L I F E S C I E N C E M A N U FA C T U R I N G

LEANING THE
BATCH RECORD
PROCESS
THE PROBLEM W I T H B AT C H R E C O R D S
Life Science manufacturing operates in a highly lead-times can be significantly improved by re-
regulated environment and significant effort is expended engineering the manual review, approval and error
in compiling and reviewing batch records. correction processes; and RFT can be significantly
improved by re-designing the batch record itself.
In fact batch records consume substantial amounts of
operator, supervisor and dedicated reviewer time. Despite
this, long lead-times for approval of the batch
documentation and poor ‘Right First Time’ (RFT)
performance are very common. In addition, there is often
a small ‘cottage industry’ built up around the correction
of errors.

Some companies have addressed these issues by

implementing an Electronic Batch Record (EBR) but for
many the cost and complexity involved makes this option
unfeasible. All is not lost however - batch record
Source: BSM’s 2007 EBR Benchmarking Report

WHAT WE OFTEN FIND IN

BATCH RECORD PROCESSES
NO REAL OWNERSHIP FOR DOCUMENT ACCURACY INEFFECTIVE INTERIM REVIEWS
We often find that manual batch record processes are In many companies, document reviews by a dedicated
mature and stable with static levels of performance over manufacturing resource or supervisor (or both) are added
several years. Long lead times and poor document in an effort to improve the RFT at the formal QC/QA
accuracy have become the ‘norm’ and are accepted. check. These ‘interim’ reviews rarely work well, with a
significant volume of errors still getting through to
Often, there is no real ownership or understanding of
QC/QA. They also add substantially to the overall lead-
document accuracy performance at operator level. This is
time.
somewhat understandable given that document design is
often poor and unintuitive and that the lead-time
QUEUES, BACKLOGS AND LONG LEAD-TIMES
between an operator completing a record and a ‘return’
for correction is typically very long. It is not unusual to find queues and backlogs before each
of the review stages in a manual batch record process.
LARGE AND OVERLY COMPLEX BATCH RECORDS It is also typical to find queues and delays associated
with the error correction process. This creates long lead
Master batch records often have a significant amount of
times and high levels of ‘Work In Progress’ (WIP). High
unnecessary duplications and transcriptions. In addition
levels of WIP inevitably lead to a lot of non value adding
they tend to get larger and more complex over time with
effort being expended in managing, prioritising
all sorts of spurious instructions and additional data
expediting and tracking batch records through the
requirements being added over the years. This is quite
process.
often as a result of CAPA (Corrective And Preventative
Action) initiatives. However, the records often do not
AVERAGE LEAD TIME (IN CALENDAR DAYS) FROM
accurately reflect the current manufacturing process. It
COMPLETION OF MANUFACTURING UNTIL BATCH RELEASE
is typical to find lots of redundant entries, and entries
Sector Average Max. Min.
not in the same order as the actual process. This
Bulk Pharmaceutical 20.2 120.0 5.00
complexity and inaccuracy increases the risk of errors.
Finished Pharmaceutical 19.2 120.0 2.00
Every manual entry is in fact an opportunity for error and
master batch records should be revised regularly to keep Medical Device 23.7 56.0 1.00
pace with process changes and to minimise the overall Biopharma 23.7 40.0 6.00
amount of data entry required.
Source: BSM’s 2007 EBR Benchmarking Report
Average of all respondents = 21 days
UNWIELDY, SLOW AND PUNITIVE CORRECTION there may be additional delays waiting for a particular
PROCESSES operator to come back on shift, etc.

Errors detected at the QC/QA review are normally routed The delay between the record being created and being
back to the originator for correction. Often these are returned for correction often means that ‘the trail is cold’
accompanied by complex CAPA type paperwork and may and the investigation and corrective actions become
involve supervisors and managers in investigations and paperwork exercises. Clearly a faster ‘flowed’ process is
corrective actions. This can result in significant delays and required.

ERROR TYPES AND CAUSES

There is a surprising commonality amongst life science The second most common error type is ‘transcription
companies in the type and ranking of manual batch errors’ (where data is transcribed incorrectly into the
record errors. The top error type is almost always ‘errors batch record from labels or print outs, etc). It is often
of omission’ (in which the required information, signature possible to eliminate the need to transcribe the data at
or ‘N/A’ is simply not filled in). Poor layout can make it all by re-engineering of the batch record or the source
easy to miss data entry requirements and this type of error material or both.
is more likely to occur when the batch record sequence
Another common ‘error’ category is ‘inadequate or
does not match the actual process. A lot can be done with
unclear entry or comment’. This is almost always because
batch record sequencing, layout, shading and the use of
the operator does not understand what detail the reviewer
data masks to reduce the propensity for errors of
expects or needs. A review process which puts the
omission. Reducing the overall volume of manual entries,
reviewer in direct contact with the originator in ‘real time’
by removing unnecessary and obsolete entries and
will short circuit this issue.
consolidating remaining entries wherever possible, will
also help.

SOLUTIONS
THE KEY LEAN PRINCIPLES OF FLOW AND WASTE ELIMINATION APPLY BUT MANUAL BATCH RECORD
PROCESSES ARE NOT THE SAME AS MANUFACTURING AND A GENERIC APPROACH WILL NOT WORK.

REDUCING LEAD–TIMES reduce errors as well. If errors can be reduced sufficiently,

the interim reviews can often be eliminated thereby
To achieve fast and consistent lead-times, queuing before
reducing costs and lead-time.
reviews must be eliminated. This requires the review
workload to be ‘level loaded’ and matched with available
review resources. Batch records should also flow between
REDUCING ERRORS
review stages and errors should be corrected without delay.
This may sound impossible but it can be done. One Re-engineering of the batch record should begin with a
method to combine these requirements is via Real Time rigorous examination of the data required followed by the
ReviewTM by which the records for active batches are removal of unnecessary and obsolete entries and
incrementally reviewed during manufacturing (every batch consolidation of remaining entries wherever possible.
every day). This avoids queuing altogether and error Every manual entry removed further reduces the risk of error.
correction is normally instant. Reviewers also get to Batch records should be designed to match the actual
communicate the standard of entry required directly to sequence of the manufacturing process avoiding any need
operators in real time. Given that the number of concurrent by the operator to skip backwards and forwards through the
batches is normally limited by the number of rooms, lines batch record when filling it in. It should also be redesigned
or reactors, the workload is often inherently level loaded. to reduce the actual effort required to complete it. Data
There are many variations on this theme and the ultimate masks, shading, and good layout should be used to help
solution will be somewhat different in each company. prevent errors of omission.
Obviously the batch record would need to be re-engineered
to support a flowed process. If this needs to be done
anyway, the opportunity should be taken to redesign it to
SAMPLE BATCH RECORD DESIGN
STEP# OPERATION DESCRIPTION DATA INITIALS/DATE BEFORE The use of shading makes data
5553 Allow the transfer lines to cool to ambient entry points more visible reducing
temperature, then transfer 62.5kg +/- 2.0kg of the Fermentor Weight Before Transfer: kg
(509-01020) Amino Acid Feed Solution to the
the risk of ‘errors of omission’. It
Target Weight = Fermentor Weight
fermentor based on the increase in weight of the before transfer + 62.5kg = kg also makes the batch record
fermentor. Fermentor Weight After Transfer: kg
Net Addition = Fermentor Weight After
easier to review. The use of data
- Fermentor Weight Before = kg masks indicating the number of
(60.5 – 64.5kg)
decimal places required avoids
STEP# OPERATION DESCRIPTION DATA INITIALS/DATE AFTER format errors. The use of these
5553 Allow the transfer lines to cool to ambient
and other error reducing
temperature, then transfer 62.5kg +/- 2.0kg of the Fermentor Weight Before Transfer: (A) . kg

(509-01020) Amino Acid Feed Solution to the +62.5 kg strategies combined with good
fermentor based on the increase in weight of the Target Weight: = . kg
general layout and sequencing
fermentor.

Fermentor Weight After Transfer: (B) . kg

and a reduction in the overall
/
volume of entries will significantly
Net Addition (B-A): . kg INT / DATE
improve RFT performance.
(60.5 – 64.5kg)

CONCLUSION
Life Science Manufacturing has always put significant effort, resources and cost into its manual batch record processes.
Despite this, Right First Time performance and lead-times are almost universally poor. Occasional improvement initiatives
may lead to temporary improvement but performance generally returns to former levels once the focus is “off”. What is
needed is a more radical approach which re-engineers the fundamental processes (and the batch record) based on key
Lean Principles. Batch Record processes are not the same as manufacturing processes and careful adaptation of the
Lean techniques is required.

An Electronic Batch Record (EBR) is probably the ultimate solution but the complexity and costs involved determine that
it is not currently a viable solution for many companies. The good news however is that careful re-engineering of the
manual processes will deliver major reductions in lead-times and costs. Improving the layout, sequencing and formatting
of the batch document itself and eliminating unnecessary entries will significantly improve Right First Time (RFT)
performance.

If a manual batch record project is to be successful and delivered within a reasonable time frame, it is necessary to
resource it properly. This should include significant senior management support and the use of external consultants with
a relevant track record and excellent project management skills. Obviously this costs money and a clear ROI (Return on
Investment) and measurable project objectives should be established prior to embarking on a full project.

To discuss any aspect of this briefing or your own batch record project or plans please contact:
TOM REYNOLDS, Operations Practice Director, E: [email protected]

BSM is a leading management and technology consulting company working in the Life science sector. We assist companies to deliver significant measurable improvement across a
range of manufacturing, testing, documentation and business processes. We develop innovative solutions via the application of best practice lean, re-engineering and change management
techniques. We have an extensive track record of successful implementations.

Copyright 2008, BSM, All rights reserved. BSM is not liable for any errors contained within this Briefing.

Ireland UK
Parkmore Business Park West, Galway 11-15 Betterton Street, Covent Garden, London, WC2H 9BP
T: + 353 (0)91 746900 E: [email protected] W: www.bsm.ie T: +44 (0)207 470 8766 W: www.bsmconsulting.co.uk