Skip to main content
This approach separates two fundamentally different questions: “How likely is the hazardous situation?” and “If it occurs, how likely is actual harm?” Keeping them separate produces more accurate risk estimates and makes the effect of specific risk controls more visible.

The Two-Factor Probability Model

Traditional risk matrices use a single probability axis. The P1xP2 model decomposes probability into two independent factors:
FactorQuestionScaleMeasurement
P1 — Hazard ProbabilityHow likely is the hazardous situation to occur?0-5Frequency ranges (1/100,000 to 1/1)
P2 — Harm ProbabilityGiven the hazardous situation, how likely is actual harm?0-5Percentage ranges (<=5% to >=96%)
diagram

P1 — Hazard Probability Scale

LevelLabelFrequency RangeExample
0UnanalyzedDefault for new records
1Very unlikely<= 1 in 100,000Rare manufacturing defect
2Unlikely<= 1 in 10,000Infrequent component failure
3Possible<= 1 in 1,000Occasional user error
4Likely<= 1 in 100Common operating condition
5Very likely<= 1 in 1Expected during normal use

P2 — Harm Probability Scale

LevelLabelPercentage RangeExample
0UnanalyzedDefault for new records
1Rare<= 5%Hazardous situation almost never causes injury
2Occasional6 - 25%Some chance of harm if situation occurs
3Probable26 - 75%Harm is likely given the hazardous situation
4Frequent76 - 95%Harm is very probable
5Almost certain>= 96%Harm is virtually guaranteed

Probability Bucketing

The raw P1 x P2 product ranges from 1 to 25. This product is bucketed into a 5-level composite probability scale:
Composite LevelLabelP1 x P2 RangeInterpretation
1Very Low1 - 3Negligible combined probability
2Low4 - 8Low combined probability
3Medium9 - 15Moderate combined probability
4High16 - 20High combined probability
5Very High21 - 25Very high combined probability
The bucket boundaries use orders of magnitude to compress the 25-point range into 5 actionable levels. The boundaries are not evenly distributed — the lower buckets cover fewer values (3 and 5) while the upper buckets also cover fewer values (5 and 5), with the middle bucket spanning 7 values. This reflects the reality that precision matters most at the extremes.

The 5x5 Risk Matrix

The composite probability (bucketed P1xP2) forms the Y-axis. Harm severity forms the X-axis. Their intersection determines the risk classification.

Harm Severity Scale

LevelLabelDefinitionDefault?
1NegligibleTemporary discomfort, no medical intervention
2MinorTemporary injury, no medical intervention required
3SeriousInjury requiring medical intervention
4CriticalPermanent impairment or life-threateningYes (conservative default)
5CatastrophicPatient death
Unassessed harms default to Critical (level 4), not Catastrophic. This is deliberate — it forces analysis while avoiding the alarm fatigue that a maximally pessimistic default would cause.

Risk Classification

The matrix intersection produces one of three risk levels:
ClassificationColorMeaningAction Required
AcceptableGreenRisk is tolerableNo further action needed
InvestigationAmberRisk is in the ALARP zoneRequires benefit-risk analysis or further reduction
UnacceptableRedRisk exceeds acceptable thresholdMandatory risk reduction required
The Investigation zone implements the ALARP principle (As Low As Reasonably Practicable) from ISO 14971. Risks in this zone are not automatically rejected — they require documented justification through benefit-risk analysis.

Pre-Mitigation and Post-Mitigation

Every risk record carries two complete probability assessments:
PhaseP1 FieldP2 FieldCompositeRisk Level
Pre-mitigationpreHazardProbabilitypreHarmProbabilitypreControlProbabilitypreRisk
Post-mitigationpostHazardProbabilitypostHarmProbabilitypostControlProbabilitypostRisk
This symmetry makes risk reduction visible. A risk control that reduces P1 (prevents the hazardous situation) shows a different pattern than one that reduces P2 (prevents harm given the situation). Auditors and risk managers can see exactly which factor each control affects.

How Risk Controls Affect the Matrix

Risk controls change the post-mitigation P1, P2, or both:
  • Inherent safety by design — Typically reduces P1 by eliminating the hazardous situation entirely
  • Protective measures — May reduce P1 (prevents exposure) or P2 (prevents harm given exposure)
  • Information for safety — Usually reduces P2 by enabling the user to avoid harm
The Risk Matrix Report generates two side-by-side 5x5 matrices from live data — one pre-mitigation, one post-mitigation — so the overall risk profile shift is immediately visible. See Read the Risk Matrix Report for details.