thebench.bio #1 Ecosystem for the Bio-Tech Industry
All Clinical Data Management questions

Question 9 of 20 Fundamentals Practical

What is SAE reconciliation, and how is it done?

Short answer

SAE reconciliation compares serious adverse events in the clinical database with the same events in the safety database, so both tell the same story before analysis and submission. Mismatches in which events exist, onset dates, outcomes, seriousness or causality are queried and resolved, not overwritten.

Why

Serious adverse events travel two routes. The site enters them in the EDC for the clinical study report and also sends an expedited report to the safety team, which enters them in a safety database for regulatory reporting. The two copies drift: a follow-up report updates the outcome in the safety database but the case report form still says ongoing, or a hospitalization is in the EDC but never reached safety.

Regulators compare the study report with safety reports, so mismatches undermine both. E6(R3) section 4.2.6 lists reconciliation of relevant databases among the activities that finalize data before analysis. Reconciliation is done regularly during the trial, not only before lock, because a missing safety case may mean a missed expedited report.

Key fields are agreed in advance, usually participant, event term, onset date, seriousness criteria, outcome and causality. Each difference is resolved at its source. Data management never edits the investigator's data to match the safety database.

How to work it through

  1. Agree the fields to compare and the matching rules with the safety team in the data management plan.
  2. Extract SAEs from both systems on the same cut-off date.
  3. Match records by participant, event and onset date, allowing for coding differences.
  4. List events found in only one system and fields that differ.
  5. Raise queries to the site or the safety team, whichever owns the wrong record.
  6. Repeat on a set schedule and document the final reconciliation before lock.

An example

At a monthly reconciliation the EDC holds 52 SAEs and the safety database 54. Matching finds 50 pairs, 2 events only in EDC and 4 only in safety. One EDC-only event, a pneumonia admission from 3 weeks earlier, was never sent to safety, so it is escalated at once for expedited assessment.

Three safety-only events were entered as non-serious in the EDC, and the site corrects them. Among the 50 pairs, 6 differ on outcome date and are queried.

Where the answer stops holding

Matching on dates and terms fails when sites split or merge events differently, so some pairs need clinical judgment. Reconciliation shows the two systems agree, not that either is right. Non-serious events and events of special interest may need their own reconciliation rules.

A strong answer covers

  • Two systems, two purposes, one truth
  • Agreed key fields and matching rules
  • Regular reconciliation during the trial, not only at lock
  • Missing safety cases escalated for expedited assessment
  • Resolve at the source, never overwrite

Common mistakes

  • Doing reconciliation only once, just before database lock.
  • Editing EDC data to match the safety database without a site query.
  • Treating a MedDRA term difference as a mismatch when the event is clearly the same.