CAPA-071 effectiveness assessment with the available observation
Filled fictional example — adapt to your procedure and actual evidence.

SOURCE PACK
- Implementation evidence identifying the change boundary, the event definition and comparable eligible runs.
- The approved observation plan and criteria; retain the source periods supporting each total.

FINDINGS
Population assessed: January–March before the change; May–July after it. April is the changeover period.
Before: 12 events / 150 eligible runs × 100 = 8.00 events per 100 runs.
After: 5 events / 60 eligible runs × 100 = 8.33 events per 100 runs.
Criteria in this illustrative plan: At least 100 post-change runs and a rate no greater than 4 events per 100. These are example inputs, not a statistical recommendation.
Current assessment: The observation is incomplete and the available rate is above the example target. Fewer events do not establish improvement.

NEXT WORK AND CLOSURE EVIDENCE
Implementation owner
Action: Confirm which runs actually used the changed operation.
Evidence needed: Attributable implementation evidence and an explicit treatment of uncertain exposure.

Effectiveness reviewer
Action: Assess the incomplete observation and whether to continue, investigate implementation or reassess the action.
Evidence needed: A reasoned next step against the original plan; do not extend observation simply until a favourable rate appears.

CAPA decision owner
Action: Keep the action’s implementation and its effectiveness conclusion distinguishable.
Evidence needed: An authorised assessment with the population and limitations retained before any separate closure decision.

CONCLUSION
Effectiveness is not demonstrated by this population. Preserve the unfavourable and incomplete assessment. A later review must identify its new evidence and must not make this earlier review appear to have had it.

WORKED EVIDENCE REVIEW
Audit which events and runs belong in the calculation.
Before aggregating the six periods, join each eligible run to the implementation state of its equipment and to the defined leakage events. Keep a dated extraction and a reviewed list of inclusions, exclusions and duplicate reports. A maintenance note and an incident report may describe one leakage event; two separate leaks during one run may instead be two events. Decide that from the event definition and source evidence, not from how many records the search returns.
The measure here is events per 100 runs. It is not the percentage of runs with a failure: if two distinct events occurred in one run, that run contributes two events but one exposed run. Do not switch between those measures before and after the action. Also compare equipment and operating conditions across periods; a shift to easier work could change the aggregate independently of the CAPA. The observed 8.00 and 8.33 rates alone establish neither statistical significance nor causation.
For the illustrative plan’s minimum of 100 post-change runs, the available 60 leave the planned observation incomplete. Record the current result now, the outstanding exposure and any recurrence requiring action under the plan. Collecting the remaining 40 runs can complete the exposure requirement; it cannot erase the five events already observed or justify changing the original acceptance criterion after seeing them.
ICH Q10 §3.2.2 describes evaluating CAPA effectiveness in commercial manufacturing and relating investigation effort to risk. The event/exposure review here is one authored way to inspect a proposed effectiveness measure; Q10 does not prescribe this example’s 100-run minimum or numerical target.
ICH Q10: Pharmaceutical Quality System: https://database.ich.org/sites/default/files/Q10_Guideline.pdf
Illustrative inclusion decisions before calculating CAPA-071
Source situation: One leak appears in both maintenance and incident records
Numerator treatment: Count one event when evidence shows the same occurrence
Denominator treatment: Count its eligible run once
Source situation: Two distinct qualifying leaks occur in one eligible run
Numerator treatment: Count two events under the stated event-frequency definition
Denominator treatment: Count one run
Source situation: A run used equipment before its action was implemented
Numerator treatment: Assign using the predefined implementation/window rules
Denominator treatment: Do not silently place it in post-change exposure
Source situation: No exposure record can be recovered for a period
Numerator treatment: Keep the event evidence
Denominator treatment: Mark rate unavailable; do not substitute zero exposure or assume no events