Skip to main content
The final quality gate for every AI-assisted medical communications deliverable. Before any content is submitted, shared with a client, or published, it must pass a structured human review that confirms accuracy, completeness, compliance, and quality.
Risk tier: Critical  ·  Review requirement: This workflow IS the review. The reviewer must be qualified for the content type and its risk tier.You, as the named reviewer, are accountable for the deliverable’s quality — regardless of how it was produced. AI assistance does not transfer accountability.

​
What AI does in this workflow

Final review is fundamentally a human activity. AI plays a supporting role only:
  • RefCheckr — automated reference accuracy check as one input to the accuracy verification
  • MedCheckr — automated compliance signal scan for promotional content
  • General LLM assistance — generating a structured comparison between the deliverable and source materials to support manual verification
AI support tools are one layer of the review. They do not replace the reviewer’s contextual judgement, completeness assessment, or accountability.

​
Who should perform this review

  • A qualified medical writer or reviewer who has not been the sole author of the deliverable
  • Someone with the appropriate therapeutic area and content-type knowledge
  • For regulated content: the named reviewer responsible for quality sign-off
Where possible, have someone other than the primary author perform the final review. Confirmation bias — reading to confirm rather than to challenge — is a significant risk when the reviewer wrote the content.

​
Before you start

  • Obtain the near-final version of the deliverable and all source materials used in its development.
  • Review the workflow documentation: which AI tools were used, which workflows were followed, and where the highest-risk steps occurred. This tells you where to focus your verification effort.
  • Have the applicable review checklist, approved messaging framework, and any claims matrix to hand.
  • Confirm you have full-text access to every cited reference. If any reference is unavailable, flag all claims it supports as unverified and resolve before sign-off.
Do not start final review without full-text access to all cited references. Reviewing accuracy without sources is not possible.

​
Steps

1

Read the deliverable in full

Before checking details, read the entire piece to assess overall quality, flow, and coherence. Does it achieve its stated purpose? Is it right for the audience and channel? Address structural or narrative issues before line-level checking.
2

Check the workflow trail

Review which AI workflows and tools were used to produce the content. Higher-risk workflows (plain language summaries, claims verification, compliance checks) require closer scrutiny in the corresponding sections of the deliverable.
3

Verify accuracy against sources

Use RefCheckr for a first-pass automated check, then manually verify every factual claim, data point, and conclusion against the cited sources. Automated tools are one layer — contextual accuracy requires human assessment.
4

Run compliance pre-screening

For promotional or compliance-sensitive content, use MedCheckr as one input to your compliance assessment. Review all flags in context against the applicable advertising code.
5

Complete the review checklist

Work through the full checklist below systematically. Do not skip sections under time pressure — the checklist exists because reviewers under pressure skip the sections where errors hide.
6

Assess AI-specific risks

Specifically check for: hallucinated content or citations, meaning drift from AI transformation, unsourced background claims, and merged findings from different study arms or sources.
7

Make corrections and document changes

Fix issues found. Track every change. If significant changes are made to claims or data, re-verify the affected sections.
8

Final sign-off

If the deliverable meets all requirements: sign off with your name, role, and date. If issues remain: return for revision with specific, actionable feedback. Document the outcome either way.

​
AI support prompt

Use this pattern to generate a structured comparison between the deliverable and its source materials as one input to your review. 
You are a medical writing QC assistant. Your task is to help me review the following deliverable against its source materials.

Deliverable:
[INSERT DELIVERABLE TEXT]

Source materials:
[INSERT SOURCE TEXTS]

Please check:
1. Does every factual claim in the deliverable have support in the source materials?
2. Are all data points (numbers, statistics, endpoints) accurately reproduced?
3. Are there any statements in the deliverable that appear to go beyond what the sources support?
4. Are qualifiers present in the sources (subgroup, post-hoc, exploratory, specific population) preserved in the deliverable?
5. Is safety information present and proportionate to the efficacy content?

For each issue found, quote the relevant text from the deliverable and explain the concern.

Note: This is a support tool for my review. I will verify all findings manually.

​
Review checklist

Work through every section. Mark each item as confirmed or document the issue found.

​
Accuracy

  • Every factual claim is supported by the cited source
  • All numerical data matches the source exactly
  • Study design, populations, and endpoints are correctly described
  • Conclusions match the authors’ stated conclusions
  • No hallucinated content, data, or citations

​
Completeness

  • All key messages from the brief are addressed
  • Safety information is present and proportionate
  • Limitations are noted where appropriate
  • Qualifiers are preserved (subgroup, post-hoc, exploratory)

​
Compliance (where applicable)

  • Claims are within the approved messaging framework
  • Fair balance is maintained
  • No off-label implications
  • References are correctly cited
  • Prescribing information requirements are met

​
Quality

  • Language is appropriate for the specific audience and channel
  • Structure and narrative flow support the deliverable’s objective
  • Length meets the brief requirements and is proportionate across sections
  • Medical terminology is used correctly and consistently
  • No grammatical, spelling, or formatting errors — including reference numbering, table formatting, and figure callouts

​
AI-specific checks

  • Sections that used AI assistance are identified and specifically reviewed
  • No unsourced content from AI training data
  • No meaning drift from AI transformation or adaptation
  • No merged findings from different sources or study arms
  • Automated verification results (RefCheckr, MedCheckr) have been reviewed and acted on

​
Documentation

  • AI tools and workflows used are documented
  • Changes made during review are tracked
  • Review outcome is recorded with reviewer name and date

​
Review outcomes

The output of this workflow is a review decision — not a content deliverable.
OutcomeMeaning
ApprovedThe deliverable meets all requirements. Documented with reviewer name and date.
Approved with minor changesSmall corrections identified and made. Changes documented. Deliverable proceeds.
Returned for revisionSignificant issues identified. Specific, actionable feedback provided. Returns to writer.
RejectedFundamental accuracy or compliance issues. Full rework required. Reasons documented.

​
Common failure modes

RiskWhat to look for
Review fatigueReviewing multiple AI-assisted documents in one session — use the checklist for every review and limit AI-assisted QC to 3 documents per session
Over-reliance on automated toolsRefCheckr returns no flags, reviewer moves on — automated tools check patterns, not contextual accuracy or whether the right reference was cited
Confirmation biasReviewer wrote the brief and expects the content to be correct — approach final review as adversarial verification. Assume errors are present
Incomplete source accessReviewer has abstract only, not full text — full-text access to all cited references is a prerequisite, not optional
Accountability gapMultiple contributors and no clear owner — the reviewer who signs off is accountable, regardless of how many people touched the document

​
Relevant tools

RefCheckr

Supporting tool — automated reference accuracy check as one input to the review.

MedCheckr

Supporting tool — automated compliance signal scan for promotional content.
Final review applies to every AI-assisted deliverable. The most common paths leading here:

Verify Claims

Reference accuracy check before final review.

Check Compliance

Compliance pre-screening before final review for promotional content.

Plain Language Summary

PLS requires expert final review before publication.