Failure mode and effects analysis (FMEA)
- Measure
Root cause analysis asks what went wrong. FMEA asks what could go wrong, before any patient is harmed, and ranks what to fix first.
What it is
Failure mode and effects analysis is a prospective, team-based method that maps a process step by step, lists how each step could fail (failure modes), what would happen (effects) and why (causes), and scores each failure to decide where to act first. The classic score is the risk priority number. Healthcare FMEA (HFMEA), developed by the US Veterans Health Administration's National Center for Patient Safety, combines FMEA with ideas from root cause analysis and uses a hazard score and a decision tree.
Formula and reading
- Formula: RPN = S × O × D, where S = severity of the effect, O = likelihood the failure occurs, D = likelihood it escapes detection.
- Reading: a higher RPN means a higher priority; compare RPNs within one analysis, not across different teams or scales.
- Steps: choose a high-risk process, gather a multidisciplinary team, map the process, list failure modes, score, act, re-score.
- Good actions lower O (redesign so it happens less) or D (make it visible earlier); severity usually changes only if the process changes.
- Other proactive methods: fault tree, event tree and bowtie analysis.
Worked example
- A made-up hospital will start a new high-alert IV infusion pathway. The team scores 3 failure modes, each S, O and D from 1 to 10.
- A: wrong concentration picked from the pump library. S 9 × O 3 × D 6 = 162.
- B: infusion line not labelled at the bag. S 5 × O 7 × D 3 = 105.
- C: rate keyed with a decimal-point error. S 10 × O 2 × D 8 = 160.
- Ranking: A (162), C (160), B (105). A and C are almost tied, and C has the maximum severity, so both get action first.
- Action for C: pump hard limits that block an implausible rate. Detection improves from 8 to 2: new RPN = 10 × 2 × 2 = 40.
- Action for A: a single standard concentration removes the choice. O falls from 3 to 1: new RPN = 9 × 1 × 6 = 54.
- Reading: the redesigned pathway's top RPN fell from 162 to 105 (B), which becomes the next target.
Prospective vs retrospective safety analysis
| Feature | FMEA (prospective) | Root cause analysis (retrospective) |
|---|---|---|
| When | Before a process runs or changes | After an adverse event or near miss |
| Starting point | A process map | A real event and its timeline |
| Question | How could each step fail, and how bad would it be? | What happened, why, and how do we stop it recurring? |
| Output | Ranked failure modes with RPNs and actions | Contributory factors and ranked corrective actions |
| Typical use | New infusion pumps, a new chemotherapy pathway | A wrong-site procedure, a fatal medication error |
Markers
- FMEA
- HFMEA
- RPN
- S × O × D
How to read it
- FMEA suits new or changing processes: a new device, a new ward, a new high-alert medicine or a move to a new record system.
- The team must include the people who do the work; managers alone miss how the process really runs.
- Re-score after the actions: an FMEA that ends with a list and no re-scoring is not finished.
Common traps
- A high detectability score is good: in FMEA a high D means the failure is hard to detect, which raises the risk
- FMEA is done after an adverse event: that is root cause analysis; FMEA is prospective, before harm
- The RPN alone decides priority: a catastrophic failure with low occurrence can have a modest RPN yet still needs action
- Two RPNs from different hospitals can be compared: scales are judged by each team; RPNs rank failures within one analysis
Test yourself
Think of the answer, then tap the question to see it.
A failure mode scores severity 8, occurrence 4 and detectability 5. What is its RPN, and how would a barcode check most likely change it?
8 × 4 × 5 = 160. A barcode check catches the failure before it reaches the patient, lowering detectability (for example to 2, giving 64).
A hospital buys new infusion pumps next month. Should it run an RCA or an FMEA, and why?
An FMEA: no event has happened yet, and the goal is to find how the new process could fail before patients are exposed.
Failure X has RPN 120 (S 10, O 2, D 6) and failure Y has RPN 150 (S 3, O 10, D 5). Why might a team still act on X?
X has maximum severity: if it happens it could kill. Many teams act on any catastrophic failure regardless of RPN, because the product can hide severity.
Why does removing a choice (one standard concentration) usually beat adding a reminder?
It lowers occurrence by design and does not depend on memory; a reminder is a weak barrier that fails when people are busy or tired.
Connections
Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.
Sources: Proactive Risk Assessment - VHA National Center for Patient Safety (US Department of Veterans Affairs, 2026); Root Cause Analysis - VHA National Center for Patient Safety (US Department of Veterans Affairs, 2025); Patient safety curriculum guide: multi-professional edition (World Health Organization, 2011).
Spotted a mistake?We are always improving for the community. Tell us what is wrong and our team will review it.Report a mistake
Representative model, for educational purposes only.
