Computer System Validation
Computer System Validation
Updated Compliance Program Guide 7346.832,7 which has as objective 3 a data integrity
audit of laboratory data.
Trained their inspectors in data integrity, which means that there is now a focus on
computerized systems and the data contained therein rather than paper output.
There is level 2 guidance for some aspects of data integrity: shared user log-ins, why
paper cannot be raw data from a computerized system, and using samples as SST
injections.8
The European Medicines Agency has started posting GMP non-compliances online, where many
cases of data integrity have been noted,9 and Health Canada has now stated that GMP inspections
will be unannounced due to data integrity issues that it has uncovered.10
MHRA Approach to Data Integrity
Our story begins in December 2013 when the MHRA gave the pharmaceutical industry an early
Christmas present via their Web site.11 This announcement stated that, from January 2014:
The MHRA is setting an expectation that pharmaceutical manufacturers, importers and contract
laboratories, as part of their self-inspection programme must review the effectiveness of their
governance systems to ensure data integrity and traceability.
However, these comments should not understate the fact that this is the first comprehensive
guidance for industry on data integrity that has been issued by a regulatory authority. Although
some may argue that the FDAs CPG 7346.8327 should be the first such document, as it outlines
a laboratory data integrity audit. However, this is intended for inspectors not industry. The
MHRA document is a guidance for industry.
1 in 28?
One question that struck me reading the document for the first time was why has the MHRA
taken the steps to publish this guidance for industry? The UK is one of the 28 member states of
the European Union, and each member state has its own regulator responsible for inspections
within its borders and for inspections outside the EU. However, the regulations and the majority
of guidance documents or concept papers are usually issued by the European Medicines Agency
(EMA), the pan European body responsible for regulations, product licensing, etcetera. What
will one EU-competent authority achieve working on its own?
Introduction to the MHRA Guidance Setting the Scene
The introduction to the MRHA guidance looks at the justification for data integrity and the first
sentence sets the scene:
Data integrity is fundamental in a pharmaceutical quality system which ensures that medicines
are of the required quality.14
It goes on to state that this guidance is complimentary to EU GMP. It also reiterates an MHRA
expectation for a data governance system, which repeats their original 2013 approach. 11 It also
warns companies not to return to paper, as this would be a breach of European Union directive
2001/83/EC15 which, in article 23, requires companies to take account of scientific and technical
progress.
Two changes have been made in the March 2015 version of the document in the introduction:
The first is informational and refines the scope of the document to active substances
(APIs) and dosage forms. Therefore, it excludes excepient from the scope of the
guidance, presumably as these are lower risk.
The second change is more far-reaching for regulated organizations. In the original
version, the guidance stated that organizations are not expected to implement a forensic
approach
to
data
checking,
However the revised version slips in four additional words to read not expected to
implement a forensic approach to data checking on a routine basis,.
This changes the whole approach to data integrity: the original version wanted a system
to provide an acceptable state of control based on data integrity risk. However, do we
now need to have CSI on standby to rush in waving their torches around looking for clues
whenever
a
data
integrity
alarm
is
raised?
Perhaps a more rational approach is that we leave the forensics to the regular self-
inspections or for cause audits and the acceptable state of control to routine operations,
such as the second-person checks of laboratory data and the reportable results?
Drowning in Integrity Definitions
The MHRA guidance document gives a definition of data integrity which is shown in Table 1
along with four other definitions (two from the FDA, one from National Institute of Science and
Technology and one from the Institute of Electronic and Electrical Engineers IEEE) of either
data integrity or integrity. I have deliberately listed all five definitions in Table 1 to illustrate that
different organizations, or even different divisions of the same regulatory organization; can have
different approaches to the same subject.
Table 1: Data Integrity and Integrity Definitions
Source
MHRA14
The extent to which all data are complete, consistent and accurate throughout the
data lifecycle (data integrity).
FDA 116
The degree to which a collection of data are complete, consistent and accurate (data
integrity)
FDA 217
Data, information and software are accurate and complete and have not been
improperly modified (integrity)
18
The property that data has not been altered in an unauthorized manner (data
integrity).
Data integrity covers data in storage, during processing, and while in transit
IEEE19
NIST
What can we learn from these definitions of integrity and data integrity? Let us attempt to
reconcile and combine them into a single approach for data integrity:
Data must be complete, consistent and accurate (MHRA & FDA 1, 2).
The first three bullet points hold for manual processes, as well as hybrid and electronic
computerized systems, and the fourth point covers hybrid and electronic systems.
Wrong Definition of Data and Integrity Criteria
In the definition section, data is defined as information derived or obtained from raw data (e.g. a
reported analytical result).14 This definition is misleading. How can be data be defined as
information? Data are processed and reduced to information, which itself can be further
interpreted to produce knowledge. However you look at it, data can never be information.
MHRAs own definition equated information as analytical results (i.e. a reduction of raw data).
In the regulatory expectation for data is the requirement to comply with ALCOA principles.
Table 2 shows these criteria in the first five rows (Attributable, Legible, Contemporaneous,
Original and Accurate). The first line of each criterion is the MHRA requirement, and underneath
are my additions to them. However, when looking at data integrity, ALCOA principles, which
were developed for paper records, are not sufficiently comprehensive. The GAMP Data Integrity
SIG has adopted the EMA GCP20 criteria for electronic source data, which are shown in Table 2
in the last four rows and summarized as ALCOA+. The four additional criteria are: Complete,
Consistent, Enduring and Available. Therefore, the data definition and the regulatory expectation
sections in the MHRA guidance need to be revised, in my opinion, to be comprehensive for
paper, hybrid and electronic processes and systems.
Table 2: ALCOA+ Criteria for Data Integrity
Criterion
Meaning
Legible (MHRA)
Legible
Can you read the data together with any metadata or all written entries on paper?
Contemporaneous (MHRA)
Contemporaneous
Documented (on paper or electronically) at the time of an activity
Accurate (MHRA)
Accurate
Complete
All data from an analysis, including any data generated before a problem is
observed, data generated after repeat part or all of the work or reanalysis
performed on the sample.
For hybrid systems, the paper output must be linked to the underlying electronic
records used to produce it.
Consistent
All elements of the analysis, such as the sequence of events, follow on and data
files are date (all processes) and time (when using a hybrid or electronic systems)
stamped in the expected order
Available
The complete collection of records can be accessed or retrieved for review and
audit or inspection over the lifetime of the record.
The next part of this series will look at the data governance system.
Review and Critique of the MRHA Data Integrity Guidance for Industry Part 2:
Data Governance System
This is the second of a four-part series reviewing and critiquing the recent Medicines and
Healthcare products Regulatory Agency (MHRA) guidance for industry document on data
integrity. The first part of the series provided a background to the guidance document and
discussed the introduction to the document. In this part, we will look at the MHRA requirement
for a data governance system.
Data Governance System
The MHRA guidance document defines a data governance system as:
The sum total of arrangements to ensure that data, irrespective of the format in which it is
generated, is recorded, processed, retained and used to ensure a complete, consistent and
accurate record throughout the data lifecycle.
Let us explore what this should entail. First, no other regulatory agency is requiring
organizations to have a data governance system. However, this is a good idea given the number
of issues involving data integrity that have been found recently. The rationale for this is based on
MHRAs interpretation of ICH Q10 on Pharmaceutical Quality Systems (PQS), which is
incorporated in Part 3 of EU GMP and that of EU GMP Chapter 1 on PQS, which is based in part
on ICH Q10.
Under the clause 1.8 for GMP for medicinal products it states
(vi) Records are made, manually and/or by recording instruments, during manufacture which
demonstrate that all the steps required by the defined procedures and instructions were in
fact taken and that the quantity and quality of the product was as expected.
There is a similar requirement for quality control laboratories in clause 1.9 which states:
(iv) Records are made, manually and/or by recording instruments, which demonstrate that all
the required sampling, inspecting and testing procedures were actually carried out. Any
deviations are fully recorded and investigated;
I believe that it is on these two clauses that MHRA bases the interpretation for a data governance
system. As required by EU GMP Chapter 4, records are evidence that instructions have been
executed correctly. However, it is a long stretch from sections 1.8 and 1.9 of EU GMP to a data
governance system. In contrast, FDA has a least burdensome approach to the interpretation of
their medical device regulations, in a risk-based world, should this not be the way forward?
More detail is provided on the data governance structure by the MHRA in the definitions section
of the guidance:1
Data governance should address data ownership throughout the lifecycle, and consider
the design, operation and monitoring of processes / systems in order to comply with the
principles of data integrity including control over intentional and unintentional changes
to information.
Data Governance systems should include staff training in the importance of data
integrity principles and the creation of a working environment that encourages an open
reporting culture for errors, omissions and aberrant results.
From this, we can derive the following elements of a data governance system, which are listed
below and shown linked in Figure 1:
management responsibilities
risk assessment
data owners, who can be equated to the process owners of computerised systems under
Annex 118 and the responsibilities combined
the accountability, will be devolved to the data owners of specific processes and
computerized systems.
Working Culture and Data Integrity Issues: This is the most important area that senior
management can foster. What is required is the creation and maintenance of an open and
no-blame culture to enable staff to raise data integrity issues. Part of this culture is the
ability of staff to raise data integrity issues without fear of retribution via reporting
mechanism to senior management.
Policies, Procedures and Training: Procedures for ensuring data integrity for all
activities (both GMP and non-GMP to avoid dual standards) followed by training in these
procedures for all staff is essential. Data integrity must be included in the regulatory
requirement for on-going GMP training to reinforce the message. Part of the policies and
procedures is the requirement for risk assessment. This needs to be undertaken to
determine the impact and criticality of the records generated by each system to determine
the controls. Then, via a gap and plan process, assess the existing controls in place to
determine what, if any, additional controls are required to ensure data integrity. The
GAMP good practice guide on Compliant Part 11 Records and Signatures 13 already has a
list of controls to protect electronic records, and this could be adapted by organizations to
include paper records as well.
Data Ownership: There is a requirement for a data owner under the MHRA guidance.
Rather that create another role, I would suggest that, for computerized systems, the
existing process owner in the laboratory for each system should also be responsible for
the integrity of data generated and managed within their systems. However, there are
potential problems what happens if data are transferred manually to a spreadsheet for
further calculations or are transferred from one system to another electronically is the
same person the data owner? However, if the responsibilities of the process owner and
data owner are combined, the issue should be resolved for the majority of processes and
systems.
In addition to the MRHA document, there is an extreme example of a data governance system in
operation today, and that is documented in the Ranbaxy consent decree that the company and the
FDA agreed upon in January 2012. This established the post of Chief Data Integrity Officer
reporting to the Board with a number of tasks to carry out to resolve the long standing
falsification issues that had arisen over the previous four to five years. Part of the setup was the
establishment of a whistle blowing phone line that any company employee can call without fear
of retribution. I am not advocating such a governance structure, as the Ranbaxy approach has
been defined to correct falsification carried out over some time. What is required is to integrate
the data governance within the pharmaceutical quality system as shown in Figure 1.
However, after writing this section, I am still reminded that this is a single inspectorate within
the European Union how effective will this request for a data governance system be? Why is
the EU not acting in unison?
Summary
In this part of the review of the MHRA data integrity guidance, we have focused in the data
governance system promoted by the UK regulator. There is a basis for this when interpreting EU
GMP Chapter 1, and an outline of the elements for such a data governance system are presented
and discussed. In the next part of this review and critique series, we will look at data criticality
and a data life cycle.