Data integrity and the ALCOA principles

A laboratory result is only as useful as the record that carries it. If no one can say who ran the test, when it was run, on which instrument, or whether the number on the report is the number the instrument produced, the result cannot be relied on, however careful the work behind it was. Data integrity is the discipline of producing records that can be relied on. ALCOA is the shorthand most quality systems use for it.

The acronym came out of United States regulatory practice and is now used by inspectors and quality units well beyond it. The data integrity guidance published by the UK’s MHRA and the FDA’s guidance on data integrity in drug manufacturing both build on it, as does the PIC/S guidance used by inspectorates in many countries. Each letter names a property a record should have from the moment it is created, not one it acquires later when someone tidies up.

The five properties

• Attributable. The record shows who created it and who changed it. On paper that means a signature or initials that can be matched to a signature log. In a computerized system it means individual user accounts, not a shared login.
• Legible. The record can be read by someone other than its author, now and for as long as it must be kept. That covers handwriting, faded thermal paper, and electronic files in formats that future software can still open.
• Contemporaneous. The record is made when the work is done. A weight is written down as it is read from the balance, not reconstructed from memory at the end of the shift.
• Original. The record is the first capture of the data, or a verified true copy of it. A value typed into a spreadsheet from a printout is a transcription, and the printout is the original.
• Accurate. The record reflects what happened, with no errors and no edits that are not themselves documented.

What the plus adds

Later guidance extended the list, and the extended form is usually written ALCOA+. The additions are complete, consistent, enduring and available. Complete means the record includes everything that happened, including repeat tests, aborted runs and results that were rejected, not only the run that was reported. Consistent means dates, times and sequence agree with each other: a sample cannot be analyzed before it was received. Enduring means the record survives for its full retention period on a medium that will last. Available means it can be retrieved for review or inspection within a reasonable time, by the people entitled to see it.

How records fail

Most data integrity problems are ordinary. They come from convenience, time pressure, and systems that make the right practice harder than the wrong one.

A shared login on an instrument computer makes every action attributable to a generic account rather than a person. A scrap of paper or a glove used to note a reading before it is copied into the notebook means the original record is thrown away. A notebook filled in at the end of the day is not contemporaneous, and its times are estimates. A correction made with correction fluid, or by overwriting a cell in a spreadsheet, destroys the value it replaced, so no one can tell what was changed or why. A chromatography system set up so that analysts can delete or reprocess data without a record leaves no way to show that the reported result was the first one obtained rather than the best of several.

The fix in each case is procedural more than technical. Correct paper records with a single line through the error, so the original stays readable, and write the new value beside it with initials, date and a reason. Give each person their own account. Record directly into the controlled notebook or system. Configure instruments so that deletion is not possible and every change is logged.

Paper and electronic records alike

The principles apply equally to paper and electronic records, but they look different in each. On paper, the controls are bound and numbered notebooks, issued worksheets, permanent ink and signature logs. In electronic systems they are user access controls, audit trails, protected storage and backups, and validation showing that the system does what it is meant to. In the United States, 21 CFR Part 11 sets out requirements for electronic records and signatures; in the European Union, Annex 11 of the GMP guidelines covers computerized systems. ISO/IEC 17025 asks accredited laboratories to control their records and protect their data in a comparable way.

Hybrid systems, where an instrument produces an electronic file and the analyst prints a summary and signs it, need particular care. The signed printout is not the original record; the electronic data is. A review that looks only at the printout cannot see a reprocessed integration or a deleted injection. The guidance documents are clear that the electronic record, with its metadata, is what has to be retained and reviewed.

For a reader looking at a certificate or report from outside the laboratory, ALCOA gives a short checklist. The document should make clear who performed the work and who approved it. Its dates should agree with each other. It should identify the original data it summarizes. A record that meets those tests plainly is usually a sign of a laboratory that takes the rest seriously.