Accred Qual Assur (2001) 6:415–420

© Springer-Verlag 2001 GENERAL PAPER

Fred D. Lasky Quality planning in the ‘in vitro diagnostic’

industry

Received: 1 May 2001 Abstract In vitro diagnostic (IVD) customer complaint monitoring to
Accepted: 13 May 2001 manufacturers play an important role identify improvement opportunities.
in helping to provide laboratory re- The processes used in the IVD in-
Presented at “Antwerp 2000: Quality sults that meet the needs of patient dustry are recognized quality prac-
in the Spotlight in Medical Laboratories”
Conference, 19–20 October 2000,
care. Industry, the primary source of tices that should be, and often are,
Antwerp, Belgium new technologies, uses established used in the clinical laboratory.
processes to assure a continuous sup-
F.D. Lasky ply of products that satisfies health Keywords Quality system ·
Ortho-Clinical Diagnostics, Inc., care needs. The processes include In vitro diagnostics · Manufacturing ·
100 Indigo Creek Drive, Rochester, validation of user needs, with well- Design
NY 14626–5101, USA
e-mail: [email protected] controlled procedures that are used
Tel.: +1-716-453 3880 to manufacture quality products.
Fax: +1-716-453 3113 Management uses routine audits and

Introduction quality process. In today’s global economy, quality ex-

pectations have become more universally consistent.
Clinical laboratory medicine is an important contributor The first quality system entailed a simple visual in-
to the health and welfare of the public. The clinical labo- spection that was used to decide if a product was accept-
ratory is dependent upon its teams of people and the ma- able for sale. This was time consuming, expensive and
terials and devices that it uses to support that effort. not very effective, allowing many defects to be passed
Manufacturers of in vitro diagnostic (IVD) devices have on to the customer. With the introduction of mass pro-
an increasing role in helping assure the quality of the re- duction and the assembly line, gauges helped assure that
sults, as laboratories become more and more dependent parts could truly be interchangeable, whether they were
on technologically sophisticated reagents and instru- installed during manufacturing or later, as a replacement.
ments. To meet the quality requirements of the customer, Henry Ford used this technique to great affect in his
the manufacturer must demonstrate that the stated and mass-produced automobiles that were manufactured in
implied needs of the user are fulfilled. high quantity and at relatively low cost. Defect rates
The concepts of quality have changed over the years. dropped still further when statistical quality control
Initially, the objectives of a quality program were to re- (SQC) processes were introduced at Bell Laboratories to
duce the number of defects in the product or service. The monitor processes and, thus, identify improvement op-
usual approach was to define the product and the param- portunities to produce more consistent and uniform prod-
eters and dimensions necessary to obtain those attributes. ucts. At Motorola, the six-sigma concept [1] for estab-
Manufacturers’ primarily focused on technical criteria to lishing quality objectives predicted that the defect rate
meet their quality goals. Now, manufacturers pay careful would reach a level of only three per million products
attention to the users’ expectations in an effort to achieve delivered if the manufacturing process was capable of
complete satisfaction. The manufacturer converts these achieving a level of uniformity such that the variability
expectations into quantitative and realistic specifications (measured as a standard deviation) was one-sixth the es-
that, in turn, are used as part of establishing an adequate tablished specification. This theoretically based objec-
416

tive is often difficult to achieve, because only the ran- system. There is some flexibility in that each manufac-
dom variability of production is considered and six-sig- turer can decide which elements should be included and
ma does not take into consideration the special causes of to what extent each applies. Twelve elements are listed
defects, mistakes and blunders. Errors due to special in Table 1. I will not itemize or detail each of these, but I
causes are (usually) rare and, thus, are not detected by will group several into three process segments: design
SQC [2]. I will describe many (but not all) of the re- control, process control and audits.
quirements that must be met to produce and maintain
high quality IVD devices using four important catego-
ries: people, process, product, and support. Design control
Poor design has been reported by the United States Food
People and Drug Administration (FDA) to be responsible for
44%of medical device voluntary recalls over a 6-year pe-
People are the key to any process. Management must be riod, and for 90% (!) of software design failures over an
responsible for the processes that are used and the prod- 8-year period [6]. The introduction of design control into
ucts that are made and sold. Competent personnel who processes and product development, as well as process
are well trained, who understand the process, and most modification, has arguably been the greatest source of
importantly, who understand the impact that their contri- improvement to the quality of medical devices [7].
bution makes to the quality of the product are essential. Design control has five general steps:
Finally, people must be held accountable for the quality
1. Define user requirements. For IVD devices, these in-
of the products they produce. Management needs execu-
clude performance attributes needed to satisfy clinical
tive authority to establish and maintain the quality policy.
laboratory users and the physicians who utilize clini-
Management is responsible for providing appropriate re-
cal laboratory services [8].
sources to assure that the quality objectives of the orga-
2. Translate user requirements into design specifica-
nization can be met. A quality process requires manage-
tions. Specifications include (but are not limited to)
ment to conduct periodic reviews of various kinds: prod-
performance requirements, reliability goals, definition
uct design, reliability, and service and support capability
of the environmental conditions in which the device
that is needed to satisfy customers. In summation, man-
will be used, and the ability of the device to meet ap-
agement is accountable for the quality of the processes,
plicable standards, such as those in United States reg-
products and services that are provided by their company.
ulation for the Clinical Laboratory Improvement
All company personnel play a part in a successful quality
Amendments of 1988 [9] and for the German stan-
system. Employees must have appropriate training, edu-
dardization law [10].
cation and background to complete their assigned func-
3. Design and develop the product to satisfy the require-
tions.
ments and design specifications previously outlined.
4. Verify the product meets the specifications.
5. Validate the product by demonstrating that the user
Process requirements are met.
An integrated process is needed for an effective quality Design control processes must accommodate for changes
system. This requires planning, commitment, actions, during the design phase of new products, that can occur.
and follow-up. For medical devices (including IVD de- Manufacturers can estimate the analytical performance
vices) sold or manufactured in the United States, confor- needs of a system by using proficiency testing or exter-
mance to the Quality System Regulation (QSR) [3] is re- nal quality assessment scheme (EQAS) data. Figure 1 is
quired. The regulation is harmonized with standards, an example of how proficiency testing data from the
such as ISO 9001 [4] and ISO 13485 [5], all of which College of American Pathologists (CAP) Proficiency
provide extensive and detailed guidance with regard to Testing Program can be used to assess the current state
what should be contained in a manufacturer’s quality of interlaboratory precision that can, in turn, be used to

Table 1 Quality system regula-

tion (QSR) elements Representative elements in a quality system
– Document controls – Corrective and preventive action
– Purchasing controls – Labeling and packaging
– Identification and traceability – Handling, storage, distribution and installation
– Production and process controls – Records
– Acceptance activities – Servicing
– Non-conforming product – Statistical techniques
 417

Fig. 1 Input comes from the College of American Pathologists define the level of performance needed for a product to
(CAP) Proficiency Survey data over 3 years for creatinine (ex- be competitive.
pressed in mg/dl) by control concentration as a function of the
among-laboratory standard deviation (SD). Each symbol repre- Risk analysis is a powerful technique that can be used
sents a different clinical chemistry system to optimize the safety and effectiveness of medical de-
vices. Two techniques are commonly used. A top-down,
deductive process is “cause and effect analysis,” which
might utilize fishbone diagrams. This technique is first
used to identify potential adverse consequences to a pa-
tient or operator, which are then tracked back to the po-
tential root cause of such an occurrence. This helps the
cross-functional development team identify opportunities
to improve design with the objective to eliminate, re-
duce, or, if mitigations cannot be identified, accept the
condition.
“Failure Mode and Effects Analysis” (FMEA) (or
“Failure Mode, Effects and Criticality Analysis”
(FMECA)) is an inductive, bottom-up process. With this
tool, the consequence of a potential component failure is
assessed against the risk to the patient or operator. This
process involves the development of a matrix, in which a
list of potential failures is recorded and an assessment is
made of the severity of the consequence of each poten-
tial failure. Figure 2 is an example of how probability of
occurrence and the severity of a failure can be used to
judged whether additional mitigations must be consid-
Fig. 2 Risk analysis matrix uses both the probability and the se- ered through changes in design, changes in labeling, or
verity of a potential failure to assess acceptability. Each critical
potential failure can be analyzed in this way so judgments can be an acceptance of the risk of such a failure and its associ-
made regarding further actions that can eliminate or minimize the ated consequences. Both these tools are “what if” tech-
effect on operator and patient. [7] niques that rely on the expertise and experience of a
cross-functional team gathered during the design phase
of a new IVD device. Risk analysis is effective for new
product or process design, as well as for modification of
products and processes.
418

Once a new product or change to a product has been

shown to be feasible, the innovative design must be veri-
fied and validated. These two concepts, as defined in the
QSR [3] are quite different. Verification is “confirmation
by examination and provision of objective evidence that
specified requirements have been fulfilled.” In other
words, specifications and/or requirements have been
met. Verification, therefore, can be conducted on compo-
nents or on the entire system if metrics are available that
can be used to compare to selected characteristics or
specifications. Note that the definition is put in the past
tense. Validation, on the other hand, is confirmation by
examination and objective evidence that the particular Fig. 3 The statistical process control (SPC) chart demonstrates
process reproducibility. The upper and lower control limits (UCL
requirements for a specific intended use can be consis- and LCL) act as statistical out-of-bounds markers for the plotted
tently fulfilled. This is a more encompassing require- results
ment, in which customer needs must be satisfied in the
environment in which the product is intended to be used.
The manufacturer must have evidence that it has confi- Purchasing controls have new importance in today’s en-
dence that these needs can be met consistently (thus, fu- vironment because of the growing use of high-technolo-
ture tense). Therefore, specifications must be developed gy, sophisticated devices, whose components are often
(for each key component) that assure the aggregate of all purchased from outside suppliers. The complexity of the
device testing limits (for verification) can meet user parts used in these devices requires that the manufacturer
needs (for validation). and the supplier enter into a working partnership. Suppliers
should be audited to assure their processes can deliver
quality materials. A manufacturer must decide whether
Process control to build components itself or to buy them from an out-
side source. In the United States, 15% of voluntary re-
A process in control will produce uniform and consistent
calls are due to unacceptably purchased materials or to
product that will minimize rework. An effective and con-
inadequate advice from consultants [6].
trolled process includes (but is not limited to):
Statistical techniques serve many purposes in a quali-
– Document control: approval procedures, change con- ty process. They are used for determining the effective-
trol and training in document content are requirements. ness of corrective and preventive actions and to monitor
– Acceptance criteria and activities: metrics must be es- validated processes to assure consistency and uniformity,
tablished to assure that products are released to meet which, in turn, increases yield. Process validation, as de-
the needs of the user over the expected life of the fined by the QSR [3] is “establishing by objective evi-
device. For IVD devices these include precision, ac- dence that a process consistently produces a result or
curacy and stability. product meeting it’s predetermined specifications.”
– Non-conforming product reports: these reports in- Therefore, acceptability criteria must be defined before a
clude deviation reports, records of scrap and destruc- process is validated, and evidence must be available to
tion of material that do not conform to specifications. demonstrate that those criteria can be met consistently
– Corrective and preventive actions: these are used to for future production events. Other examples of statisti-
help determine where process improvement opportu- cal tools include the use of Shewart plots to monitor pro-
nities exist to minimize and eliminate causes of devia- cesses so they remain in control. Figure 3 illustrates a
tions from specification. process in which material is monitored. The upper and
– Labeling and packaging: for IVD devices, there are lower control limits (UCL and LCL, respectively) are the
specific requirements for content and organization specifications that must be met for the material to be in
from the FDA [11] and the European IVD Medical conformance. Selecting a critical parameter that can be
Device Directive [12]. quantified and tracked helps to verify that the production
– Handling, storage, distribution, and installation: oper- process is in control. A goal of 100% yield can be
ating procedures and work instructions are needed for achieved, depending on the process capability. The same
acceptance of raw materials customer installations statistical data can be used to assess the process capability
etc. and help to determine if improvements must be initiated.
– Records: records must be maintained to track products Figure 4, left illustrates a process under good control,
during production and to provide traceability in distri- which would be expected to provide high yields. A pro-
bution and, if possible and as needed (depending upon cess whose data is illustrated on the right has significant
the criticality of the device), into the hands of the user. variability and, as such, would give lower yield because
 419

Fig. 4 Process capability (Cpk)

blends statistical results with
process specifications. A pro-
cess with a Cpk of 2 (left) is
more robust than a process with
Cpk of 1 (right)

a small but significant number of materials exceed the turing process should be mistake-proofed to avoid blun-
specification, that is, would be out of conformance. Non- ders and errors (“special cause” failures) during manu-
conformance does not necessarily mean that the product facture. Products must also be stable and expected obso-
is unacceptable for its intended use, but the risk does in- lescence should be considered. This is particularly criti-
crease if the specification is appropriately linked to cus- cal in our age of sophisticated electronics, where obso-
tomer expectations. Tracking the number and type of lescence of components occurs frequently. Variation in
non-conformities is an important tool in helping to pro- component lifecycles presents a significant challenge for
duce quality products. The critical process steps and maintaining the long-term viability of (often expensive)
product components must be identified to assure proper instrumentation. Lastly, products must be easily main-
attention is paid to key aspects of product design and the tained and serviceable by both the user and the compa-
manufacturing processes. ny’s technical representatives. This adds to the effective
use of the product by the customer (uptime) and decreases
the cost of service to the manufacturer.
Audits
Audits are essential for assuring that quality system pro- Support
cesses are maintained. Internal audits, which can be con-
ducted as frequently as deemed necessary and by person- The manufacturer has a responsibility to support the
nel who are familiar with the processes and products, medical devices that it produces. Support includes appro-
prevent problems from becoming endemic. Their objec- priate and proper training for the user and any support
tives include having a mechanism to help assure that personnel who are needed to maintain the equipment and
procedures and instructions are adequately understood supplies. Adequate and reliable sources for materials,
and adhered to, and that the process is in compliance. parts and service are important in maintaining effective
Audits provide metrics for management to use for as- use and operation of medical devices. All complaints
sessing adequacy of product and process during quality must be tracked and investigated, and used as a tool to
reviews. Audits are also necessary to continuously moni- identify opportunities for improvement.
tor qualified suppliers. Audits should be conducted of A complaint, as defined by the QSR [3], is “any alle-
post-market information as well. A complaint-tracking gation of deficiency related to the identity, quality, dura-
system must be maintained and audited for effectiveness bility, reliability, safety, effectiveness, or performance of
for capturing and categorizing all expressions of dissatis- a device.” This definition illustrates the broad expecta-
faction by customers. External auditors assure that as- tion that government authorities place on the manufac-
sessments are made without prejudice and by “unin- turer of IVD devices. Such expectations require signifi-
formed” personnel who can be especially helpful in con- cant resources to remain in compliance. Each and every
firming that procedures, instructions for use, and docu- individual valid complaint must be investigated and as-
mentation are clear and understandable. sociated, if possible, with corrective and/or preventive
actions (CAPA). Additionally, service records are moni-
tored and analyzed for trends to identify more-frequent-
Product than-expected service needs. CAPA drives changes in
design, in process or in labeling. If non-conformities do
Product requirements encompass a broad range of attri- occur, the personnel in the production line and those re-
butes. Performance and reliability are especially impor- sponsible for the design must be informed of their poten-
tant for IVD devices. Products should be designed so tial role in causing the non-conformity.
that they are robust; that they operate adequately in envi-
ronments that are reasonable for the global regions in
which they are to be marketed. Although seemingly ob-
vious, a product must be manufacturable. The manufac-
420

Conclusions municating with customers when necessary, can escalate

into a significant issue with negative outcomes: lost cus-
The establishment, use, and maintenance of a quality tomers and lost revenue. Quality manufacturers do not
system requires vast resources and persistent attention. strive to satisfy their customers, but to delight them!
Quality is not free! But it is far superior to the alterna- When a manufacturer understands and designs IVD de-
tive. The costs of repairs, field corrections, recalls, and vices suitable for their intended use, customers will re-
lost goodwill are enormous. A simple problem that is not turn again and again, with even higher expectations for
resolved rapidly with adequate attention to corrective meeting their quality needs.
and preventive actions and with proper attention to com-

References
1. Harry M, Schroeder R (1999) Six sig- 5. ISO 13485 (1998) Quality systems – 10. Qualitätssicherung der quantitativen
ma: The breakthrough management medical devices – particular require- Bestimmungen im Laboratorium. Neue
strategy revolutionizing the world’s top ments for the application of ISO 9001, Richtlinien der Bundesärztekammer
corporations. Doubleday & Company, installation and servicing. ISO, Geneva (1988) Deutsches Ärzteblatt 85:
New York 6. United States Department of Health & B517–532
2. Hinckley CM (1997) Clin Chem Human Services, Food and Drug 11. United States Department of Health &
43:873–879 Administration, (7 October 7, 1996) Human Services (1992) 21 CFR Part
3. United States Department of Health & 21CFR Part 820, Medical devices cur- 809.10 In vitro diagnostic products for
Human Services (1992) 42 CFR Part rent good manufacturing practice human use; Labeling. Washington D.C.
820, Quality system regulation. (cGMP), Final Rule, Fed Regist (1996) 12. Directive 98/79/EC of the European
Washington D.C. 61:52601–52662 Parliament and of the Council. Brussels
4. ISO 9001 (1994) Quality systems – 7. Lasky FD, Boser RB (1997) Clin
model for quality assurance in de- Chem 43:866–872
sign/development, production, installa- 8. Powers DM, Greenberg N (1999)
tion and servicing. International Stan- Scand J Clin Lab Invest 59:539–543
dards Organization (ISO), Geneva 9. United States Department of Health &
Human Services (1992) 42 CFR Part
405 et al. Clinical laboratory improve-
ment amendments of 1988. Washing-
ton D.C.