Review and evidence

A review decision only means something if it is tied to the exact data that was reviewed. Each review decision records the cohort, the source revision, the reviewer, the decision, the date it was made, the reason and an evidence reference. If the cohort changes after the decision, the change is recorded as a new revision, and the earlier approval covers only the revision it names.

The construction workflow opens a review decision with Decision = 'Pending' when the cohort needs approval. The Clinical Reviewer, who is not the cohort’s owner, reads the accepted facts, the calculated results and the attached evidence, then records Approved, Returned or Rejected with a reason. A returned revision goes back to the Outcomes Analyst with the reason attached. Only an approved revision can be published.

• Independent reviewer. The reviewer must be a different person from the cohort Owner, and the app records who decided.
• Evidence verified by the reviewer. Evidence documents move from Verified = 'No' to Yes and record who verified them.
• History that cannot be edited. State history keeps every status change with the actor, the revision, the operation key and the reason.

Agree which outcome definitions require independent review before publication and which do not. Agree who can act as a Clinical Reviewer for each organization, and whether a second reviewer is needed for particular measures. Decide what evidence a reviewer expects to see, such as a source extract log, a data-quality check or a clinical sign-off memo, and how each is referenced. Agree the reasons a reviewer may give for returning or rejecting a revision.

Take an analyzed cohort with a pending review decision. Sign in as the cohort Owner and try to approve it; the decision should be refused. Sign in as a Clinical Reviewer, open the evidence and mark the extract log verified. Approve the decision with a reason and an evidence reference, and confirm that the decided at date and reviewer are recorded.

Now ask the Outcomes Analyst to reject one more fact and record a new revision. A new review decision should be needed, and the earlier approval should still be visible against the old revision in the “Cohort state history” view. Publication should be blocked until the new revision is approved.

• Review decisions awaiting reviewer — pending decisions with the cohort, source revision, assigned reviewer and evidence reference.
• My review decisions — decisions the signed-in reviewer has made, with date, reason and evidence reference.
• Cohort state history — every status change with its actor, revision, operation key and reason.

• Independent review to publication: Approve the exact revision and publish its outcome rates.
• Run reconciliation and correction: Resolve failed runs and correct a published revision without losing the original.

Review the implementation plan and the data model before preparing your import files.