Originated by the United States Department of Defense (MIL-P-1629, 1949), now set by the AIAG and VDA FMEA Handbook, and set out for practitioners by Robin E. McDermott, Raymond J. Mikulak, and Michael R. Beauregard (The Basics of FMEA, 1996).
Lists the ways each part or step can fail, scores each failure for severity, likelihood, and detectability, and ranks the fixes by the combined score.
Four ways a ward's medication cabinet can fail, scored and ranked by risk number
Each failure mode is scored 1 to 10 on how bad its effect is, how often it happens, and how likely it is to slip past the checks (10 is almost never caught). The risk number is the product, and the list ranks them. Fix the top one (a barcode scan at the cabinet brings occurrence and detection to 2) and rescore: it drops to the bottom and the next one takes the top. Illustrative.
Nurse picks a look-alike drug from the next bin
SeverityOccurrenceDetectionRisk216Dose entered in milligrams where micrograms were meant
SeverityOccurrenceDetectionRisk140Override used at night with no pharmacist review
SeverityOccurrenceDetectionRisk200Label prints for the patient in the next bed
SeverityOccurrenceDetectionRisk63Highest risk number216
VerdictScore every row to see the verdict.WatchNothing above 100: keep the controls and rescore at the next review.Plan fixesGive the top failure modes owners and dates this quarter.Fix the top one firstA risk number above 150 needs an owner and a date now; fix it, then rescore it.
Highest risk first
- 216Nurse picks a look-alike drug from the next bin
- 200Override used at night with no pharmacist review
- 140Dose entered in milligrams where micrograms were meant
- 63Label prints for the patient in the next bed
An example
After two near misses in a month, the medication safety committee of a community hospital ran a failure mode and effects analysis on one step: nurses taking drugs from the automated cabinet on a medical ward. The team was a pharmacist, two night-shift nurses, the cabinet's technician, and a quality analyst.
They walked the step and listed every way it could go wrong, then scored each on severity, occurrence, and detection from 1 to 10. Four modes mattered. A look-alike drug in the neighboring bin scored 9, 4, and 6, a risk number of 216. Night overrides with no pharmacist review came to 200. A milligram and microgram mix-up was the most severe, but rare and partly caught, at 140. A label printed for the wrong bed came to 63. The committee funded barcode scanning at the cabinet first. Rescored a month later, the look-alike mode fell to 36, and the night override became the top risk, with an on-call pharmacist review as its fix.
Why it works
Teams fix the failure they remember, usually the last incident or the most frightening one. FMEA makes them look at every failure mode before choosing, and the three scores separate questions that memory blurs: how bad the effect is, how often the cause occurs, and how likely the failure is to reach the patient or customer before someone catches it.
Detection is the score that changes the ranking most often. A rare, severe failure that the existing checks catch can matter less than a moderate one that slips through unseen. Multiplying the scores gives one number for ordering the work, and rescoring after each fix shows whether it actually moved the risk or only felt like progress.
The thinking behind it
Robin E. McDermott The method began with the United States military, which set out a procedure for analyzing failure modes and their effects in 1949 (MIL-P-1629). Robin McDermott, a quality consultant, was lead author of The Basics of FMEA (1996), written with Raymond Mikulak and Michael Beauregard: a short practitioner's guide for teams running the analysis themselves. It covers listing how a product or process can fail, scoring each failure's severity, occurrence, and detection, and acting on the highest risks first.
Michael R. Beauregard Michael Beauregard, a coauthor of The Basics of FMEA, worked in quality engineering. The guide walks a team through the analysis in ten steps, short enough to keep beside the work. Its second edition (2009), from Productivity Press, was aligned with the automotive industry's 2008 AIAG guidelines; the joint AIAG and VDA FMEA Handbook now sets the standard for that industry.
How to run it
- Pick one process step or part. Keep the scope small enough to finish in two or three sessions.
- Gather the people who do the work. Include the front line and the night shift, who see failure modes managers never hear about.
- List every failure mode and its effect. For each, write what goes wrong, what it does downstream, and what causes it.
- Score severity, occurrence, and detection. Use 1 to 10 on agreed scales, where a high detection score means the failure is hard to catch.
- Multiply into a risk number and rank. Read the ranking alongside severity alone, so a rare catastrophic mode is never buried.
- Assign fixes to the top modes. Give each an owner and a date, and prefer fixes that make the failure impossible over warnings.
- Rescore after each fix. Score the mode again once the fix is in place, and let the next one up take the top.
Where it goes wrong
The most common failure is trusting the arithmetic too far. The scores are rankings, so a risk number of 200 is not twice as risky as 100, and a severity of 10 can hide behind low occurrence. Always look at the most severe modes on their own, whatever their product.
The second is an analysis done in a conference room. Without the people who run the step, the list holds the failures managers imagine and misses the workarounds. Walk the process where it happens.
The third is a spreadsheet that ends the effort. Teams finish the ranking and file it. Without fixes, owners, and a rescore date, the analysis changes nothing.
When to use something else
- Premortem. When the subject is a plan or project and the team needs a quick story of how it failed, without a full scoring.
- Poka-Yoke. When the top failure mode is known and the work is designing it out at the step where it happens.
- Hazard and Operability Study. When the system is a continuous process plant and deviations from design intent need a guided-word review.
Where it sits in SOLVE eX
Go to the source
- MIL-P-1629, Procedures for Performing a Failure Mode, Effects and Criticality Analysis, 1949
- The Basics of FMEA, Quality Resources, 1996