Calibration Drift and the Report Due in Three Days
A water-treatment laboratory reports 1,200 turbidity readings per month to a municipal client. The line 2 inline analyser was due for its 12-month calibration on 1 March; today is 12 March and the calibration vendor cannot attend until 20 March. The analyser's diagnostics show 4.1% drift against its internal reference since the last verification, and the client contract requires reported data to come from instruments within their calibration interval. A handheld field meter with a valid certificate is available; its expanded uncertainty is plus or minus 0.5 NTU, against the analyser's plus or minus 0.1 NTU. Of the month's 1,200 readings, 240 fall in the 1-12 March window. Nine of those 240 exceed the 4.0 NTU permit threshold, and all nine lie between 4.1 and 4.6 NTU. The report is due 15 March. The lab manager says to backdate the calibration date because the vendor is booked anyway.
1. What can and cannot be reported from the line 2 analyser for 1-12 March? Show the reasoning for the nine borderline exceedances.
Read the worked answer
Neither instrument can carry those nine results as confirmed exceedances.
- Analyser drift: 4.1% applied at 4.0 NTU is about plus or minus 0.16 NTU before the analyser's own plus or minus 0.1 NTU is added. A reading of 4.1 NTU could plausibly sit near 3.94 or near 4.26. A reading of 4.6 is closer but still inside the combined band. On that evidence the nine are indeterminate, not proven exceedances.
- Handheld meter: plus or minus 0.5 NTU means a field reading of 4.2 NTU spans roughly 3.7 to 4.7 - straddling the threshold in both directions. It also cannot reconstruct the 1-12 March time series.
What to do. Report the 240 readings as qualified or estimated, with the drift magnitude and the out-of-interval status disclosed in the report body, not a footnote. Do not present the nine as confirmed exceedances and do not delete them. Re-sample those nine locations now using the handheld as a screen, and send confirmatory samples to a bench method. If the analyser can retain or re-run archived samples after the 20 March calibration, do that and amend. The contract requires data from in-interval instruments, so the client must hear about the qualification before the 15 March deadline.
2. How should the analyst handle the lab manager's instruction, and what should be recorded?
Read the worked answer
Do not alter the calibration record. Backdating a calibration date is a false record, not a clerical convenience: it breaches the contract's instrument condition, defeats the traceability an accredited laboratory is required to maintain, and creates a document that a client audit or regulator would treat as misrepresentation rather than error. The internal diagnostic log will contradict the altered date anyway, which converts a data-quality problem into a credibility problem.
Response. Decline, state the reason once in plain terms, and immediately offer the workable path: qualify the 240 readings, request a short reporting extension or issue the report with the qualification and an amendment to follow, use the handheld for screening, and ask the vendor to prioritise an earlier slot or send a substitute technician.
Recording. Enter the data-quality issue in the quality system with times, readings and the qualification applied. Record factually that an instruction to amend the calibration date was received and declined, without editorialising about motive. Escalate to the quality manager if the instruction is repeated. Separately check whether other instruments are near their intervals, since the same pressure will produce the same shortcut elsewhere.
3. Draft the data-qualification note for the client report and state what must accompany it in the client file.
Read the worked answer
Draft note. 'Turbidity results for line 2 covering 1-12 March 2025 are reported as qualified. The inline analyser's calibration interval expired on 1 March 2025 and recalibration is scheduled for 20 March 2025; the instrument's diagnostic recorded 4.1% drift over the same period. The 240 affected results should be read as indicative rather than confirmed, and the nine results between 4.1 and 4.6 NTU are not confirmed exceedances of the 4.0 NTU threshold given instrument uncertainty. Confirmatory sampling of those locations was carried out on 12 March with a calibrated field meter and samples were submitted for bench analysis. Amended results will be issued following recalibration.'
Accompanying file contents. The diagnostic drift printout; the calibration certificate history showing the lapsed interval; the certificate for the handheld meter; the re-sampling log and chain-of-custody forms; the bench results when returned; the quality-system entry recording the qualification decision and who approved it; and the correspondence with the client acknowledging the qualification and any agreed amendment date. The report and the file must reconcile exactly - any figure that appears in one and not the other will be read as concealment.