
Closing the Traceability Gaps That Derail Medical-Device Software V&V
A recurring QA problem in medical-device software is not the absence of testing, but the inability to prove it: you may have many test cases and results, yet still struggle to demonstrate that every requirement was verified, every risk control was implemented, and every change was assessed for impact. Industry guidance and regulations are commonly understood to expect structured verification and validation (V&V) evidence—planned tests, recorded results, and controlled approvals—plus traceability that ties user needs, requirements, risks, design outputs, and test evidence into one coherent story.
When that story is assembled late—just before a design review, an audit, or a technical documentation milestone—teams find missing links, ambiguous requirement scope, and retesting decisions made by intuition instead of impact analysis. This article focuses on practical decisions QA leaders can make to prevent traceability gaps from forming, and to surface them early when they’re cheapest to fix.
Define the V&V Objective in Operational Terms
V&V can be described as two complementary questions:
- Verification: did we build the software right against its specified design inputs and requirements?
- Validation: did we build the right software for intended use and user needs?
To make that actionable, convert it into an objective your team can execute:
- Each requirement has objective evidence of verification (and validation where applicable).
- Each risk control has evidence it is implemented and effective (as appropriate for the risk control type).
- Each change has documented impact analysis and a clear retest scope decision.
- Each test result is attributable, reviewable, and approval-controlled (including electronic signatures where required by your quality system).
This turns V&V from “a pile of test artifacts” into an auditable network of relationships and decisions.
Choose a Traceability Model Before You Choose a Tool
Teams often select a test management tool first and then attempt to “bolt on” requirement and risk traceability. A better order is:
- Decide the item types you must control (typical set below).
- Define the mandatory links between them.
- Define state transitions and approval rules (draft → reviewed → approved, etc.).
- Only then evaluate software support.
A practical baseline information model for medical-device software V&V:
- User needs / intended use statements
- System/software requirements (with stable IDs)
- Risk analysis items (hazards, hazardous situations, harms, risk controls) per your process (often aligned to ISO 14971 practices)
- Design outputs (architecture elements, units, interfaces) as needed for your lifecycle approach (often aligned to IEC 62304 practices)
- Verification test cases (manual/automated), test procedures, and test steps
- Test runs / executions and results
- Defects/anomalies and dispositions
- Releases/baselines (what exact set of items was tested)
Minimum link rules (examples):
- Each requirement links to ≥1 verification test case (or a justified alternative evidence record).
- Each risk control links to the requirement(s) and/or design output(s) that implement it, and to test evidence (or other verification evidence) that supports effectiveness.
- Each test result links to the exact test version and requirement version it claims to verify.
If your current process cannot answer “show me all tests and risk controls affected by this requirement change,” that’s a signal the model is incomplete or not being followed.
Build “Gap Visibility” Into Your QA Routine
Traceability problems become expensive when discovered late. The practical countermeasure is to create gap visibility as a routine QA signal—not an end-of-project scramble.
Implement two layers:
1) Real-time completeness checks (daily/weekly)
Create a short list of dashboards or reports your QA team reviews regularly:
- Requirements with no linked verification evidence
- Risk controls with no linked implementation or verification evidence
- Tests not linked to any requirement (often “nice-to-have” tests that don’t support V&V claims)
- Executed tests missing reviewer approval, missing objective evidence attachments, or missing environmental/configuration data
- Open anomalies that block requirements from being claimed “verified”
Even if you are using spreadsheets, you can approximate this with pivot tables and conditional formatting; however, spreadsheet-based approaches often struggle as volume grows or as many contributors edit in parallel.
2) Stage-gate checks (before formal reviews)
Before design reviews, verification readiness reviews, or submission packaging:
- Freeze a baseline of requirements and tests.
- Produce a trace report showing requirement → test case → last execution result.
- Produce a risk control evidence report showing risk control → implementation item → verification evidence.
Make this a checklist with an owner for each line item. When a gap is discovered, capture it as a tracked work item with a due date, not a “we’ll fix it later” note.
Make Change Impact and Retesting a First-Class Workflow
A major failure mode is treating retesting as ad hoc. When requirements change, teams often rerun a subset of tests based on memory, availability, or “what feels risky.” Instead:
- Require a change record that references the changed requirement(s) and design output(s).
- Require a linked impact analysis: list impacted tests, risk controls, and validation activities.
- Decide the retest scope using consistent criteria:
- Directly linked tests must be rerun unless there is a documented rationale.
- Indirectly affected tests (dependent requirements/interfaces) are evaluated based on dependency and risk.
- Automated regression suites may be mandatory for certain component types or safety classes per your internal policy.
This workflow matters regardless of tooling. Some V&V platforms claim they can automatically surface downstream impacts when an item changes; if you evaluate such a capability, test it using your own artifacts and realistic changes.
Treat Tool Evaluation Like a Mini-Validation Exercise
Medical-device teams are often expected to validate software tools used in the quality system to the extent appropriate. Without asserting any specific product behavior, a pragmatic approach to tool selection is to run a time-boxed evaluation using your own data and your own process expectations.
A 45-minute, scenario-based demo structure can expose whether a tool supports your needs:
- Import or create sample requirements with stable IDs.
- Link requirements to test cases and to risk controls.
- Execute a test run and capture step-level results and objective evidence.
- Introduce a requirement change and demonstrate impact analysis and retest selection.
- Produce trace outputs: requirements coverage, risk control evidence, and a test report suitable for your documentation format and approvals.
Make “show me, don’t tell me” the rule—especially for traceability and change impact.
Plan the Migration: IDs, Mapping, and Incremental Cutover
Most organizations already have artifacts in documents, spreadsheets, and issue trackers. Migration fails when teams attempt a “big bang” import without agreeing on identifiers and link mapping.
A safer migration plan:
-
Stabilize identifiers
- Assign immutable IDs to requirements and key risk items.
- If IDs already exist, preserve them.
-
Map columns/fields to item types
- Decide which spreadsheet columns become attributes vs. separate linked items.
- Decide what becomes a controlled field (e.g., verification method) vs. free text.
-
Trial import a small subset
- Start with a representative set (e.g., 50–100 requirements) including edge cases.
- Verify link integrity and reporting outcomes before scaling.
-
Incremental cutover
- Move active development first (where change impact matters most).
- Then migrate legacy artifacts needed for submissions/technical documentation.
This approach minimizes the risk of losing trace links or ending up with a tool full of untrusted data.
Hypothetical Example: A Requirement Change That Should Trigger Targeted Retesting
Hypothetical scenario: A software requirement for alarm escalation timing is modified to handle an additional edge condition.
Recommended QA flow:
- Update the requirement (new revision) and document the change rationale.
- Identify linked items:
- Verification tests directly linked to the requirement (alarm timing tests)
- Risk controls tied to alarm behavior (e.g., mitigations relying on timely alarms)
- Interface requirements affected (if alarm escalation crosses subsystem boundaries)
- Decide retesting:
- Rerun the linked alarm verification tests.
- Add/adjust a test for the new edge condition.
- Rerun a small regression set for related alarm routing components if dependencies exist.
- Capture results with configuration details (software build, test environment, timing measurement method).
- Update trace outputs so the new requirement revision is demonstrably covered.
The key is that the retest decision is trace-driven and reviewable, not based on memory.
Next Steps: A Practical QA Action List
- Document your traceability model (item types + mandatory links) in one page.
- Define completeness checks and review them on a recurring cadence.
- Standardize change impact analysis with a required template and approval step.
- Pilot tool support using your own data and change scenarios; focus on link integrity, impact visibility, reporting outputs, and approval controls.
- Plan migration incrementally with stable IDs and small trial imports.
- Train the team on “trace hygiene”: creating links is part of “done,” not an afterthought.
When these steps are in place, V&V becomes a predictable, inspectable process: you can show what was tested, why it was tested, what changed, and what you retested—without heroic effort at the end.


