Review Software Changes and Supplier Releases with Defensible Evidence
A change control worksheet and two synthetic examples for recording what a software change or supplier release touches, what you decided about its validation impact, and how the decision was followed through.
Written by CSV/CSA Maturity Editorial TeamPublished 12 min read
On this page9 sections
A software change is ready to rely on when your own records show four things: what changed, which of your functions and intended uses it touches, what assurance you decided was needed and why, and that the decision was carried through to closure. This holds for an internal configuration change and for a release a supplier schedules for you. Supplier release notes are one input to that record. They describe the product. They cannot describe your configuration, your process or your risk reasoning, so the validation impact decision stays with a named owner inside your organization.
This guide is for change owners, system owners and quality leads who handle changes to systems covered by computer system validation (CSV) or computer software assurance (CSA). CSV here means validation, not comma-separated values. It covers internal changes and supplier-initiated releases, including software as a service (SaaS) where the supplier sets the release date. It gives a review sequence, a worksheet and two synthetic examples. It does not prescribe one outcome for every deployment, and it does not decide which regulations apply to your systems.
Regulatory context, kept to what is relevant
Two sources frame the topic, each for its own setting.
FDA's final guidance on computer software assurance, issued on February 3, 2026, gives nonbinding recommendations for software used as part of medical device production or the quality management system (FDA guidance page [S02]). It describes assurance as keeping such software in a state of control throughout its life cycle, which it calls the "validated state" (FDA guidance, PDF [S03]). The guidance does not set out a general change control procedure. Its section on software changes deals with reporting for devices under a premarket approval (PMA) or humanitarian device exemption (HDE), so this guide does not use it as one.
For medicinal products in the EU, Annex 11 says changes to a computerized system should be made in a controlled manner in accordance with a defined procedure (clause 10) and that systems should be periodically evaluated (clause 11) (Annex 11 [S36]). As of November 12, 2026, the EudraLex Volume 4 index [S04] lists Annex 11 as the January 2011 revision.
Our reading is that neither source fixes how deep the review of a particular change must be, and that both expect the decision to be deliberate and recorded. The rest of this guide is about making it so.
Supplier release notes and your own impact decision
Release notes, a supplier's test summary and a change classification are useful. They come from the party that knows the product best. What they cannot contain is knowledge of how you use it.
| Topic | What supplier information can tell you | What only your own review can tell you |
|---|---|---|
| What changed | The features, fixes and known issues in the release, as the supplier describes them. | Whether each item is enabled, configured or used in your instance. |
| Classification | The supplier's label for the change across all its customers, such as "minor" or "no functional impact". | Whether the changed element is one your intended use depends on. |
| Testing | What the supplier tested, in its own environment. | Whether that testing covers your configuration, integrations, roles and data. |
| Timing | The release date and any preview period. | Whether you can assess before relying on the release, and what you do if you cannot. |
A supplier label is a prompt for where to look. It is never the decision. A change the supplier calls cosmetic can alter a screen your procedure depends on, and a change it calls major can touch a module you do not use. The assessment model makes the same point in its guidance for the supplier-release question: supplier evidence can be useful, and it does not remove the need to understand your own intended use.
Some releases arrive on the supplier's schedule with little or no preview. The model's statement scored 2 for supplier releases is that an owner reviews relevant release information and determines required actions "before reliance where practicable". Where that is not practicable, the defensible record says so, names who authorized continued use, and shows the review that followed and when.
A seven-step review sequence
1. Identify what changed
Describe the change in your own words from primary information: the change request, the configuration difference, the supplier's release notes and, where one exists, a preview environment. List discrete items. "Quarterly release" is not a description. "Approval routing now supports delegation, enabled by default" is. Record the source and date of the information you used.
2. Determine the affected intended use and functions
Map each item to the functions you rely on and to the intended-use statement for the system. Mark each item as affected, not affected or not yet known, with a reason for the first two. An item that touches nothing you use still gets a line, because the reason is the evidence.
If the change extends what the system is used for, the intended use itself changes. Update it before anything is tested against it.
3. Review supplier information
Read what the supplier provides against your map from step 2, not in place of it. Note what the supplier tested and under which conditions, any known issues, and what the notes leave out. Record what you reviewed, with its version or date, so that a later reader can retrieve the same information.
4. Decide what internal evidence is needed
For each affected function, decide what evidence would give confidence that it still does what you rely on it for. The depth follows the risk reasoning for that function. A change to a function whose failure could affect product quality may call for scripted retesting with independent review. A contained change to a lower-risk function may be covered by a targeted check with a short record. Connecting functional risk to an assurance activity describes how to make and record that choice.
Supplier evidence can reduce what you repeat, where you have checked that it applies to your use.
5. Document the assurance decision
Write the decision as a statement someone could disagree with: what will be done, what will not, and why. Name the owner and the approver, and obtain the approval before the change is implemented or relied on wherever the process allows. "No testing required" can be a defensible decision when the reason is recorded and the boundary is explained. Without the reason it is an omission.
6. Track open issues
Anything unexpected during the review or the testing gets an issue record with an owner, an assessed impact and a disposition. State whether the system may be used while the issue is open, and under what restriction. An issue raised with the supplier still needs an internal owner, because the supplier's ticket does not track your risk.
7. Confirm closure and continued use
Close the change when the planned evidence exists, issues have a disposition, and the records that depend on the change have been updated: the intended-use statement, affected needs, the links from needs to evidence with their versions, procedures and training material. Then confirm continued use in a sentence that names who decided.
After closure, check a sample of real use against what the impact review expected. Following one evidence chain after a change is a practical way to do that.
Change-review worksheet
| Worksheet field | What to record | Prompt for the reviewer |
|---|---|---|
| Change reference and source | Identifier, internal or supplier-initiated, date notified, planned date. | Was this known before it took effect? |
| What changed | Discrete items in your own words. Source and version of the information. | Could someone else list the same items from the same sources? |
| Affected intended use and functions | Each item mapped to functions as affected, not affected or not yet known, with reasons. | Does any "not affected" rest only on a supplier label? |
| Supplier information reviewed | Documents read, their version or date, what they cover and what they leave out. | Does the supplier's testing match your configuration? |
| Internal evidence needed | The activity for each affected function and why that depth fits. | Would the activity reveal the failure you are concerned about? |
| Assurance decision | What will and will not be done, the rationale, owner, approver and approval date. | Was approval given before reliance? If not, who authorized that? |
| Open issues | Issue reference, owner, impact, disposition and any restriction on use while open. | Does each issue have an internal owner? |
| Records to update | Intended use, needs, evidence links and versions, procedures, training. | Do the links point to the new evidence version? |
| Closure and continued use | Closure date, who confirmed continued use, the post-change check and its result. | Did actual outcomes match the expected impact? |
Two synthetic examples
| Worksheet field | Example 1: presentation-only change | Example 2: workflow-affecting change |
|---|---|---|
| Change reference and source | REL-24.2, supplier-initiated. Notified three weeks ahead, with a preview environment. | REL-24.3, supplier-initiated. Notified two weeks ahead, with a preview environment. |
| What changed | Navigation menu regrouped. Status labels restyled with new colors. Supplier label: "UI refresh, no functional change". | Approvers can delegate approval to another user, and delegation is enabled by default. Supplier label: "minor enhancement". |
| Affected intended use and functions | Approval, routing and record content: not affected, confirmed by comparing fields, statuses and permissions in preview. Boundary: label wording and status meanings are unchanged; only position and color differ. Work instruction screenshots: affected. | Approval authority: affected. The intended use says only users in the approver role may approve, and delegation would let a user outside that role approve. Approval history and audit trail: not yet known. |
| Supplier information reviewed | Release notes, with their date recorded. No test summary offered for this item. | Release notes and a supplier test summary. The summary covers delegation between approvers, not delegation to other roles. |
| Internal evidence needed | Targeted check in preview of the two screens used for approval: same fields, same statuses, same actions for each role. Short record with tester and date. No retest of the approval workflow, with the reason stated. | Decide first whether delegation is wanted. If so: scripted negative tests for each role on who can receive and use a delegation, a check that a delegation expires, and a review of how a delegated approval appears in the approval history. Independent review. |
| Assurance decision | Proceed. Rationale: no change to data, workflow, permissions or defaults, and status meaning does not depend on color alone. Owner: system owner (role). Approver: quality lead (role). Approved before release. | Disable delegation at release. Adopt it later only after the intended use and the approval-authority need are updated and the tests pass. Owner: system owner (role). Approver: quality lead (role). Approved before release. |
| Open issues | None. | ISS-031: in preview, a delegated approval shows the delegate's name without the delegator's. Owner: system owner (role). Raised with the supplier. Delegation stays disabled while the issue is open. |
| Records to update | Work instruction screenshots and a note to users. Evidence links unchanged, with the reason recorded. | Configuration baseline (delegation off). If adopted later: intended use, need wording, risk rationale, test evidence version and its links. |
| Closure and continued use | Closed after release. The system owner confirmed that the two screens in production match preview. | Closed for this release with delegation off, confirmed in production. Adoption is tracked as a separate change. |
What the examples show
In Example 1 the supplier's label and the team's conclusion agree, but the conclusion is the team's. It rests on a boundary that was checked: no change to data, workflow, permissions or defaults, and status meaning that does not depend on color alone. If statuses in this organization were told apart only by color, the same release would affect a function and the decision would differ.
In Example 2 the supplier's label is "minor" and the team's conclusion is that the change reaches approval authority, which is one of the things the system is relied on for. Disabling the feature is this team's choice, not the only defensible one. Another organization might adopt delegation at release with the tests complete. Leaving it on by default because the notes called it minor would not be defensible.
Automation supports the review and does not replace it
Automated regression tests, release-monitoring feeds and tools that flag changed components can shorten steps 1, 3 and 4. They are worth having where they are kept current. Their limits are specific:
- An automated suite covers what it was written to cover. A new feature such as delegation has no test until someone decides it needs one.
- A passing run shows that existing checks still pass. It does not show that the intended use is unchanged.
- The suite needs an owner who reviews its scope when the system changes.
So the impact decision and the confirmation of continued use still carry a person's name. The assessment model takes the same position from the other side: it gives no points for a tool as such, and its guidance says manual updating can be mature when ownership, review and evidence are reliable.
How the worksheet maps to the assessment
In model version csa-maturity-2.0.0, the change management practice area has three questions, each answered with one of four practice statements scored 0 to 3 or with "I do not know yet".
| Question | What a 2 describes | Worksheet fields that show it |
|---|---|---|
| Q11 | "Changes have an appropriate pre-implementation impact review, responsible owner and approval." | Affected intended use and functions; assurance decision |
| Q12 | "An owner reviews relevant release information and determines required actions before reliance where practicable." | Supplier information reviewed; assurance decision and its approval date |
| Q13 | "A defined review approach examines changes, issues and continued suitability at appropriate intervals." | Closure and continued use, read across several changes |
A 3 on Q11 describes a process that checks whether actual outcomes matched the expected impact, which is what the last worksheet prompt asks. Q13 concerns review over time: the completed worksheets for a period are a natural input to a review of continued suitability. The model does not prescribe a review frequency.
In the assessment's published worked example, which is a synthetic profile, this area scores 44.4. Q11 and Q12 are answered 1 and Q13 is answered 2, which is 4 of the 9 available points. The score describes reported practices for a chosen scope and nothing more.
You can answer the change management questions for your own scope. The assessment asks for a scope, takes you through the seven practice areas and a review step, then asks for your first name, last name, work email and company before it shows the profile, which you can download as a PDF. The result is a self-reported profile for planning, not an audit or a certification.
Limitations
- The worksheet is a working aid. Your change procedure governs, including how emergency changes are authorized and followed up.
- A complete worksheet shows that the decision can be reconstructed. It does not show that the decision was correct or that the testing was sufficient.
- The FDA guidance cited here is nonbinding and addresses software used in medical device production and quality management systems. Annex 11 is EU GMP guidance for medicinal products. This guide makes no determination about which applies to you.
- Supplier qualification and the content of supplier agreements are outside this guide.
Related reading
- Risk-based software assurance: a worked decision record: how to set and justify the depth of evidence that step 4 asks for.
- Review validation traceability with one evidence chain: how to check that links and evidence versions were updated when a change closed.
- Build a system inventory you can maintain: a change that alters how a system is used may also alter its inventory entry and relevance decision.
This article describes assessment model csa-maturity-2.0.0.
Sources and references
- S02Computer Software Assurance for Production and Quality Management System Software. U.S. Food and Drug Administration.
Final guidance (nonbinding), dated February 2026; covers software used in medical device production and quality management systems.
- S03Computer Software Assurance for Production and Quality Management System Software: Guidance for Industry and Food and Drug Administration Staff. U.S. Food and Drug Administration.
Final guidance document (PDF), issued February 3, 2026; nonbinding; supersedes the final guidance issued September 24, 2025.
- S04EudraLex - Volume 4 - Good Manufacturing Practice (GMP) guidelines. European Commission.
Official index of EU GMP guidelines; listed Annex 11 as 'Computerised Systems (revision January 2011)' when checked on November 12, 2026.
- S05Stakeholders’ Consultation on EudraLex Volume 4 - Good Manufacturing Practice Guidelines: Chapter 4, Annex 11 and New Annex 22. European Commission, Directorate-General for Health and Food Safety.
Consultation on draft texts, open July 7 to October 7, 2025; status shown as closed. The drafts are not operative guidance.
- S36EudraLex Volume 4, Annex 11: Computerised Systems. European Commission.
EU GMP guideline, revision 1, with a stated deadline for coming into operation of June 30, 2011; the version linked from the EudraLex index on November 12, 2026.