- Published
- Updated
Raising a threshold can reduce alert counts immediately. That reduction does not by itself show a more reliable process. A threshold revision changes how a risk becomes visible and should be treated as a management decision, with a clear rationale and limits on comparison with previous periods.
Scope of the decision
The proposal should consider false-positive workload, important events that might be missed and alternative monitoring methods together. Avoid selecting a new value merely to clear the current alert list. Record authority, effective date and the sites affected by the proposed change.
Worked review
Suppose a laboratory-delay indicator produces frequent alarms. The team replays historical examples using the proposed threshold and finds that some important delays would disappear. Instead of simply raising one value, it designs separate review routes based on impact and examines them during a limited pilot.
Practical checks
- Compare the old and new rules using the same historical data cut.
- Examine whether important events are among the alerts that disappear.
- Mark the threshold revision date and version explicitly on trend charts.
Acceptance and limits
Accept the new rule after evaluating both review workload and visibility of important events. Situations outside the pilot remain uncertain. Do not present rates before and after a definition change as one uninterrupted success curve; explain the contribution made by the changed rule.
Frequently asked questions
- Is changing a threshold always a mistake?
- No. It can support justified learning and proportionate oversight when its effects are evaluated.
- Should old alerts disappear under the new rule?
- No. Preserve historical decisions and document the rationale for any retrospective reclassification.