FR

Human error expertise · SENTINELLE method

Reducing human error in pharmaceutical production

Errors that come back deviation after deviation, retraining as the only response, checks added for reassurance: human error isn’t inevitable. We identify its real causes and put in place on the shop floor the rituals and practices that prevent it.

Results measured on our assignments

A ranger watches over the workshop while the team removes the warning flags−47 %human-error deviations, in one year
Two production teams in front of their boards of handled weak signals−46 %human errors across two production units
An overflowing in-tray next to a tidy one−27 %deviations, all causes combined
An operator points at a box label while her colleague confirms−76 %human-error deviations in packaging

Results measured on several production sites.

Raising awareness isn’t enough to eliminate human error.An error isn’t the fault of the person who makes it: it’s the symptom of causes that can be addressed.

85% of errors come back after 6 months following awareness training, a document change or an added check. Adding checks doesn’t strengthen control: the double signature creates an illusion of safety.

Do you recognise your human errors?

  • Human errors that come back, deviation after deviation.
  • Retraining operators as the first action, every time.
  • Checks added for reassurance, and a double signature where each person relies on the other.
  • CAPAs that make the process heavier without addressing the cause.
  • Weak signals that never come up from the shop floor.
  • The belief that there will always be some.

The 5 causes of human error, and the 3 factors that change everything

A human error always has an explanation. Our DTMIC tool looks for it in five broad families of causes.

  1. D
    Documentation

    Having the activity described isn’t enough: the information must be accessible, understandable and correct.

  2. T
    Training

    Passing on knowledge isn’t enough: the information must remain available at the moment of action.

  3. M
    Mental load

    Nobody can concentrate on two tasks at once: multitasking breeds error.

  4. B
    Behaviour

    A reactive quality culture waits for errors instead of anticipating them.

  5. C
    Change

    Changing an ingrained habit requires genuine human change management.

The 3 factors that adjust the analysis

They are what makes the difference. Depending on the task and the person, the same causes carry very different weight: the 3 factors adjust the weight of each and target the real leads, instead of reviewing everything.

1

Complexity or criticality

A simple, low-stakes task doesn’t call for the same analysis as a complex operation or one critical for the patient.

Complexity or criticality
2

Frequency

A task done every day exposes people to routine and autopilot; a task done twice a year exposes them to forgetting.

Frequency
3

Experience

An operator who has been here for ten years and a newcomer don’t make mistakes for the same reasons.

Experience

An operator who arrived in the sterile area two weeks ago isn’t exposed to the same risks as an experienced operator who works every day in a non-sterile area. Same apparent error, different analysis.

Our approach: tackle errors before they happen

No single tool eliminates human error. It’s the accumulation of the right tools, at the right time, that reduces it drastically.

Find the real causes
1

Find the real causes

The DTMIC Tool analyses each error against 5 causes and 3 influencing factors. No more retraining by default: CAPAs finally address what produced the error.

DTMIC Toolroot cause

Bring weak signals to the surface
2

Bring weak signals to the surface

Teams spot risky situations before they become deviations. A quick huddle in the workshop, a weekly meeting to decide collectively on simple actions that can be done within the month.

weak signalsshort-term actions

Observe tasks
3

Observe tasks

Quick, regular, informal observations, anything but an audit. Managers see the gap between what is written and what is done, and correct it with the teams.

task observationshop floor

Talk quality as a team
4

Talk quality as a team

15 minutes for Quality: every two weeks or every month, the manager and their team discuss a quality topic, using ready-to-use sheets.

15 minutes for Qualityquality culture

Make critical checks genuinely effective
5

Make critical checks genuinely effective

The V.A.K. check: you look, you say it out loud, you point at it. 75% of packaging errors are attention errors: fewer checks, but real ones.

V.A.K. checkcritical checksstart-up

Anticipate before acting
6

Anticipate before acting

The pre-job briefing: 5 minutes before a batch start-up or a shift change to go over the day’s 2 or 3 critical points out loud. Instructions that say everything don’t make what matters stand out.

pre-job briefingshift change

Two sites, two results

Clinical R&D site, Packaging & Labelling unit of 200 people

Nearly half as many human errors in one year

408 deviations in 2022, a quarter of them due to human error, and retraining with no lasting effect. A human error expert per area, 4 shop-floor rituals and 6 months of coaching.

  • Human-error deviations in packaging: −76%
  • Deviations from all causes: 34 → 25 per month
  • 157 weak signals reported in 3 months
−47%human-error deviations (9 → 5 per month)
Production site, two units

A proactive culture driven by front-line managers

Recurring errors in production and strong growth. Weak signals, task observations, 15 minutes for Quality and action management, run by the teams.

  • Deviations and events: −24%
  • 272 weak signals reported and handled
  • 166 task observations carried out
−46%human errors, from April to September, year on year

See all our case studies

How an engagement works

  1. Flash assessment
    Flash assessmentUnderstand the site’s context: tools in place, organisation, premises and recurring errors.
  2. Introduction
    IntroductionA day to introduce your teams through hands-on experience: V.A.K., pre-job briefing, weak signals.
  3. Adaptation
    AdaptationOur tools adapted to your site: DTMIC Tool, observation checklists, 15-minute sheets, V.A.K. checks.
  4. Tapering shop-floor coaching over 8 months
    Shop-floor coaching 8 monthsA human error expert per area, coached on the shop floor during rituals, task observations and 15 minutes for Quality. Tapering support: very present at the start, then increasingly in the background, until your experts run the approach on their own.
    Month 1Month 8
    Coach presenceYour experts’ autonomy →

Frequently asked questions about human error

What is human error in the pharmaceutical industry?

It’s a deviation that results directly from a person’s action: a wrong setting, a forgotten step, a poorly performed check. But human error isn’t the fault of the person who makes it: it’s the symptom of causes that can be identified and acted upon.

Why isn’t retraining operators enough?

Because it doesn’t address the cause. In our benchmark, 85% of errors come back after 6 months following awareness training, a document change or an added check. As long as the documentation, training, mental load, behaviours or changes involved aren’t addressed, the error comes back.

How do you find the real cause of a human error?

With the DTMIC Tool, which screens the error against 5 major causes: documentation, training, mental load, behaviour and change. It cross-references them with 3 influencing factors: task complexity, its frequency and the person’s experience.

What is the V.A.K. check?

It’s a way of performing critical checks using three senses: you look at the actual item (visual), you say out loud what you read (auditory) and you point at or touch the item (kinaesthetic). Inspired by aviation and Japanese railways, it takes the operator out of autopilot: the check becomes conscious again.

What is a pre-job briefing?

A short moment, about 5 minutes, taken by the line lead before a batch start-up, a shift change or a restart after an interruption. It goes over out loud the objective, who does what, the day’s 2 or 3 critical points, what has changed and what to do in case of doubt. It doesn’t replace the documentation: it brings out what the documentation buries among everything else.

How quickly do you see results?

The first weak signals come up within the first few weeks. On our assignments, the drop in human-error deviations can be measured over a few months of coaching, with up to −47% in one year.

Do the same errors keep coming back on your site?

Tell us about your situation in thirty minutes. We’ll tell you where to start.

Tell us about your situation