What’s So Good about Good Documentation Practice?

When I first started working in clinical research, “Good Documentation Practice” sounded like a preoccupation for archivists or administrative staff. Of course, documentation should be “good,” meaning neat, accurate, and captured in a somber blue or blank ink rather than a garish pink gel pen. By the time I had finished my first fraud case, though, I understood that documentation is not “good” because it’s pretty, but because it serves as adequate, if imperfect, evidence that a particular event occurred.

Why imperfect? Think about how prone documentation is to falsification. For example, let’s say you were required, for insurance purposes, to produce documentation that you brushed your teeth twice a day. You might complete a log entry every morning and every evening. How easy it would be to fake such documentation!  For example, you might state that you brushed your teeth when you didn’t.  Or, you could complete the log all at once, at the end of the month, even though you couldn’t really remember if you had brushed your teeth earlier that month.  Or someone else could complete it for you, even though they didn’t see you brushing your teeth.

To increase the chances that a document is a faithful record of an event, our industry has implemented ALCOA+ principles. The first four principles – Attributability, Legibility, Contemporaneousness, and Originality – do not guarantee the fifth principle, Accuracy,  but their adherence increases the probability that a document will be accurate:

• Attributability:  If you initial or sign a record, it is thought that you will be less likely to falsify the record, because your handwriting links the record to you.
• Legibility: If the record is legible, it reduces the risk that it will be misinterpreted.
• Contemporaneousness: If you complete the record at the same time the event takes place, you are more likely to capture accurate information than if you complete the record later
• Originality:  Copies introduce the opportunity for falsification or loss of fidelity.

Falsification of records by sponsor and CRO study team members is pretty rare, mostly because we would derive little benefit – no one’s bonus would increase appreciably if a Data Management Plan were backdated by a few days, and we don’t get paid by the eligible patient. More commonly, we fail to document events in a timely manner or create non-compliant documentation with inadequate computer systems.  For example,

• A site monitor forgets to capture training on the monitoring plan prior to the first monitoring visit, and creates and signs a record at a later date indicating that the training took place earlier (lack of contemporaneousness)
• An employee departs the company without documenting study-specific training, so the manager completes a training form on their behalf, documenting the days that training is likely to have occurred (lack of contemporaneousness and attributability)
• A team member indicates approval of a plan via email, but the document attached to the email is a non-final version of the plan (lack of attributability)
• A plan contains links and embedded documents for attachments, but the attachments cannot be rendered from the version of the file stored in the TMF (lack of originality)
• A team attends an important training remotely, and the project manager captures attendees in a list, but trainees do not sign to indicate they attended training (lack of attributability)
• A team stores study-specific in a shared file for easy retrieval, because the TMF is difficult to navigate.  When they need to update their protocol they pull the most current version from the shared file, update it, and send it out for approval; however, the version on the shared file was not equivalent to the final approved version, so the new version of the protocol released was not consistent with the prior version in ways unintended by the study team (lack of originality)
• A data manager performs a back-end operation to invalidate some eCRFs generated by sites in error, but the audit trail for the EDC system captures this action as performed by “System” (lack of attributability)

Our reliance on computers has improved the legibility of our documentation, but has increased attributability issues. Below are some best practices:

• Implement a Part 11-compliant electronic signature tool that can be applied easily to study-specific plans and training records. If you don’t have the funds or time to validate a Part 11-compliant tool, adopt a non-compliant tool – it’s better than not capturing signatures because staff can’t figure out how to work their home printers and scanners.
• Use a Learning Management System for study-specific training. Yes, this is a huge effort, but so is manually tracking and documenting training.
• Use controlled repositories (like validated TMFs, Quality Management Systems, and Regulatory Information Management Systems)for storage and retrieval of all documents. Ensure relevant team members are provided access to systems during onboarding, even if they are controlled by CROs.  Develop views and reports that help staff navigate these systems easily.  When a document is edited, institute a check against the authoritative version to ensure that only intended changes were made.
• When selecting data capture systems, UAT the “back end” functions to determine whether they generate accurate audit trails.