← Free resourcesGet the toolkit

Free tool · NC treatment decision

The Proportionality Matrix

Not every nonconformity deserves a CAPA, and some need one without discussion. A common practice to rationalize your decision is the so-called risk-based approach. Traditionally, you look at a combination of occurrence and severity level, and based on the combination decide your next steps. Two values can’t cover everything. Once you pick a cell, you answer four more questions about how you found the NC, earlier investigations, related CAPAs and your risk management file. Your answers can raise the default.

You can deviate from the default, as long as you write down why.

Click the cell that matches your issue

* means the default response. You can decide differently, but only with a documented rationale.

Pick a cell to see the default treatment, what to document, and when a deviation is defensible. Then answer four more questions that can raise it.

Example definitions

Severity: product (patient & user impact)

LowNo impact on safety or performance: cosmetic, packaging blemish, labeling aesthetics. The device does what it should.
MediumPerformance, usability, or reliability could degrade. The device is still safe, but it no longer fully does what it should.
HighPlausible path to patient or user harm: a safety function is affected, or misdiagnosis or mistreatment becomes possible. Possibly reportable.

Severity: process & compliance

LowDocumentation slips or minor deviations from work instructions with no downstream effect.
MediumA requirement not fulfilled or a control skipped, with audit-finding potential. Caught before it affected product.
HighA QMS control failed systematically: release without approval, data integrity breach, or a gap an auditor would rate as major.

Occurrence: qualitative

LowUnlikely or rare: first time seen, or has rarely or never occurred in similar processes.
MediumHas happened before, or an early trend signal is visible.
HighRecurring pattern or systemic: almost certain to happen.

Occurrence: quantitative (example values)

LowSingle event, or < 0.1% of units/lots affected; no repeat within the last 12 months.
MediumRoughly 0.1–1% of units/lots, or 2–3 events within 12 months.
High> 1% of units/lots, or repeating monthly/quarterly despite corrections.

All definitions and thresholds are examples. Calibrate them to your production volume and risk acceptance criteria.

Calibrate this to your own system. The matrix helps you argue proportionality. It doesn’t replace your NC/CAPA procedure or your risk acceptance criteria (ISO 14971).
Every cell assumes a confirmed nonconformity that you correct. Each cell tells you whether corrective action should follow.

Take the toolkit with you

Landed on CAPA, or already have one open? The record still has to hold up in an audit, and the CAPA Stress Test toolkit helps you check that. Subscribers get:

Delivered by email, along with the MedTech CAPA Journal: frameworks for NCs and CAPAs that solve the problem and hold up in an audit. No fluff. Unsubscribe anytime.

Want to work on your NC/CAPA process together? See how I can support you