This is a ready-to-use log for a batch record reviewer calibration session, the exercise where multiple reviewers independently review the same record so that inconsistency between reviewers becomes visible and fixable instead of staying invisible until an inspector or an incident surfaces it. Replace every <<FILL: ...>> placeholder. A filled specimen follows.
Session record
| Field | Entry |
|---|---|
| Session ID | <<FILL: e.g. CAL-2026-Q3-01>> |
| Session date | <<FILL>> |
| Trigger | Scheduled quarterly / New reviewer first 90 days / Divergent rejection or finding rate observed / Procedure change |
| Facilitator (not among reviewers calibrated) | <<FILL>> |
| Participating reviewers | <<FILL: list names>> |
| Batch record(s) used | <<FILL: batch/lot number(s)>> |
| Records carry known, already-closed findings | Yes / No |
| Independent review conducted without prior discussion | Yes / No |
Findings matrix
List every finding any reviewer raised, known or emergent, and mark each reviewer’s outcome.
| Finding | <<FILL: Reviewer 1>> | <<FILL: Reviewer 2>> | <<FILL: Reviewer 3>> | <<FILL: add columns as needed>> |
|---|---|---|---|---|
<<FILL: finding 1>> | Caught / Missed | Caught / Missed | Caught / Missed | <<FILL>> |
<<FILL: finding 2>> | Caught / Missed | Caught / Missed | Caught / Missed | <<FILL>> |
<<FILL: finding 3>> | Caught / Missed | Caught / Missed | Caught / Missed | <<FILL>> |
<<FILL: add rows for every finding, including anything a reviewer raised that was not a known finding>> | ||||
| Individual catch rate | <<FILL: n of total (%)>> | <<FILL>> | <<FILL>> | <<FILL>> |
| Group agreement rate (findings every reviewer caught, of total) | <<FILL: %>> |
Divergence analysis
For every finding where reviewers did not agree, record the cause and the resulting action. A session with no divergence recorded is treated as a facilitation defect (reviewers likely compared notes before submitting), not as a clean result, unless the facilitator documents specifically why none occurred.
| Divergent finding | Reviewer(s) who missed it | Cause traced to (ambiguous procedure / technique gap / other) | Action | Owner | Due date |
|---|---|---|---|---|---|
<<FILL>> | <<FILL>> | <<FILL>> | <<FILL: procedure/checklist fix under change control, or documented retraining>> | <<FILL>> | <<FILL>> |
<<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> |
Sign-off
| Field | Entry |
|---|---|
| Session summary reviewed with all participants | Yes / No |
| All divergences traced to a cause and assigned an action | Yes / No |
| Facilitator (name, signature, date) | <<FILL>> |
| QA management approval (name, signature, date) | <<FILL>> |
Cumulative trend log (append one row per session)
| Session ID | Date | Records used | Reviewers | Group agreement rate | Divergences with actions closed | Notes |
|---|---|---|---|---|---|---|
<<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> |
<<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> |
Acceptance criteria
A calibration program is defensible when: a defined cadence exists and sessions actually occur on it; the records used are representative and, where feasible, carry known findings; every session produces a documented matrix, individual and group catch rates, and a divergence analysis; every divergence traces to a stated cause (procedure or technique) and closes with a recorded action; and the cumulative trend log shows the group agreement rate tracked across sessions over time, not a single session filed and forgotten.
Retention
Retain calibration session records per <<FILL: retention period>>, per the records retention schedule, separate from the batch record reviews themselves.
References
21 CFR 211.192 (production record review by the quality control unit; the basis for expecting a consistent, defensible standard across reviewers). ICH Q10, Pharmaceutical Quality System (management review and continual improvement of the quality system, which a reviewer calibration program feeds). ICH Q9 (quality risk management, for prioritizing which records and how often to calibrate against).
Confirm the current version of each reference before issue.
Filled specimen
The following shows a completed session record for three reviewers calibrated against one batch record carrying four known, previously closed findings. The company, batch, and numbers are illustrative.
Session record:
| Field | Entry |
|---|---|
| Session ID | CAL-2026-Q3-01 |
| Session date | 20 August 2026 |
| Trigger | Scheduled quarterly |
| Facilitator | J. Reyes, QA lead |
| Participating reviewers | A. Okoye, B. Lindqvist, C. Marsh |
| Batch record(s) used | B-7605 (carries four known findings from a prior real investigation) |
| Records carry known, already-closed findings | Yes |
| Independent review conducted without prior discussion | Yes, reviewed separately during the same 90-minute window, no discussion before submission |
Findings matrix:
| Finding | A. Okoye | B. Lindqvist | C. Marsh |
|---|---|---|---|
| Missing second-check signature on step 12 | Caught | Caught | Missed |
| Granulation yield 9 percent loss, below the 5 percent action limit, unflagged in the record | Caught | Missed | Caught |
| Deviation referenced in comments, no deviation record on file | Caught | Caught | Caught |
| Audit trail gap, 40-minute system-disabled window | Missed | Caught | Missed |
| Individual catch rate | 3 of 4 (75%) | 3 of 4 (75%) | 2 of 4 (50%) |
| Group agreement rate | 1 of 4 findings caught by all three (25%) |
Divergence analysis:
| Divergent finding | Reviewer(s) who missed it | Cause | Action | Owner | Due date |
|---|---|---|---|---|---|
| Audit trail gap | A. Okoye, C. Marsh | Technique gap: both skipped opening the audit trail as a routine step | Targeted coaching on the “open the audit trail every time” step; add as a mandatory line item on the reviewer checklist | J. Reyes | 15 September 2026 |
| Granulation yield below action limit | B. Lindqvist | Procedure gap: checklist did not force a lookup against the documented action-limit table, so the value was eyeballed as “close enough” | Update the batch record review checklist to require the actual action-limit value be recorded, not just a pass/fail judgment | J. Reyes | 15 September 2026 |
Sign-off:
| Field | Entry |
|---|---|
| Session summary reviewed with all participants | Yes |
| All divergences traced to a cause and assigned an action | Yes |
| Facilitator | J. Reyes, signed, 20 August 2026 |
| QA management approval | S. Whitfield, signed, 21 August 2026 |
The value of this session was not that reviewers disagreed, disagreement was expected and is why the session exists, but that both divergences traced to concrete, fixable causes: one technique gap addressed with coaching, and one checklist gap addressed with a procedure fix that benefits every reviewer, not just the one who missed it this time.
Common inspection findings this log prevents
- A calibration program named in a procedure but with no session records showing it was ever actually run.
- Reviewers compared notes before submitting, producing a session with no divergence and no evidence the exercise tested anything.
- A divergence identified but never traced to a cause or closed with an action, so the same gap recurs at the next session.
- Individual reviewer performance discussed informally with no documented record, leaving no trend to show whether calibration is actually improving consistency over time.
How to adapt this log
- Set your cadence and trigger list in the session record to match your team size and risk profile.
- If your batch record system supports seeding known findings into a copy of a real record for training purposes, state that method here; otherwise use closed real records with their findings already known to the facilitator.
- Extend the findings matrix and divergence analysis for however many reviewers participate.
- Keep the cumulative trend log current after every session so the group agreement rate is genuinely trended, not just reported once.
- Confirm every regulation in the References section against the current published version before issue.