Calibration interval: how to define it, justify it and let it evolve

A choice that is not made just once

The most widespread mistake in managing calibration intervals is treating them as a fixed attribute of the instrument: decided at inventory entry, carved into the system, copied over the years. It is an approach that holds up poorly in a GMP audit because it does not answer the fundamental question: on what basis was that interval chosen? An answer like "we do it every year because we have always done it that way" is not a justification.

A correct calibration interval is the conclusion of a documented reasoning, applied to the individual instrument, reviewed over time. The reasoning may lead to "once a year" just as it may lead to "every six months" or "every two years": what matters is that the choice is traceable to an explicit logic.

The factors to combine

Five variables contribute to defining the interval, and none of them alone is sufficient.

The instrument's GMP criticality

How serious is the impact if the instrument drifts without us noticing? A temperature probe on a sterilization autoclave has very high impact: an undetected drift may have compromised the sterility of entire batches. A service pressure gauge on an auxiliary line has low impact. The critical/non-critical classification done at inventory is the starting point.

The instrument's historical stability

In the previous calibration cycles, has the instrument always been within tolerance? Has it shown predictable or erratic drift? Has it had any OOT? It is the most reliable and most neglected factor. Three consecutive cycles of calibrations well within tolerance justify a cautious extension; an observed drift suggests shortening.

Conditions of use

Is the instrument used in nominal conditions or under stress? Vibration, thermal excursions, humidity, contact with aggressive chemicals, frequency of use: all affect the drift rate. A laboratory balance in a climate-controlled room has different needs from a process balance in a hot, humid environment.

Manufacturer's recommendations

The manufacturer knows the instrument's typical drift. Its recommendation is a reference, not a constraint. It must be taken seriously but can be modified according to the real context of use, with documented justification in both directions (whether to shorten or lengthen).

Regulatory and customer requirements

Some regulatory authorities or some customers may impose maximum intervals for instruments on specific processes. These are external constraints that prevail over the internal choice.

How to arrive at an initial interval

For a new instrument, with no internal history, a reasonable sequence:

1. GMP classification: critical or non-critical, and in what function.
2. Conservative default: you typically start from an interval aligned with or shorter than the manufacturer's recommendation.
3. Adjustment for conditions of use: if the instrument operates in harsh conditions, shorten it further.
4. Adjustment for criticality: for GMP-critical instruments you can shorten even further while building up history.
5. Documentation: the choice is reported on the metrological data sheet with an explicit justification.

After the first 2–3 calibration cycles, internal data are available and the review phase begins.

The periodic review

This is the phase that separates a mature GMP system from a merely formal one. Periodically — typically on each instrument's anniversary or in an annual global review — the historical calibration data are examined:

• observed drift: by how much did the instrument move from the tolerance center between one calibration and the next?
• trend: is the drift stable, increasing, oscillating?
• margin: did the instrument always stay within tolerance with a wide margin, or did it approach the limits?
• OOT: were there Out of Tolerance events? How often?

Three typical behaviors and the appropriate responses:

Instrument always well within tolerance, stable and contained drift. An extension of the interval can be proposed, with documented justification. The extension must be done in steps (e.g. from 12 to 18 months, then possibly to 24), not in large jumps, and must be monitored on the following cycle.

Instrument within tolerance but with increasing drift or decreasing margin. The interval is kept and monitoring is reinforced. A preventive replacement may be planned if the trend is worsening.

Instrument with OOT or erratic drift. The interval is shortened, the cause is analyzed (natural drift, conditions of use, structural instrument problem), and replacement or repair is assessed.

The documentary justification

In an audit, the inspector does not ask "why 12 months" — they ask "on what basis did you decide 12 months and on what basis do you verify that 12 months are still adequate?" The answer cannot be verbal. There must be:

• a procedure describing the method for defining and reviewing intervals;
• for each critical instrument, a metrological data sheet reporting the current interval and its justification;
• for each periodic review, a review record with the historical data analyzed and the decisions taken;
• for each extension or shortening, a written justification with evidence of the supporting data.

ILAC G24 (the International Laboratory Accreditation Cooperation guideline) is the international methodological reference for defining and reviewing calibration intervals. Although intended for calibration laboratories, it provides frameworks that are also applicable to the site's internal management.

Frequent errors

The same interval for all instruments of the same type. Identical instruments in different conditions of use behave differently. Family-based management is a simplification that should be justified, not assumed.

Extension without supporting data. The site proposes moving from 12 to 24 months "to reduce costs". Without documented historical stability data, the extension is an arbitrary choice that does not hold up in audit.

Failure to react to OOTs. The instrument goes OOT, the case is closed, but the interval is not changed. The same event will recur.

Review only "if needed". The periodic review must be planned, not triggered only when there are problems. A reactive system is not a managed system.

Confusion between nominal and actual interval. The instrument has a nominal interval of 12 months, but it is calibrated after 14 because "there was never a good moment". Late calibrations must be managed as formal exceptions, not as routine.

The Eurosystem experience

On the instrument fleets we manage, interval management is one of the processes that pays off most from having a dedicated system. Dimacs keeps the complete history of each instrument, automatically calculates the drift between successive calibrations, flags the cases where historical data would suggest a review, and traces every extension or shortening decision with an associated justification. It does not replace the metrologist's judgement — that remains human — but it eliminates the manual work of extracting and analyzing data that, on thousands of instruments, is materially impossible to do well with generic tools.