Putting a Label on Clinical Trial Success by asafwewe


More Info
									                      Putting a Label on Clinical Trial Success

February 2006
Author: Derek McKay – Business Development Manager – CTS Craigavon
Using automated label verification technology in clinical trial supplies is a consequence
of the drive to shorten lead times and get drugs to market more quickly against an
increasingly challenging regulatory background, contends Derek McKay at Clinical Trial

Licensing authorities are demanding                    Regulatory requirements
additional data to demonstrate drug safety
and efficacy prior to issuing the marketing            While various machine-readable formats
authorisation (MA). As a result, clinical              such as barcodes and RFID can be used to
trials are becoming ever more complex in               identify the pack number or kit type
terms of design, patient populations,                  throughout the supply chain, the key focus
geographical coverage and labelling                    from a regulatory perspective is that of the
requirements. Some blockbuster drugs able                Figure 1: In Clinical Trials,   end user i.e. the
to achieve annual sales in excess of $1                  Every Patient is Unique, so     patient. In the
billion therefore any delays in getting drugs                  are the Labels            case of an
to market can mean lost revenue of millions                                              investigational
of dollars on a daily basis.                                                             medicinal
Compared with marketed products, packing                                                 labelling must
and labelling of investigational medicinal                                               ensure
products are more complex and liable to                                                  protection of
errors. This complexity stems from the need                                              the subject and
to provide blinded supplies with unique kit                                              traceability, to
or patient identifiers on each pack (see fig                                             enable
1). In addition these supplies may be packed                                             identification of
over several smaller runs compared with                                                the product and
larger commercial scale batches. One                                                     the trial and
potential bottleneck in the manufacture of                                               facilitate proper
clinical supplies is the printing and checking                                           use of the
of labels. Large phase III trials often require                                          investigational
millions of unique labels to be printed over                                             product1. It is
30 – 40 languages, this presents a stern                                                 therefore
challenge to those responsible for the                                                   imperative that
manufacture, packing and labelling of the                                                the human
clinical supplies. Fortunately technology                                                readable format
now exists which enables some of the most                                                of any pack is
labour intensive aspects (i.e. label checking)         accurate and fully legible.
to be automated. This article assesses the
challenges and potential benefits of                   Barcodes may be used to electronically
implementing such a system.                            verify that the correct pack number was

                                              Page 1 of 5
scanned, packed and shipped by the                           •   Comparison against fixed
distribution depot. The electronic                           •   Variable text including batch number,
verification of the pack removes the                             expiry date
possibility of the operator misreading the                   •   Colour and shade of text and labels
pack number to be shipped to the                             •   Correct formation and shape of letters
investigator site. In addition, RFID can                         or numbers (variances may arise due
provide additional electronic back up that                       to printer settings)
the correct material has been issued at                      •   Relative positioning of text and
various stages of the packaging and                              objects
distribution process. While the above                        •   Letter spacing, line spacing and
technologies may reduce errors and increase                      margins (can be caused by crooked
efficiencies in delivering the correct pack to                   paper feeds into the printer).
the correct site, it is the investigator and
patient who must be able to read the label to            For blinded, randomised studies,
ensure they are using the correct medication.            comparative checks also need to be
Unfortunately sites are unlikely to have                 performed against label sets from other
electronic verification technology available             treatment groups (assuming they are printed
to them, the printed text on the labels                  in separate runs). The checking of one label
therefore should be clearly visible to the               set against another is performed to ensure
patient. By ensuring the correct text is on the          there are no differences such as boldness of
label, the sponsor is facilitating the                   text or label shade which may be caused by
protection of the patient, traceability of the           the issue of a new batch of labels to the
pack and adherence to the regulatory                     printing operation. Individual patient
guidance.                                                numbers or Med Id numbers, week numbers
                                                         and visit numbers also have to be checked
        Figure 2: Set up of automated                    field by field. Once the above checks have
             inspection process                          been performed another person, typically a
                                                         QC function, must repeat all the checks
                                                         Often the personnel performing the manual
                                                         inspection of the labels will only be familiar
                                                         with 1 or 2 languages, the range of
                                                         languages used in a large trial may exceed
                                                         30. The checking personnel are therefore
                                                         faced with a situation where they have to
Traditional approach to label printing                   check letters, accents and symbols that have
and checking                                             no intrinsic meaning to them. The check is a
                                                         literal comparison of the printed ‘shape’
Printed labels are normally subjected to at              against the master label text. Chinese,
least a 200% manual inspection within a                  Japanese and Hebrew fonts for example
production process. This approach is very                present some of the toughest challenges to
slow, labour intensive, potentially prone to             label checkers who are not native speakers.
error, difficult to validate and ultimately              Personnel who are checking label text for
subjective depending on the personnel                    languages with which they are familiar face
performing the work. In addition, due to the             a slightly different set of problems. Even
human element involved, regular breaks                   when looking at text character-by-character,
need to be factored in to any anticipated                letter-by-letter, humans will inadvertently
checking rates. The label checks will be                 try to assign meaning to the text – it is part
performed against an approved master,                    of the natural reading process. Any
checks will typically include:                           inadvertent interpretation of the text to
                                                         ‘make sense’ can result in the human brain

                                               Page 2 of 5
seeing what it wants to see as opposed to                systems scan a character or string of
what is actually printed on the label. For a             characters and try to interpret the
machine, there is no conflict of interest, it            character(s) by comparing against a pre-
will not add or remove letters intuitively in            trained library of characters. On the other
order to make a sense of a sentence or                   hand OCV systems scan a character or string
statement.                                               of characters and compares these to a
                                                         predefined character or string of characters.
Challenges                                               This can be a more robust system than OCR
                                                         as each scanned character is compared to a
When designing and implementing an                       single character as opposed to all pre-trained
automated checking system the total scope                characters – see fig. 2.
of the challenge faced by the human checker
must be addressed in a machine friendly                  Many of the rules associated with the
format. Due to the nature of clinical supplies           manufacture and packaging of clinical
labels will vary greatly in their overall                supplies hold true when specifying the
dimensions, while individual vial labels may             requirements of an automated checking
only cover a few square mm, patient pack                 system. Key considerations including
labels may cover 1,000’s of square mm in                 flexibility, traceability, ease of set up and
area. Some aspects of the checking process               batch turnaround times must be factored in
present no great difficulty to the human                 to the system design. In the author’s
checker e.g. size of label. Whereas the ‘field           experience, comparative evaluations of
of vision’ is theoretically unlimited for a              human versus machine checking rates show
human, there has to be a finite value                    that automation can improve on human rates
associated with an automated checking                    by 10 times or more on like for like labels.
system. Fortunately with some lateral                    However, checking efficiencies over
thinking and advanced software solutions,                multiple batches can be diminished if the
the maximum label dimension need not be                  time taken to ‘train’ the machine per label
restricted to the capacity of a single camera            type is extensive. In addition the time
or detection unit. However, an automated                 required to reset the machine for a different
system should be designed with other issues              set of labels, but the same label type (e.g.
in mind:                                                 different treatment group) must also be
As a product-critical process, an automated              minimised. The flexibility in the system
label verification system must be subject to             therefore must ensure that it can be set up as
validation in accordance with standard                   quickly to check a label with 3 specified
policies and procedures to provide                       fields of text as one with 10 fields. Each set
operational confidence and to demonstrate                of labels that are checked must also have an
compliance with regulatory requirements                  associated identifier preferably with version
(e.g. EMEA and FDA – 21 CFR Part 11).                    number for the checking run, an operator ID
It is also worth clarifying the difference               including user name and password in order
between Optical Character Recognition                    to provide an easy GMP trace as to who did
(OCR) and Optical Character Verification                 what, when and where. The overall process
(OCV), as both are employed in the                       flow for a typical automated inspection
automated inspection process. OCR refers to              process can be seen in fig. 3.

                                                Page 3 of 5
While the                  Figure 3: Process Flow for a Typical Automated Inspection Process                         velocity but with
system needs                             Start                                                                       variable label
to be able to                                                                                                        dimensions (in
handle a                 Generate labels for system set-up
                                                                                                                     the line of
range of                                                                                                             transfer) will
fonts, texts              Generate verification template
                                                                                                                     result in variable
and                                                                                                                  hourly
characters                   Approve template set-up
                                                                                                                     throughput, the
associated                                                                                                           conveyor speed
with many                           Print label set
                                                                                                                     therefore must be
different                                                                                                            capable of
languages,              Log label batch details eg database                                                          adjustment in
                                name, variables etc.
provision also                                                                                                       order to temper
needs to be            Select verification template, database
                                                                                                                     the throughput
made for                                                                                                             rate.
confirmation                    Verify batch details

of graphics,                                                                                                                  Overprinted
barcodes and                   Start label inspection
                                                                                                                              booklet labels and
any electronic                                                                                                                the associated
markers e.g.                         Inspect label                  Yes
                                                                                                                              materials used in
RFID where                                                                                                                    their construction
applicable. In                           Label
                                                                       More labels to                                         present another
                                                              Yes      be checked in
addition it                           accepted?
                                                                                          No            Generate batch report
                                                                                                                              challenge to the
should also                                    No
                                                                                                          Log out of system   automated
be possible                    Manually inspect label
                                                                                                                              checking process.
for the                                                                                                                       In comparison to
system to                                Label
                                                                                                                              booklet labels,
measure and                                                                                                                   single panel
confirm the               Remove label. Enter reason for                                                                      labels are
overall label                     rejection & e-sig

                                                                      Enter reason for acceptance & e-sig
                                                                                                                              relatively easily
quality i.e.                     Re-start inspection
                                                                                                                              analysed by the
check for                                                                                                                     system’s viewing
marks on                                                                                                                      unit. Protective
labels. In order to assess and verify the                                         plastic covers, tear off sections, perforations
criteria above a number of software                                               and a range thermal transfer print resolutions
solutions need to be employed to analyse                                          add to the complexity of the checking
and interpret the digital image obtained by                                       process for booklet labels. The automated
the system’s viewing mechanism. The speed                                         system therefore should be able to handle all
at which the software processes the images                                        of the above variables so that it can be used
also needs to dovetail with the transport                                         with the full range of clinical trial labels that
system that displays the labels in front of the                                   it is likely to encounter.
viewer. If the software cannot process one
label before the next arrives, the checking                                   Conclusion
process will grind to a halt. Alternatively, if
the conveyor is moving slower than the                                        The use of automated label verification
system can check then the process is not                                      technology for clinical trial supplies has
maximising its output potential. The wide                                     significant benefits. By specifying,
variation in the size of clinical trial labels                                designing and implementing the right
means that a variable conveyor speed will be                                  system contract packagers and sponsors can
required. A conveyor running at constant                                      realise substantial savings in time and

                                                               Page 4 of 5
money for clinical trial supplies. Automation
of the label printing and checking process
during the manufacture of supplies can take
this step off the critical path. Meeting the
ever more ambitious study start dates can
also be achieved more easily using this
approach. Ultimately, large, late phase trials
can be started with reduced lead times and
the blockbuster drugs of tomorrow are
brought a little closer to today.

The author can be contacted at


1. Commission Directive 2003/94/EC,
Article 15.

                                            Page 5 of 5

To top