Reviewing and retaining monitoring records
A monitoring record nobody reviews is not a control. Define the review frequency, what the reviewer checks, how it is recorded, and how long it is kept.
GDP requires temperature records to be reviewed and retained. Review turns data into evidence: someone competent looked, checked against limits, noticed gaps and drift, and signed. Retention makes the evidence available to later investigations and inspections.
A review checklist
- All sensors reporting; no unexplained gaps.
- All values within alert limits; excursions matched to deviation records.
- No sensor flat-lining, stuck values or implausible steps.
- Calibration due dates not passed.
- Alarm log reviewed; acknowledgements and responses present.
- Reviewer name, date and conclusion recorded.
Retention
Retention periods are set by the applicable regulation and company policy; for medicinal products they typically extend at least one year past product expiry, and often five years or more. Records must remain readable — a proprietary export format with no viewer is not 'available'.
What review actually adds
Data accumulating in a system is not a control; it is a liability, because it establishes that the information was available and nobody looked. Review turns it into evidence: a competent person examined it against defined criteria, noticed what was wrong, and recorded that they did. The distinction matters because the second version is what an inspector asks for.
| Frequency | What is checked |
|---|---|
| Daily or each working day | All sensors reporting; any alarm from the previous day acknowledged and actioned |
| Weekly | Full trace scanned for gaps, flatlines and implausible steps; alert-limit breaches that did not reach alarm |
| Monthly | Trends: is a position drifting relative to the others? Calibration due dates; battery status |
| Annually, or at periodic review | Comparison against the mapping baseline; excursion and alarm history; whether the qualified state still holds |
The checks that catch the quiet failures
- Gaps: every sensor should have produced a known number of readings. Counting them is the fastest check there is, and a missing block is invisible on a plotted trace at monthly zoom.
- Flatlines: a stalled sensor reporting the same value hour after hour looks like excellent control. Long runs of an identical value are the signature.
- Sensors that stopped reporting entirely, which looks like nothing at all unless the system has a heartbeat.
- Alert-limit breaches that never became alarms. These are the early warning the alert limit exists to give, and they are routinely ignored because nothing escalated.
- Excursions matched to deviation records, both ways. An excursion with no deviation is a gap in the quality system; a deviation with no excursion in the data is a records problem.
Retention, and the part that gets forgotten
Periods come from the applicable regulation and company policy — for medicinal products typically at least a year past expiry, often five years or more, and longer where a customer's own GMP obligations reach you. The number is the easy part.
The part that fails is readability. A record retained in a proprietary logger format whose software no longer runs, or exported to a system that has since been replaced, has not been retained in any useful sense. Archive the raw export in an open format alongside whatever the system holds, and keep the audit trail with it: moving a record to cold storage does not lift the obligation that amendments remain traceable.
Frequently asked questions
- Does the review have to be by quality personnel?
- The reviewer must be competent and independent enough from the operation to be credible; quality oversight of the process is expected, but day-to-day review can be delegated with training.
References
- [1]EU Guidelines on Good Distribution Practice of medicinal products for human use (2013/C 343/01)
- [2]PIC/S PI 041-1: Good Practices for Data Management and Integrity in Regulated GMP/GDP Environments (2021)
General technical guidance written against the cited sources. It is not regulatory or legal advice and does not replace the applicable standard, guideline or a qualified reviewer's judgement.
Have a question on this topic?
Ask ValiTrac AI and see the evidence and calculation behind the answer.
Ask ValiTrac AIFree calculator: MKT & logger data check
Paste logger readings from storage or a cargo shipment: MKT per ICH Q1A(R2), time out of range, and which figure actually governs the excursion.
Open the calculatorRelated articles
Data integrity for temperature records: ALCOA+ in practice
Temperature records are GxP records. Attributable, legible, contemporaneous, original and accurate — plus complete, consistent, enduring and available.
What a continuous temperature monitoring system must do
Capture at an appropriate interval, alarm reliably, store data securely with an audit trail, and keep working through power and network failures.