Are you sure you want to logout?

Confirm Cancel

Smart Patch Series – Build the Evidence While You Patch, Not When the Auditor Calls

18 November, 2026 | Post by

The SQA2 Blog: General

The problem: evidence that isn’t there when someone asks

Most teams verify their patches. The trouble is that the proof of that verification is scattered, temporary, or never written down. At patch volume, with dozens of small changes a quarter, each one is easy to check in the moment and hard to reconstruct later.

Evidence usually goes missing in a few predictable places:

  • Test results that expire. Build logs and pipeline runs are kept for weeks, not years.
  • Approvals in chat. A release was approved in a message thread, not in the system of record.
  • Screenshots without context. An image shows a passing screen but not the build, the environment, or the date.
  • An inventory that fell behind. A component was patched, but the software bill of materials (SBOM) still lists the old version.
  • People who moved on. The engineer who ran the tests and knew why a risk was accepted is now on another project.

The cost doesn’t show up on patch day. It shows up later, as an audit finding, an exception in a SOC 2 report, a customer security review that stalls a deal, or engineers pulled off current work to rebuild an old story.

Why the rules make evidence part of the job

For most health tech companies, the evidence questions come from software rules and from customers, not from device regulators. The details depend on what your product is, so confirm how each applies with your regulatory and legal advisers. Every source below asks the same thing: show the work.

Where the question comes fromApplies toWhat it asks you to show
HIPAA Security Rule, in force nowCovered entities and business associates, including most EHR and health software vendors that handle patient dataWritten records of required actions and assessments, kept for six years, and documentation that is reviewed when operational changes affect security.
HIPAA Security Rule update, proposed January 2025The same organizationsAs proposed: penetration testing at least yearly, scanning on a set schedule, and fixes within set timeframes. Final action is projected for July 2027.
ONC Health IT Certification ProgramDevelopers of certified health IT, including certified EHRsThat certified capabilities still meet the criteria as the product changes. The HTI-5 final rule (October 2026) cut 14 criteria and scaled back real-world testing, but kept the privacy and security criteria.
SOC 2 and HITRUST assessmentsHealth software vendors whose customers require themThat each change was authorized, tested, and approved. In a SOC 2 Type 2 audit, the auditor samples changes from the period and traces each one from ticket to deployment.
Health system security reviewsVendors selling to hospitals and health systemsHow patches are tested, approved, and released, often on a questionnaire at purchase and again at renewal.
FDA rules for software that is a medical deviceSoftware as a medical device and connected device softwareSection 524B patch and SBOM obligations, change control under the Quality Management System Regulation (effective February 2, 2026), and IEC 62304 maintenance.

The pattern holds even as the rules move. ONC’s certification requirements just got lighter and the HIPAA update keeps slipping, but customers and auditors still ask how each change was tested and approved. Evidence built into the work answers them no matter which rule is current.

How QA solves it: evidence built into every patch

The fix is to stop treating evidence as something gathered at audit time. When the record is produced by the same work that verifies the patch, it is complete on the day the patch ships. That is how our Quality Engineers set up patch work:

  1. Agree on the evidence standard before the work starts. For each kind of change, we define what counts as proof, where it is stored, who reviews it, and how long it is kept. Testers stop guessing, and reviewers stop rejecting work after the fact.
  2. Link every patch to one record. The advisory or finding, the change, the components and functions it affects, the tests, the results, the defects, and the approval all connect in one place. The component inventory tells us which product functions a patch touches.
  3. Let the pipeline write the record. Automated runs save their results with the build, the environment, and the date. Each fixed flaw gets a retest case that stays in the permanent suite, so the proof keeps running.
  4. Review the package before release. A Quality Engineer checks the record for gaps, such as a missing approval or a test that ran against the wrong build. Evidence is part of the definition of done, so a patch with an incomplete record is not done.
  5. Verify in production and record that too. After a staged release, we run smoke checks in the live environment and add the results to the record.
  6. Write it for the person who wasn’t there. Plain-language change summaries, runbooks, and handoff notes mean the record still makes sense after the people who did the work have moved on.

The result is an evidence package for every change. It shows:

  • what changed, and why
  • which components and functions it affects
  • which tests ran, and what they found
  • which defects were found, and how each was resolved or accepted
  • who approved the release, and when
  • how the release was verified in the live environment

AI can help assemble that package and summarize the results. AI accelerates execution, not judgment: a Quality Engineer decides whether the evidence is enough.

Why we treat evidence as a deliverable, not paperwork

We work in regulated environments, so we plan patch work around a question that always arrives eventually: can you show us? It rarely comes on patch day. It comes from an auditor, a new customer’s security team, or an engineer who joins next year, and none of them were in the room.

Evidence built this way pays off well beyond the audit:

  • Audits become retrieval. Pulling the record for a patch is a search, not a project.
  • Customer reviews move faster. Security questionnaires ask how changes are tested and approved, and a record answers with proof instead of a promise.
  • Assessments go more smoothly. A SOC 2 auditor’s sample, a HITRUST assessment, or a certification review draws on records that already exist.
  • The next patch goes faster. When a component changes again, the last record shows what it touched and which tests protect it.
  • Knowledge stays with the project. The record holds what people would otherwise carry in their heads, so a staffing change doesn’t erase it.

That is why a team with this habit also moves faster. It stops rebuilding the story each time someone asks.

Four questions to ask your team this week

  1. Pick a patch from last quarter. How long would it take to produce what changed, what was tested, the results, and who approved it?
  2. Where does the evidence for a patch live, and who owns it? If the answer is “several places,” that’s the gap.
  3. Is a complete evidence record part of the definition of done? If a patch can ship without one, some will.
  4. If the person who ran the tests left next month, could someone else find and explain the record?

If any answer gives you pause, start with the evidence standard. It is the cheapest step, and every other step depends on it.

The bottom line

In regulated health tech, a patch isn’t done when it ships. It’s done when you can prove how it was verified. The rules differ by product, but every one of them asks you to show your work, and the question usually comes long after the people who did the work have moved on.

QA closes that gap by making evidence a by-product of every patch: a standard agreed up front, one linked record, results the pipeline writes itself, a review before release, and verification in production. Build it that way, and an audit or a customer review becomes a retrieval exercise instead of a rebuild.

SQA² is QA-only. We don’t find vulnerabilities, write patches, or give legal advice. We verify that fixes work, that nothing else broke, and that the evidence exists. If you want a second set of eyes on how your team records its patches, our 30-minute strategy call is open to anyone who asks.

Part 5 looks at the seams where other people’s code meets yours.

This is a part of our overall Health Tech Patch Smart series.

Sources

Let's discuss how we can help you! GET IN TOUCH

Please to View This Content.

Not a Member? Register Now

Create New Account