1. Define one monitoring question and quantity
Ask what change the chart should expose. Reference EC50, control recovery, stable-control relative potency, slope, replicate CV, and signal window monitor different parts of an assay. Name one quantity, unit, transformation, direction, and interpretation per chart.
Do not chart product samples that intentionally differ and interpret their differences as process drift. Prefer a stable control, reference parameter, or normalized metric whose expected behavior matches the monitoring question.
2. Admit only comparable run records
Define the method and settings version, assay conditions, control or reference lot, matrix, instrument state, result status, and time order required for a record to enter the sequence. Preserve failed, invalid, repeated, and excluded runs even when the approved calculation omits them.
Each point links to its source analysis, report, suitability outcome, exclusions, analyst, instrument, material lots, and software version. Stratify incompatible populations or annotate justified changes instead of widening one baseline until unlike runs appear stable.
3. Approve the baseline, calculation, and rule
Select a stable representative period, define baseline membership, calculate the center and dispersion by the approved method, and retain the values, formulas, effective date, and approver. A later baseline creates a new version rather than rewriting the limits shown with earlier points.
Control limits summarize expected baseline variation. Method acceptance limits or product specifications answer a different question and remain separately labelled. Define the exact sequence logic and response for each monitoring rule before applying it.
4. Generate a visible, reproducible signal
When a point or pattern meets a rule, display the rule version, involved runs, observed values, baseline and limits, calculation time, and active filters. The same retained records and versions should reproduce the flag.
A signal begins the defined review; it does not silently invalidate a run, change a disposition, exclude a point, or prove a cause. Per-run suitability and longitudinal monitoring remain separate decisions.
5. Investigate the signal with its run context
Compare relevant reagent, reference, cell, and control lots; analysts; instruments and maintenance; plate patterns; incubation and preparation details; method and software changes; prior deviations; and neighboring runs. Record competing explanations and the evidence used to test them.
Keep annotations and investigation notes linked to the signal without altering the source result or baseline. If no assignable cause is found, retain that conclusion and any approved monitoring response.
6. Retain the decision, action, and follow-up
Record triage, investigation owner, disposition, correction or preventive action, affected records, baseline decision, approval, and follow-up effectiveness. If a new baseline is justified, preserve the bridge, effective point, and old baseline.
Review false-signal burden and missed or late signals periodically. Change the quantity, baseline, or rule only through a versioned assessment; do not tune history until prior flags disappear.
Limits and current product boundary
Levey–Jennings, Shewhart, SPC, and Westgard-style rules are possible monitoring methods, not interchangeable universal requirements. The linked guides explain their distinct questions and assumptions.
Current Team trending is bounded to method, date, and metric views with one explainable rule-based signal. Broader rule libraries, customer-defined multirules, automated investigation workflows, and assay-specific decision criteria are not current standard capabilities.
Frequently asked questions
Are Westgard rules automatically appropriate for potency assays?
No. They were developed in a particular laboratory QC context. Any transferred rule needs a justified charted quantity, error model, false-signal assessment, and procedural response.
Is every point outside a control limit an invalid assay?
No. A statistical signal indicates behavior that merits evaluation. Validity and release decisions follow the approved method and quality procedure.