Top 5 Tips for Quality control (QC) labs are well aware of data integrity and the
compendia; however, several questions still arise about how to
Data Integrity maintain and ensure the regulations are met. Here are the top five tips
to help ensure your data meets the current regulations.
Data Ownership
1
Data ownership is crucial when it comes to understanding a process and releasing products. Being the owner of data means
ensuring data is accurate and complete. This includes making sure the data is not modified or deleted without a record of the
change. Some instrument vendors can conduct a full IOPQ on site per regulations, ensuring data will be accurate and complete
and relieving customers from these concerns.
Risk vs. Control
Risk assessments should be conducted regularly to determine where data gaps or errors may occur. To determine the level of
risk, data that includes the following areas needs to be collected and reviewed: risks of unwanted data actions (deletion or
2
alterations); the type of data collection (keyboard entry, scanned); and how/where the data is stored (flat files, database, local,
remote). Other areas to review involve the level of interaction that is allowed (filtering, sorting, exporting), and what analytics
tools are used to facilitate risk detection. Understanding the full data process will inherently increase the overall control of data
analysis. If a data owner is not aware of the above risks within their process, it could be considered a lack of control which
could ultimately lead to non-compliance and regulatory questioning or fines. In short, data can only be verified if there is a
record of its generation, modification, or deletion - this requires controls around data.
Data Analytics
Data can be used in four main ways: descriptive, diagnostic, predictive, and prescriptive. Descriptive data can be used to
3 determine what happened by using analysis and graphs or other visual tools, while diagnostic data is used to determine why
something happened. Predictive data is exactly as it sounds. It is used to determine what will happen in the future by using
historical data to identify trends and determine if they are likely to reoccur. Finally, prescriptive data focuses on what should be
done by extracting insights from predictive data.
Metadata Importance
4
Metadata is a set of data that describes and gives information about other data. In the lab, it includes audit trails and alarms,
information about process setup, operator behavior and capability, relative performance of process inputs, overall process
performance, and instrument capabilities (ex: maintenance and calibration dates). Collecting and reviewing this information
can reveal opportunities to improve processes and results alike.
Employee Training
While all of the above is integral to data integrity, one area where effort should be made is with the people who work with the
data daily. While it is crucial to understand that automation can enhance overall processes, people are still needed to properly
5 set up and maintain them - and always will be. Providing comprehensive initial training and periodic re-training will benefit both
the employer and employee. Training is a way to ensure rules and procedures are adhered to, which leads to consistency and
control, and an employee who feels confident. Employees who do not feel as comfortable or valued might hide errors, delete
data, or deny knowing about an issue. When investing in new instrumentation, be sure to ask your vendor about what kind of
training they can offer and take advantage of their expertise, so your employees are set up for success from the beginning.
