The Dirty Dozen: Twelve Factors Hiding Behind "Human Error" in Every CAPA
Today, we sat in a round table discussion on CAPA management, and Talia Kleiner-Dayagi from Hishuley Carmel introduced a concept from the aviation world: The Dirty Dozen. For years, I've seen CAPA investigations stop at "human error" (various reasons) and close with retraining, and suddenly I had a structured name for everything that happens beneath that reason.
So I went to read about it in more depth, and I'm here to summarize what I found and how I think it can be used in medical devices.
Where it Came From
The list was developed by Gordon Dupont, an accident investigator at Transport Canada, in 1993.
After the Air Ontario Flight 1363 accident in 1989, Transport Canada decided to build a human factors training program for aircraft maintenance technicians.
Dupont and an industry steering committee received thousands of maintenance error reports from the Canadian military, filtered out those simply classified as "negligence," and analyzed what remained.
From this analysis, 12 recurring factors emerged. These are not types of errors; these are the preconditions in whose presence a skilled and competent person makes a mistake.
The point Dupont described is familiar to anyone who reads non-conformance reports: accidents repeated themselves, and only the dates, names, and places changed.
Since then, the list has become a fundamental part of human factors training in aircraft maintenance worldwide, and it is also migrating to other industries.
Why is this specifically relevant to CAPA?
The CAPA requirement (ISO 13485 clauses 8.5.2 and 8.5.3, and from February 2026 also the FDA's QMSR) states that the cause of the non-conformance must be determined and action taken to prevent recurrence. The problem is that in most investigations I see, the "reason" recorded is a re-description of the event in different wording.
The pattern repeats itself:
Description: The employee did not fill in a field in the DHR.
Root Cause: The employee did not follow the procedure.
Corrective Action: Retraining the employee, signing a training form.
Effectiveness Check: No similar non-conformance was received in three months.
Six months later, it happens again, to another employee, at the same station.
The reason is simple: "did not follow the procedure" is not a cause; it's the fact we already knew.
The real question is why a person who was trained, knows what the procedure requires, and wants to do a good job, still failed to do it at that moment.
The Dirty Dozen is the list of possible answers to that question.
It also has value with auditors, because when a Notified Body or FDA investigator sees a series of corrective actions that all end in retraining, it's a red flag in itself, as it shows that the CAPA system is not identifying root causes.
An investigation that points to an environmental or organizational condition and addresses it, even if the conclusion is less flattering to the company, looks much better in terms of system maturity.
The Dirty Dozen, in the language of a medical device company
For each factor, I've added how it looks in the field for us, and what usually hides behind it in an investigation.
Lack of Communication. The failure here is usually in information that was delivered but not received, and not in information that no one bothered to deliver. The classic example is shift changes in manufacturing, or information transfer between development and production in an engineering change. Information overload also falls here: if you updated a procedure and the notification of the change was sent in one email out of thirty that day, the information was delivered but did not reach its recipient.
Complacency. An action that is repeated hundreds of times without incident loses its status in the eyes of the performer. Receiving inspection of a component from a long-standing supplier, visual inspection at the end of a line, review of a batch record that looks like all previous records. The more the expectation of a finding decreases, the lower the chance of identifying it.
Lack of Knowledge. Beyond the absence of formal training, this factor also covers a situation where the employee knows how to perform the action but does not understand why it is critical, and therefore does not know how to identify when it deviates. An employee who knows they need to record temperature, but does not know what happens to the product if the temperature deviates, will not stop the process.
Distraction. The factor Dupont found most common in maintenance. For us, this happens every time someone is called in the middle of an action: a technician asked to help on another line, a quality manager entering a clean room to ask a question, a phone call in the middle of filling out a form. After a break, the natural tendency is to resume from a point later than where we actually stopped.
Lack of Teamwork. Occurs when departments have goals that pull in different directions. Manufacturing is measured by output, quality is measured by preventing deviations, and no one is measured by the joint outcome. In investigations, this appears as "we assumed they checked it."
Fatigue. Long shifts, night work, a week of closing batches before the end of a quarter. Fatigue primarily impairs the ability to detect that something is wrong, and this is the ability upon which all visual inspection and documentation control rely.
Lack of Resources. A tool sent for calibration and not returned, spare parts stock running out, a printed version of a procedure that was not updated, and especially lack of time and people. When resources are lacking, employees improvise, and most improvisations are neither documented nor checked.
Pressure. Real or imagined time pressure: a shipment that needs to go out, an audit approaching. Under pressure, people intentionally shorten processes, based on a quick judgment that the shortcut is safe. This is a conscious decision and therefore manageable.
Lack of Assertiveness. Someone saw something wrong and didn't say anything. This is the hardest factor to uncover in an investigation, because no one will raise their hand and say they remained silent. It appears especially in hierarchy gaps, when a junior employee identifies a problem with a manager or a more senior employee.
Stress. Pressure originating outside the task itself, such as company cutbacks or personal conflict within the team. It reduces the available attention capacity for the task.
Lack of Situational Awareness. The employee performed the action correctly but did not see its impact on something else. A change in a process parameter that affected a downstream process, a change in raw material that affected cleanliness. In our world, this often appears in changes that did not undergo a systematic impact assessment.
Norms. The way work is actually done, which is not the way written in the procedure. This is the most dangerous factor because it is learned from employee to employee, it seems legitimate because everyone does it this way, and it is completely invisible in document reviews. It is only exposed when observing people working.
What to Actually Do With It
This tool comes after, not instead of, structured root cause analysis.
After you have completed a 5 Whys or another root cause analysis and you conclude that a person is involved, go through the Dirty Dozen and ask about each one: Was it present here? In most cases, you will find two or three, not just one.
The following table is what I would include in the investigation form, because it translates the factor directly into a corrective action that is not training:
The Factor | What to Look For in the Investigation | Possible Corrective Action |
|---|---|---|
Lack of Communication | How information was delivered, and how receipt is ensured | Structured shift handover form, confirmation of receipt for procedure change |
Complacency | How many times the action was performed without a finding | Rotation among inspectors, targeted audit samples with known findings |
Lack of Knowledge | Does the employee understand the implication, not just the action | Adding rationale to the work instruction, performance-based certification |
Distraction | Was the action interrupted midway | Rule of no interruption at critical stations, marking a stop point on the form |
Lack of Teamwork | Who assumed someone else checked | Unambiguous responsibility definition at the interface, RACI for the process |
Fatigue | Actual working hours on that day | Limiting hours at visual inspection stations, splitting long tasks |
Lack of Resources | What was missing at the time of the event | Minimum stock for tools and materials, advance calibration planning |
Pressure | What was the schedule like that day | Separation of release authority from shipping authority, defined escalation |
Lack of Assertiveness | Did someone know and not report | Anonymous reporting channel, work stoppage procedure with authority for every employee |
Stress | Was there an unusual event in the company or team | Avoid placing employees in critical tasks during an unusual period |
Lack of Situational Awareness | Was an impact assessment performed for the change | Expanding change control form to downstream processes |
Norms | How work is actually done versus what is written | Observation of actual work, updating the procedure to reality or correcting reality |
Note that no action in the third column is "retraining."
Training only addresses factor number 3 (lack of knowledge), and even then, provided the training is different from what the employee has already undergone.
Where it Fits in the Quality System
There is no regulatory requirement here, and no standard mandates this list.
It's a thinking tool, so the cost of implementing it is small:
In the CAPA investigation form: One field where you mark which of the Dirty Dozen factors were identified, with a justification line. This is an addition of a field to an existing form.
In the CAPA procedure: A sentence stating that when an investigation indicates human involvement, the list of factors should be reviewed before determining the root cause. No new chapter is needed.
In an internal audit: When you observe a process, the Dirty Dozen are your list of questions. Especially factor number 12 (Norms), which is only revealed through observation.
In management review: Classification of non-conformances by factor. If ten corrective actions in a quarter point to time pressure, addressing this is within management's authority and requires their decision.
One Caveat
The list originated from the analysis of reports in aircraft maintenance, not from research on the medical device industry, and it is not a statistically validated taxonomy.
One can argue about the distinction between stress and fatigue or between complacency and lack of awareness, and this argument doesn't really matter.
Its value lies in giving the investigator twelve lines of inquiry instead of one.
I intend to try it in the upcoming investigations I oversee, and I'd be happy to hear from anyone who has already done so about how it worked for them.
Sources: SKYbrary, Human Factors "Dirty Dozen"; Integrated Human Factors, The "Dirty Dozen"; Aviation Safety Blog, Origin of Dirty Dozen
And thanks to Talia Kleiner-Dayagi.