QC laboratory
LIMS: specifications, samples and QC results
Every in-process, release, stability and raw material result, recorded against an approved specification on instruments in calibration and standards in date, calculated and judged by the system, reviewed by a second person and approved by QA.
In the portal: /app/lims
From specification to certificate
- E-signature
1. Approve the specification
Tests, methods, limits; one version in force
- Person
2. Register the sample
Lot, testing stage, sampling point; tests copied from the specification
- Person
3. Receive the sample
Chain of custody: time, condition, storage
- Person
4. Record each result
Determinations as read, instruments, standards, raw data
- System
5. Calculate and judge
Rounding, limits, out of specification and out of trend
- E-signature
6. Second-person review
Nobody who recorded a result
- E-signature
7. QA approval
Conclusion set by the system; Certificate of Analysis
- An out-of-specification result stands until an investigation linked to the sample finds an assignable cause and invalidates it; a passing repeat never replaces it.
Open LIMS. The overview shows samples in testing, ready for review, in review, with an open out-of-specification result or overdue, testing by stage, instruments due for calibration, standards close to expiry, and the data integrity indicators across all samples.
Specifications
- Step 1: Choose New specification. Enter the title, specification number and version, the material and its code, the material type, and the testing it governs: In-process control, Release, Stability, Raw material acceptance, Investigation or retest, or Other.
- Step 2: Attach the registered specification, monograph or analytical method under Documents and choose Draft the tests with AI, or add each test with Add a test. Tests the AI drafted are marked Drafted by AI with the words they came from; check every one.
- Step 3: For each test give the method and version, the technique and the result type: Numeric, Conforms or does not conform, or Text (for information). A numeric test also has its unit, the decimals reported, the number of determinations, whether the mean or each determination is reported, the specification limits (inclusive, or exclusive with More than or Less than), optional alert limits for out-of-trend results, a maximum %RSD and a limit of quantitation.
- Step 4: Tick what each result must name: the instrument used, the reference standard used, the raw data.
- Step 5: Route the specification for Specification approval. A draft never routed is edited without reasons; after routing every change needs one.
To change an approved specification choose Revise: create version 2 (or the next number). The new version starts as a draft with the same tests; the current version stays in force until the new one is approved. Samples keep the version they were registered under, and their results are never judged again against a later version.
Instruments and calibration
- Register an instrument with its asset ID, name, type, manufacturer, model, serial number, location, whether it needs periodic calibration and how often, and its qualification reference. Tick Already qualified and in service for an instrument in use; otherwise it starts as Awaiting qualification.
- Record a calibration: the date, who performed it, the result (Pass (as found), Pass after adjustment, or Fail), the next due date, the certificate number, and the certificate under Documents.
- A failed calibration takes the instrument out of service at once and lists the samples with results on it since its last passing calibration, for the impact assessment.
- Void… a calibration entered in error, with the reason. It stays on the record, struck through.
- Change the status to In service, Out of service or Retired with the reason. An instrument that needs calibration goes back into service only with a passing calibration covering the day.
Rule: A result is refused when an instrument named on it was not in service, or not covered by a passing calibration, on the day of the analysis.
Reference standards and reagents
- Register a standard or reagent: name, kind (primary, secondary or working, impurity standard, reagent, volumetric solution, culture or media), supplier, catalogue number, lot, potency, storage, received and opened dates, and the expiry or retest date.
- It starts in quarantine. Tick Release it for use now (the certificate is checked) when the certificate has been checked.
- Statuses are Quarantine, Released for use, Depleted and Discarded, each change with a reason. A standard past its expiry or retest date cannot be released until the date is extended, with a reason.
Rule: A result is refused when a standard named on it was not released, or was past its expiry or retest date, at the time of the analysis.
Register and receive a sample
- Step 1: Choose Register a sample and pick the approved specification in force for the material and testing stage. Its tests are copied onto the sample as they stand.
- Step 2: Enter the lot or batch, the sampling point, when and by whom the sample was taken, the quantity and container, storage, the results due date and the priority. Stability samples also carry the study, condition and time point.
- Step 3: Record the receipt at registration, or later under Chain of custody with Record the receipt: the time, condition on receipt and storage location. Results can be recorded only on a received sample.
A test beyond the specification is added to the sample with Add a test and a reason. A test not performed is marked Not tested… with the reason; a test with an out-of-specification result cannot be dropped.
Record results
- Step 1: On the test choose Record the result. You are the analyst: the result is saved under your name with the system time.
- Step 2: Enter when the analysis was performed: not in the future, not before the sample was taken or received. A result recorded more than 24 hours after the analysis needs the reason for the late entry.
- Step 3: Type each determination exactly as read, with a point as the decimal separator, or <LOQ when below the limit of quantitation. The form shows the reportable value and the %RSD; the system judges it against the criterion only when it is saved.
- Step 4: Tick the instruments and standards used. Those that cannot be used today are shown with the reason and cannot be ticked; the server checks each one again for the time of the analysis.
- Step 5: Tick the raw data files already attached, or choose Attach raw data file: each file is fingerprinted (SHA-256) when it arrives and is ticked for this result.
- Step 6: Choose Save the result.
How results are calculated and judged
- Determinations are kept exactly as typed and calculated in exact decimal arithmetic, so 1.005 rounds to 1.01.
- The reportable value is the mean of the determinations, or each determination, rounded to the reported decimals (half up, USP General Notices 7.20) before it is compared with the limits.
- A %RSD above the maximum, or any determination outside the limits when each is reported, makes the result out of specification.
- Values below the limit of quantitation are reported as below it; a mean of values above and below it is refused.
- A result inside the specification but outside the alert limits is out of trend, and needs a written evaluation before review.
Rule: The server decides whether a numeric result conforms when it is saved, from the determinations and the limits. The form only previews the calculation.
Out of specification, corrections and repeats
- Step 1: When a result is out of specification the sample shows it in red. Choose Open an OOS investigation: the investigation opens with the result, the determinations, the instruments, the standards and the analyst filled in, linked to the sample both ways.
- Step 2: Only an assignable cause found by that investigation invalidates the result: choose Invalidate, give the reason and choose the investigation. The result stays on the record, struck through, with the reason.
- Step 3: Choose Record a repeat and give the reason (the investigation step it follows). Repeating a test whose out-of-specification result still stands needs the investigation linked.
- Step 4: Only the analyst who recorded a result can Correct it, with a reason; the old and new values stay in the audit trail. An out-of-specification result cannot be corrected into compliance.
Rule: A sample with an out-of-specification result standing cannot be cancelled, and goes to review only once the investigation that invalidated a result is approved.
Review, approval and the certificate
Before routing for review lists what is missing: the receipt, a valid result for every test, the raw data each test requires, an evaluation for each out-of-trend result, an investigation for each out-of-specification result. Assign the reviewers and approvers and route the sample for Results review and approval. Nobody who recorded a result on the sample can review or approve it.
At QA approval the system sets the conclusion from the results: Meets specification, Does not meet specification, or Results reported for a sample without a specification. The Certificate of Analysis is then produced from the approved record with the signatures; before approval only a draft certificate is available. Every export is recorded in the audit trail.
Reports and trends
Open Reports. Laboratory indicators for 30 days, 90 days or 12 months count results, out-of-specification and out-of-trend results, invalidations, late entries and corrections, and show right-first-time approvals, turnaround from receipt to approval by testing stage, overdue samples, and results by test, analyst and instrument.
Result trend draws the reported result of each sample for a test of a specification, oldest first, as a control chart: limits from the moving range, Nelson run rules, the specification and alert limits, and capability. Tick Every valid result to include repeats. Save to charts keeps it with your charts; Results as CSV downloads the rows. Statistics and BI tools read the same results through GET /api/v1/lims/results.
Rule: Trends support review and continued process verification. The approved record of each sample remains the basis of its disposition.
Connected systems (interfaces)
Every other system connects through a controlled interface, under Interfaces. An instrument data system (a CDS or instrument software) sends results with their raw data and reads the worklist of open tests. An ERP or manufacturing system requests samples for a lot and reads the dispositions of approved samples. A document repository opens document references, such as the controlled document a specification implements. The quality system is built in: investigations open from the sample or the instrument.
- Step 1: An administrator registers the interface: its kind, the system and version, the owner and what it exchanges. It starts in validation and is not heard.
- Step 2: A super administrator creates an API key with the LIMS interface scope under Settings and binds it to the interface.
- Step 3: For an instrument data system, assign the instruments it may report for. A result naming any other instrument is refused.
- Step 4: Record the validation reference, then make the interface active with the validation as the reason.
- Step 5: Watch the message log: every message is accepted, rejected with the reason, or recognised as a duplicate of one already accepted. Rejections on an active interface raise an alert.
A result received through an interface is checked exactly like a typed one: calibration, standards, late entry, out-of-specification rules. It is attributed to the analyst it names and shows Received from the interface and the message on the sample. It can only be corrected by the system that sent it. A sample requested by an ERP shows Requested by with the ERP reference and still needs receipt in the laboratory.
Tip: Suspend an interface (with a reason) during an upgrade of the connected system; it is not heard until it is revalidated and made active again.
System architecture
System architecture shows the LIMS in layers, each with its live state: the people who work in it, the browser interface, the application layer from samples through tests and results to review, the central laboratory database with its audit chain verified on opening, and reporting. Beside them are the systems connected through controlled interfaces, with their status and last message.
A failed calibration lists the results recorded on the instrument since its last good calibration. Open a deviation investigation from that list starts an investigation pre-filled with the instrument and every sample to assess, linked to the instrument.
Data integrity
| Principle | How the LIMS meets it |
|---|---|
| Attributable | Every result under the signed-in analyst's own account; reviewers and approvers sign with their password |
| Legible and enduring | Results are kept on the record with their history; nothing is deleted |
| Contemporaneous | System time stamps; analysis time checked; late entries justified |
| Original | Determinations as typed; raw data fingerprinted and linked to each result |
| Accurate | Calculated by the system; instruments in calibration and standards in date checked at the time of analysis |
| Complete and consistent | Invalidated results kept with the reason and the investigation; every change in the hash-chained audit trail |
The data integrity panel on each sample and the overview count late entries, invalidated results, corrections and out-of-specification samples without an investigation. Audit trails for the whole laboratory are under Audit trails.