability in the process or product. Going
further, it does not make sense to even talk about doing
a risk assessment if you do not have a proposed control
strategy. This sometimes comes as a surprise to people
when they hear it for the first time. So QRM plays an
important role in variability reduction because if used
properly, it identifies areas of high risk in the process that
could lead to unacceptable variability in the process or
product performance. Further, it provides actions to miti-
gate these risks so that when the product is manufactured
and consumed by the patient, it has an acceptable level
of quality. QRM can help us get it right from the start. In
fact, for organizations that are committed to six sigma
levels of quality for their processes and products, QRM is
an essential activity in process and product development.
The relationship between QRM and the control strat-
egy can be represented by the diagram in Figure 2.
Example: Environmental Control
We start by having a proposed control strategy as
described in Figure 2. Sometimes this initial control
strategy may have no controls at all. Consider a product
or process that is sensitive to environmental conditions.
The initial control strategy might be to have no environ-
mental controls. Then we perform the risk assessment
and realize that because of the changes in temperature
and humidity across the year, the risk to product quality
(and hence patient safety) is too high. Thus we have a
gap in the control strategy. What we are really saying is
that if we go ahead with this proposed control strategy
(no environmental controls), product quality will be
too variable. This would then lead us to propose a new
control strategy. For example, we propose a humidity
14 Journal of Validation T echnology [Summer 2011]
Figure 2: Relationship between QRM and the
control strategy.
control system. This control system is designed so that
the variation of environmental conditions across a year
is so small that it does not impact product quality. This
is the control strategy that would be implemented to
produce the product. Of course we recognize that even
this new control strategy has some residual risk associated
with it. This is equivalent to saying that no matter how
good our control strategy is there will always be some
variability in the process.
Example: Powder Filling
As an example of this use of QRM to reduce variability,
a new product was being developed that involved filling
a small amount of product into a vial. The product was
in a powder form that would be reconstituted later with
water. An important part of the control strategy was to
control the weight of product filled into each vial. Based
on previous experience with filling vials with liquid,
an initial control strategy was proposed where the vial
weights would be checked by sampling a small number
of vials on a periodic basis and checking the weights on a
gravimetric laboratory balance. Once weighed, the vials
are discarded so this testing is inherently destructive. This
works well for liquids because the variability of adding a
target weight or volume of liquid to a vial is very stable.
However, powders are a different matter. The density of
a powder can vary through the filling equipment and
this leads to higher levels of variability in the weights
of powder filled into the vials. A risk assessment based
on this control strategy showed that the risk of produc-
ing out-of-specification (OOS) product weights in the
vial was too high. Once this was realized, the simplest
change to the control strategy was proposed—increase
the number of vials sampled. However, the risk assess-
iv thome.com
 John McConnell, Brian Nunnally, and Bernard McGarvey.
ment showed that in order to reduce the risk of vials with
product weights that were out of specification, so many
vials would have to be weighed that the process would
not be economic.
An alternative approach to checking the weights was
needed before this process could be implemented. The
solution chosen was to implement a process analytical
technology (PAT) solution to weigh the vials. A non-con-
tact check weigher (NCCW) system was implemented.
Because the system was non-contact, the weighing was
non-destructive, and weighed vials were not discarded.
This technology directly measured vial weights during
filling with the following risk reduction benefits:
• Because each vial is weighed, any vial with an OOS
weight can be discarded automatically.
• The weights of each vial could be used as part of an
automatic controller to adjust the filling system to
reduce the variability of the product weights. This
was the real win for this approach because not only
did we keep OOS product from reaching the patient,
we were able to minimize the variability of the prod-
uct weights for the vials patients did use. Deming
and Taguchi would be proud of us (4).
Gravimetric balance weights will always be more
accurate than the NCCW weights. However, because
the NCCW can measure the weights of all vials, risk is
reduced. With the NCCW system, the revised risk assess-
ment showed that the risk to the patient from OOS prod-
uct weights in the vials was insignificant, and the process
was eventually implemented with this control strategy.
Experience from the implemented process showed that
the PAT solution did in fact deliver the reduced vari-
ability expected.
QUALITY SYSTEMS,
CONTROL STRATEGY, AND QRM
In the previous section we looked at how QRM can
reduce variability by addressing the first issue that can
arise with a control strategy—that is, there is a flaw
in the proposed control strategy. QRM can also help
reduce variability by addressing the second issue, that
is, where the control strategy is good but is not imple-
mented correctly over time for the actual production
processes. Once a process is implemented and product
is being produced and delivered to the patient, the
effectiveness of the control strategy is managed by the
quality systems. The quality systems will react to both
inadequate control strategies and control strategies that
are not implemented correctly over time. In QbD, four
such quality systems are recognized (5). Figure 3 shows
these in a diagrammatic form.
gxpandjv t.com
Figure 3: Relationship between QRM and the
control strategy.
The four quality systems are as follows:
• Product and process monitoring
• Corrective action and preventive action (CAPA)/
improvements
• Change management
• Quality systems monitoring.
QRM gives us a risk model of the process, and as the
quality systems are used, they should be used in the con-
text of this risk model. To see how this works, consider
the following hypothetical situation for the vial filling
process described in the last section.
Example: Powder Filling #2
The control strategy using the NCCW was implemented
and production started. Because the NCCW was new
technology to the company, it had been decided as part
of the risk analysis to do some spot-checking with a
gravimetric balance just to compare the NCCW results
with the balance results. For the first three months of
production, the output from the filling process matched
the predictions of the risk assessment. However, follow-
ing this initial period of good production, the weights
from the gravimetric spot checks began to rise, and the
rate of OOS weights rose to 0.03% (on the high side).
The weights from the NCCW measurements remained
similar to that of the initial three months. This is the
“product and process monitoring” quality system in
action, where the trending of the weight data picks up
a signal that would not be expected. In the language of
statistical process control (SPC), the weights are not in
a state of statistical control. QRM improved the quality
system by suggesting adding an extra measurement for a
period of time until full confidence in the NCCW system
could be established.
Journal of Validation T echnology [Summer 2011] 15
Analysis and Control of Variation.
Once this trend has been observed, a deviation
would be raised and the CAPA system would take
over. This system would conduct a root cause analysis
on the deviation. As the engineer begins to look for
causal factors that would cause the observed devia-
tion, the engineer can use the QRM analysis to rank
where to look first. That is, start by looking for causal
factors in those aspects of the process that have the
higher risks of adding variability to the vial weights.
Given that the NCCW weights were still in a state of
statistical control, and the NCCW is used as an input
to an automatic controller that adjusts the weights if
they stray too far from target, the process engineer
looked at the controller action to see if there were
any changes. What the engineer found is that at the
same time as the gravimetric weights began to rise,
the output of the automatic controller began to rise.
This led the engineer to look at the NCCW system. It
was noticed that at this time a calibration had been
performed on the NCCW. The NCCW was retested
and found to be reading low. This explained what was
being seen—the NCCW was reading low, so in order to
get the NCCW weights back on target, the controller
would have to increase its output. Once the reason
for the incorrect calibration was determined, fixes
could be put in place to prevent it occurring again.
Change Management
Now suppose that the product sales increased and
production rates must be increased. We have been
asked to increase the vial filling rate. In order to make
such a change, we must use the change management
quality system. And of course, we want to make sure
that when we implement the change, we do not create
any unintended consequences and increase the vari-
ability of the process. How would the QRM analysis
help here? Because we know the risks we found when
we analysed the current process, we would want to
make sure that we check the impact of the change on
these risks. So we would want to answer the follow-
ing questions:
• How will the increased filling rate impact the mea-
surement capability of the NCCW?
• How will the increased filling rate impact the con-
troller performance?
• How would the increased filling rate impact the
ability of the system to discard out of specifica-
tion vials?
So QRM can be integral to ensuring that variability
is not introduced via the change management qual-
ity system.
16 Journal of Validation T echnology [Summer 2011]
Finally, with respect to quality system monitor-
ing, management reviews should focus on those
areas of highest risks as identified by the QRM
analysis. This will make the management reviews
more successful.
CONCLUSIONS
QRM can play a vital role in managing and reducing the
variability of pharmaceutical processes and products.
It does this by identifying risks that can add variability
as well as actions that can be taken to mitigate these
risks and so control variability to acceptable levels. As
the process is implemented and product is delivered
to the patient, it can aid the quality systems because it
provides information that speeds up the identification
of causal factors for deviations. It also helps ensure that
changes can be made to the process without introduc-
ing unintended consequences that will subsequently
increase variability.
REFERENCES
1. Joseph Juran, Juran on Quality by Design: The New Steps for
Planning Quality into Goods and Services, 1992.
2. ICH, Q9 Quality Risk Management, November 2005, avail-
able on the ICH website,
http://www.ich.org/fileadmin/Public_Web_Site/ICH_
Products/Guidelines/Quality/Q9/Step4/Q9_Guideline.pdf
3. ICH, Q8(R2) Pharmaceutical Development, available on the
ICH website, http://www.ich.org/fileadmin/Public_Web_
Site/ICH_Products/Guidelines/Quality/Q8_R1/Step4/
Q8_R2_Guideline.pdf
4. J. McConnell, B. Nunnally, and B. McGarvey, “Meeting
Specifications Is Not Good Enough, The Taguchi Loss
Function,” Journal of Validation Technology, Volume 17, Num-
ber 2, Spring 2011.
5. ICH, Q10 Pharmaceutical Quality System, June 2008, avail-
able on the ICH website, http://www.ich.org/fileadmin/
Public_Web_Site/ICH_Products/Guidelines/Quality/Q10/
Step4/Q10_Guideline.pdf JVT
ARTICLE ACRONYM LISTING
ICH International Conference on Harmonisation
NCCW Non-Contact Check Weigher
OOS Out of Specification
PAT Process Analytical Technology
PSD Particle Size Distribution
QbD Quality by Design
QRM Quality Risk Management
SPC Statistical Process Control
iv thome.com