ALCOA+ data integrity checklist

What data integrity means

Data integrity is the extent to which data are complete, consistent and accurate throughout their life, from the moment they are captured until they are destroyed at the end of the retention period. A quality system runs on records: a batch is released because its records say it passed. If the records cannot be trusted, neither can the release.

Data integrity failures are not only fraud. Most are ordinary habits: a shared login because it is quicker, a week of entries written up on Friday, a printout kept while the electronic file is overwritten. Each one makes it impossible to show what actually happened.

ALCOA and ALCOA+

ALCOA is a memory aid for the properties a good record has. The original five are attributable, legible, contemporaneous, original and accurate. ALCOA+ adds four more: complete, consistent, enduring and available. The table in the tool says what each one means in practice and what to look for, on paper and on screen.

Paper, electronic and hybrid records

• Paper records fail through good documentation practice lapses: overwriting, correction fluid, blank fields left open, entries made later.
• Electronic records fail through system setup: shared or generic logins, audit trails switched off or never reviewed, clocks that users can change, data held only on a local drive, files overwritten by the next run.
• Hybrid systems, an instrument with a printout that is signed, are the riskiest, because each half assumes the other is the record. A printout is not a complete copy of the electronic original: it loses the metadata and the audit trail.

Electronic records and signatures used for regulated records have their own requirements, such as the FDA's 21 CFR Part 11 in the United States: validated systems, secure audit trails, unique user IDs and controlled signatures.

How to run the assessment

1. Choose the sample: a lot, a period, a process. Pick at random where you can, and follow the data from the instrument to the release decision.
2. For each record, answer each attribute Yes, Partly, No or N/A. Look at the electronic original, not only the printout, and ask to see the audit trail.
3. Write a finding for every No, and for every Partly that matters, with the objective evidence.
4. Read the attribute chart. A weakness that shows up across several records points to a system cause, such as a procedure, a system setting or a habit, rather than one person's slip.

The score counts Yes as one and Partly as a half, over the attributes answered. It is a summary to compare areas and track progress, not a pass mark. One record that was not contemporaneous can matter more than a high score.

When a finding is serious

A data integrity finding raises a second question: what decisions were made on these data? If product was released on results that cannot be shown to be true, the finding needs containment and an assessment of the product, as well as a corrective action.

Related tools

The audit checklist and working papers hold the wider audit; this tool is the data integrity part of it. Findings go into an audit nonconformity report. The data collection plan is where integrity is designed in, before any data are collected.

On the exam

CMDA II.B.3 asks you to examine record-keeping for data acquisition systems and evaluate audit data against ALCOA+. CMDA III.A.1 includes 21 CFR Part 11 on electronic records and signatures, and III.D.3 document and record control. CQPA III.B.4 covers data quality attributes and the methods that prevent and detect data integrity problems, such as audit trails and record management training, and CCT III.G the integrity of calibration records. Expect to name the attribute a described lapse breaks.

Other tools and templates