Reconciliation
After extractors complete their work, a quality assurance step reviews the extracted data before it is finalized. Which step you use depends on your project's screening type:
- Single DE + QC - a QC reviewer audits completed extractions for accuracy
- Double DE + Reconciliation - a reconciler compares the two independent extractions and resolves any differences
The QC and reconciliation steps in data extraction mirror the quality assurance steps in the screening phases. The workflow is the same, but applied to extracted field values rather than include/exclude decisions.
QC review (single DE)
In single-extraction projects, a QC reviewer checks completed extraction forms for errors, omissions, or inconsistencies. The QC reviewer has read access to the extraction form and the study PDF.
QC workflow
Open the QC queue
Navigate to Data Extraction and open the QC Review tab. This lists all studies that have been marked complete by an extractor and are awaiting review.
Open a study for review
Click a study to open the side-by-side view. The extraction form appears on the right, filled in by the extractor. The PDF is on the left.
Review each field
Work through the form, comparing the extracted values against the source PDF. Verify that numbers are accurate, text entries are complete, and dropdown selections are appropriate.
Approve or flag for revision
If the extraction is correct, click Approve. If you find issues, use Flag for Revision to send the study back to the extractor with a note explaining what needs to be corrected.
Flagging for revision
When you flag a study for revision, you can leave a comment on specific fields to explain the problem. The extractor receives a notification and the study reappears in their extraction queue. After they resubmit, the study returns to the QC queue for you to re-review.
Be specific in revision notes. Instead of "check sample size," write "the sample size on page 3 is 240, not 204 as entered." Clear notes mean fewer revision cycles.
Reconciliation (double DE)
In double-extraction projects, two extractors independently complete the form for each study. The reconciler reviews the two versions side by side and resolves any fields where the values differ.
What triggers reconciliation
A study enters the reconciliation queue when both extractors have marked it complete. Fields where both extractors entered the same value are auto-accepted. Only fields with differing values require the reconciler's attention.
Reconciliation workflow
Open the reconciliation queue
Navigate to Data Extraction and open the Reconciliation tab. Each study in the queue shows the number of fields with differences.
Open a study
Click a study to open the reconciliation interface. Fields are grouped by section, with differing fields highlighted.
Review differences field by field
For each differing field, you see the value entered by Extractor A and the value entered by Extractor B. The study PDF is available in the left panel for reference.
Select or enter the final value
Click Accept A or Accept B to adopt one extractor's value, or type a different value entirely if neither is correct.
Mark as reconciled
After resolving all differing fields, click Mark as Reconciled. The study moves to the completed extraction set.
Marking incomplete for revision
If the reconciler identifies an error in a field that both extractors agreed on, they can mark the study as incomplete and send it back to one or both extractors for correction. Use the Send Back for Revision option and specify which extractor should correct the issue.
Reviewing finalized extractions
Once a study is approved (QC) or reconciled (double DE), it appears in the Completed tab of the Data Extraction section. The final extracted values are locked and available for export.