Document rendition verification checks whether an output such as a PDF faithfully represents the identified source for its intended use. A successful Word export does not establish that the right revision was converted, every page is present or an important symbol survived. Verify the final file that will actually be issued, not only the Word source or the export settings.
This checklist is for pharmaceutical document-control, regulatory-operations and quality teams producing static PDFs from Word. It separates content fidelity from approval status and destination-specific technical requirements. It does not establish that a static PDF replaces a dynamic laboratory record, or that a verified document is an accepted regulatory submission.
Define the verification record first
Record the source document ID, controlled revision, intended use, output filename and repository version, conversion application and version, platform, export route, material settings, operator and conversion time. Name the reviewer and the authority that will accept the rendition. Capture a hash or another controlled identifier when your process uses one to distinguish the exact reviewed output.
A hash can help show that a file has not changed since verification. It does not prove that the PDF means the same thing as the Word document, and the source and PDF are not expected to have matching hashes.
Specify the expected output: clean approved instruction, marked review copy, submission document or another defined purpose. If the destination, source revision or intended representation is unknown, resolve it before selecting technical criteria. Do not infer “final approved” from a filename.
For every applicable row below, record Pass, Fail, Unknown or Not applicable, an evidence location, reviewer and date. Pass means the criterion was met; Fail means observed evidence does not meet it; Unknown means evidence is missing or inconclusive. Not applicable needs a specific reason accepted by the designated reviewer. Mandatory Fail or Unknown results block acceptance for the stated use.
Source and conversion checks
| ID | Observable acceptance criterion | Evidence to inspect | Owner |
|---|---|---|---|
| S1 | Source is the exact authorized revision for this conversion | Controlled source record and conversion input identity | Document controller |
| S2 | Source approval or review status fits the intended output | Approval record or authorized review-copy instruction | Document owner |
| S3 | Tracked changes and comments have an authorized disposition | Source review state and the agreed clean/marked-copy requirement | Document owner |
| S4 | Expected sections, appendices and included objects are defined | Source inventory and intended output scope | Content reviewer |
| S5 | Conversion environment and settings are identifiable | Application/version, platform, route, page range and settings record | Conversion operator |
| S6 | The file opened for verification is the recorded output | Repository version or controlled identifier matched to review record | Verification reviewer |
Microsoft documents Word export options for page ranges, markup, bookmarks and other PDF properties. Its desktop, Mac and web guidance describes different routes and option availability. Record the route actually used; do not assume another platform's settings were applied. Microsoft PDF conversion guidance, Word PDF conversion on Mac
One concrete reason to record the route: Microsoft's guidance identifies a Word for Mac issue in which hyperlinks may not convert correctly when using Best for printing. If links are required, inspect them in the final output from that route. Do not mark the link checks complete because another platform's export passed. Microsoft's Mac export note
Hiding tracked changes does not remove them from the Word document. For a clean controlled output, establish that the source's changes have been resolved through the authorized review process. Do not let the conversion operator accept every change merely to make the file look clean. A marked review PDF can legitimately retain markup when that is its declared purpose. Microsoft: Track changes in Word
Content and presentation checks
| ID | Observable acceptance criterion | Evidence to inspect | Owner |
|---|---|---|---|
| C1 | All intended sections and appendices are present in order | Source-to-output section reconciliation | Content reviewer |
| C2 | No content is clipped, overlapped or obscured | Page review, including page boundaries, tables and figures | Content reviewer |
| C3 | Critical numbers, units, signs and symbols retain their meaning | Direct comparison of defined critical values and expressions | Technical reviewer |
| C4 | Tables retain correct row/column association and footnotes | Table-by-table comparison, including continued tables | Technical reviewer |
| C5 | Figures and legends remain readable and associated | Output at intended reading scale and appropriate zoom | Content reviewer |
| C6 | Headers, footers, identifiers and status labels are correct | Output across section breaks and page orientations | Document controller |
| C7 | Fields, numbering, references and contents entries are consistent | Output entries checked against actual referenced content | Content reviewer |
Page count is a useful signal, not sufficient proof. Reflow may change pagination legitimately, while a missing appendix can be hidden by an extra blank page. Reconcile the content, then explain any expected pagination difference.
Choose critical comparisons from the document's use: a dose unit, temperature range, inequality, decimal, mathematical subscript or footnote can matter more than a cosmetic line break. Define the coverage before review. If sampling is justified for a qualified conversion process, record what was sampled, why and what risk it covers; do not report unexamined content as verified.
Microsoft notes that fonts that cannot be embedded may be substituted, or represented as bitmap text depending on the selected option. Check both appearance and required text functionality rather than assuming a visually similar page satisfies every use. Microsoft export font options
For EU human-medicinal GMP documents, Chapter 4 section 4.2 addresses avoiding errors when reproducing working documents from masters. The comparison method above is an operational way to examine rendition fidelity; Chapter 4 does not prescribe these checklist IDs. EU GMP Chapter 4, January 2011
Function, approval context and destination checks
| ID | Observable acceptance criterion | Evidence to inspect | Owner |
|---|---|---|---|
| F1 | Required text search and selection work accurately | Search/select representative critical text and inspect result | Verification reviewer |
| F2 | Required bookmarks navigate to correct destinations | Bookmark labels, hierarchy and actual landing points | Publishing reviewer |
| F3a | Required internal hyperlinks resolve to intended content | Open internal links in the final PDF | Publishing reviewer |
| F3b | Required external hyperlinks resolve to intended accessible targets | Open external links in the intended delivery context | Publishing reviewer |
| F4 | The rendition preserves required approval information | Source approval context and output manifestations, where applicable | Quality/document controller |
| F5 | Properties, annotations and embedded content match sharing scope | PDF properties, annotations and attachment inspection | Document owner |
| F6 | Applicable destination technical criteria are satisfied | Identified specification/version and technical-check result | Publishing or records owner |
| F7 | The accepted file is the exact file issued or handed off | Final repository identity and release/transfer record | Document controller |
Under Part 11 where applicable, signed electronic records have specified signature manifestations, including signer name, signing date/time and meaning; those manifestations are subject to requirements for human-readable forms. A pasted signature image alone does not establish the electronic-record and signature controls. Identify whether the PDF is a rendition of an approved record, a separately signed record or a review copy. 21 CFR Part 11, sections 11.50 and 11.70
For an FDA eCTD destination, use the applicable current submission specifications and validation criteria in addition to this checklist. FDA's eCTD v3.2.2 standards page currently lists PDF specifications version 4.1, with support beginning September 21, 2016; the PDF's revision history dates version 4.1 September 14, 2016. Those specifications address matters such as security settings, fonts and navigation; they are not a universal EDMS export profile. Do not add password protection or select a PDF variant merely because it sounds more secure or archival. Apply the actual destination criteria. FDA eCTD v3.2.2 standards, FDA PDF specifications
FDA's drug-CGMP data-integrity guidance distinguishes accurate copies preserving content, meaning and relevant metadata from incomplete representations. A static PDF may omit functionality or information needed from a dynamic electronic record. Passing these Word-to-PDF checks is not authorization to discard that source or other required evidence. FDA data-integrity guidance, questions 9 and 10
Worked verification with a failure and an unknown
Use this fictional example to practice the decision. It is a documentary exercise, not a claim of testing Word or a particular EDMS.
The authorized source is method M-730 revision 5, intended as a clean reference rendition. Its approved content includes Appendix A, a four-column results table and the critical expression “≤ 0.10 mg/mL.” The review record identifies PDF output R-1 and requires content fidelity, searchable text and working internal links. External hyperlinks are absent. No regulatory submission destination has been assigned.
| Check | Observed evidence in the example | Result | Action / owner |
|---|---|---|---|
| S1 source identity | Conversion input record identifies M-730 revision 5 | Pass | Retain input identity with R-1 |
| C1 completeness | Section reconciliation shows Appendix A is absent | Fail | Operator corrects conversion scope and creates a new output |
| C3 critical expression | Comparison shows the exact approved symbol, number and unit | Pass for recorded comparison | Technical reviewer retains comparison evidence |
| F1 search | Reviewer finds critical expression and confirms selected text is accurate | Pass for recorded text check | Preserve result and coverage |
| F3a internal link | Link to the missing appendix cannot reach intended content | Fail | Recheck after regeneration |
| F3b external links | Source inventory confirms no external hyperlinks | Not applicable | Record absence; internal-link failure remains |
| F4 approval context | Required approval record was not supplied to reviewer | Unknown | Controller obtains evidence; do not infer approval from footer |
The outcome is not accepted. Content fidelity failed and required approval evidence is unknown. The three recorded passes cannot outweigh either block, and this partial example does not imply the remaining checklist rows passed.
The operator fixes the export scope and produces R-2. The reviewer confirms Appendix A is present, checks the internal link, obtains the required approval evidence and completes every other applicable row. Because regeneration creates a new output, the review covers R-2's identity and the defined verification scope; it does not merely attach R-1's passes to a different file. Once all applicable mandatory criteria pass, the authorized reviewer accepts R-2 for the declared reference use and records the exact issued file.
Change the scenario: R-2 is complete and attractive, but the conversion input was revision 4. S1 fails, even if its filename says revision 5. Correct the source selection and repeat the defined checks. A correct-looking export of the wrong source is still the wrong rendition.
Also challenge the critical expression itself. If the output changes “≤ 0.10 mg/mL” to “< 0.10 mg/mL,” C3 fails because the boundary meaning changed, even though the number and unit match. If the display looks correct but copying the expression yields a different symbol, the visual comparison and F1 text-function check have different results. Record both and resolve the failure for the declared use; do not let one successful view conceal an incorrect text representation. These values are synthetic document-checking content, not a recommended analytical limit.
Resolve discrepancies without losing traceability
Log the affected check, source location, output location, observed difference, consequence and correction owner. Distinguish a source-content issue from a conversion defect. A reviewer who finds an incorrect source value should route it through document correction or change control; silently editing the PDF would create an unexplained divergence.
After corrections, retain the relevant failed-output evidence according to your process, identify the new candidate and rerun the affected checks plus any checks impacted by regeneration or changed settings. If a new converter or platform is introduced, assess the process change rather than assuming earlier assurance automatically covers it.
At handoff, state the accepted purpose and remaining limits. “Verified reference rendition” does not mean “approved for operational use,” and document-level verification does not replace package-level submission validation. The eCTD validation guide covers that next boundary. Keep the source, rendition, review evidence and final delivery identity connected so another reviewer can reproduce the decision.
About the author
Assyro Team
Expert regulatory operations consultants helping pharmaceutical companies navigate complex compliance challenges.

