Psychological safety is about creating a climate in which we feel able to take interpersonal risks in order to communicate our ideas, concerns and issues – and we want to be able to speak up in a way that we feel comfortable to. But that doesn’t mean we shouldn’t make an effort to articulate what we want to say in clear, accessible ways that allow the recipient to understand our intention and act on it appropriately. PACE was created as part of CRM, in response to disasters such as Tenerife in 1977.

We want to communicate in ways that give the recipient the best chance of hearing and understanding the message. So it’s important to avoid communicating in ways that can come off as aggressive or rude; partly because it’s good to be kind, but also because most of us struggle to receive information that’s delivered aggressively or harshly – we’re more likely to become defensive in those cases – hearing not *what* is said, but *how* it was said instead.

But sometimes time is of the essence, and we need to quickly communicate an important message. This can be especially challenging if we’re communicating with someone across a power gradient such as to a surgeon in an operating theatre or the captain in a cockpit of a passenger plane. How can we do this respectfully and effectively, ensuring our words have the intended impact?

PACE: Probe, Alert, Challenge, Emergency

One way is to use “graded assertiveness”, which there are various models for (including a good one called CUSS), but my personal favourite is PACE. PACE allows us to begin with a low-threat question (a “Probe”). This may well be enough to cause the recipient to pause, consider and adjust their course of action. If it’s not, we can incrementally escalate to higher levels of urgency and assertiveness. It also means that if we’re wrong about our concern, beginning with a simple probe reduces the interpersonal cost of that mistake, and poses less risk of causing a so-called “cry wolf” incident by going straight to emergency-level communication with a mistaken concern.

Using PACE, we first “Probe”, then “Alert”, then “Challenge” if necessary, and if we have to, we can resort to “Emergency”. I recently delivered a talk at a national highways safety conference, and introduced PACE as a method to effectively raise concerns to colleagues and peers as below:

As an example, let’s say Jane has noticed that her colleague, Bob, looks very tired and has already necked two cans of Red Bull. He has just climbed into the cab of his truck to begin a long journey. Jane believes that Bob is too tired to drive safely, and knows she should speak up, but wants to make sure the message is received well.

This is how Jane might use the PACE framework:

Probe – “Hey, Bob, you’ve got a long drive ahead, are you feeling rested enough to do this journey?”

Bob says he’s fine, and continues to prepare the cab of the truck, so Jane escalates to:

Alert – “Bob, I’m concerned that you might be too tired to complete this journey safely. Maybe you should take some time to rest before heading off.”

Of course, Bob wants to get cracking, so insists he’s ok and fires up the engine. Jane needs to make it clear that she’s not comfortable with Bob driving this tired, so moves to:

Challenge – “Wait Bob. I don’t think this is safe. Take a rest first.”

Bob’s lack of sleep is probably making him a bit irritable too, so he again dismisses Jane’s concern, and belts up ready to go. Jane knows this is her last chance to prevent a potential accident, so escalates to the emergency level:

Emergency – “Bob, STOP. You’re not safe to drive.”

At this point, there’s no ambiguity. Bob knows that he’s about to do something dangerous if he drives off, so stops the engine and gets out of the cab. Whilst Bob might be a little put out at having to rest before doing his drive, we’ve avoided potential disaster. 

Of course, hopefully in most situations, we don’t have to escalate all the way to Emergency. Probing in itself can be a very effective trigger to cause someone to consider their actions or pause long enough to check they’re safe. It can be any low threat question that raises the issue at hand:

• “Are we on track with the patient’s breathing management?”
• “Have we successfully executed all the integration tests on this code?”
• “Do we have ATC clearance for departure?”

Alerting can take many forms too, and can be enough to prevent an incident. It’s where we clearly point out our concern:

• “Oxygen is down to 90%.”
• “This release hasn’t yet passed all the integration tests.”
• “ATC say wait for clearance.”

Challenging someone in authority can itself be challenging, but knowing this framework and going through the steps can make it easier to do so, and bring a situation under control. For this, we clearly state the concern and suggest an action:

• “This is the fourth attempt at intubation. We should ask for assistance now.” 
• “Deploying this code now could saturate our load balancers and bring down the service. Let’s pause and run further tests.”
• “We don’t have clearance. Hold position.”

And if these don’t work, we can escalate to the Emergency level, with an unambiguous, decisive and forceful command. This can be uncomfortable, but it is much easier to do if we’ve already taken the previous steps. This usually consists of a decisive “STOP. This isn’t safe.” or similar.

Advantages of PACE

The very existence of a framework like PACE signals to people that it’s ok, and even expected that we speak up before something reaches an emergency state. Of course there might be extreme time-critical situations in which we need to escalate straight to emergency, (if there’s a fire, it’s ok to shout “Fire! Everyone get out!“) but when there is time to make a more considered and graded intervention, PACE is an invaluable tool. 

Another distinct advantage of PACE is that once we’ve learned it, we may recognise when others are using it with us. That in itself is a powerful signal that they’re trying to draw our attention to something important, and we can catch it before they have to escalate to emergency communication.

Because it’s most valuable in crisis or pre-crisis situations, it’s best to try to make it second nature before you find yourself in those situations, so practise it in simulations with your team, or if that’s not possible, simply practise it yourself so that it comes more naturally should you need to employ it. You can also display it on posters and discuss it with colleagues to help others to adopt it. 

As a framework, PACE is used extensively in healthcare, aviation, maritime, and other domains. The PACE framework serves as a valuable tool for fostering psychological safety and effective communication about things that are important or safety-critical in high-stakes contexts. By structuring the escalation of concerns, PACE facilitates clear, respectful dialogue even across power gradients, and can significantly improve safety as well as day-to-day collaboration.

Further Links:
Psychological Safety: Structure and Power
Healthcare and CRM
Amplifying Weak Signals
Civility Saves Lives
Normalisation of Deviance
PSIRF in Healthcare
Speaking up at work
Psychological Safety in Aviation
Types of Human Error
The Andon Cord
Non-Violent Communication
Power over and power to

Psychological Safety in practice

Quality and Safety in Health and Care

The “Cambridge Elements” series of guides to key approaches to improving quality & safety in health & care are written by leading authors and open access. They offer a comprehensive and authoritative overview of various improvement approaches, exploring the thinking behind them, examining evidence for each approach, and identifying areas of debate.

Topics include Statistical process control (which we also covered in this newsletter issue), Health economics, Simulations, Positive Deviance and more. Element 7 on workplace conditions was written by the fabulous Jill Maben, Jane Ball and Amy Edmondson. Thanks to Mary Dixon-Woods for sharing.

Psychological Safety in Education

Our wonderful placement student Bea finished her placement last week, and we’re going to really miss her. After some great conversations about what psychological safety looks and feels like from a student perspective, she’s shared her own experiences in this fantastic article. It’s a personal reflection, but makes points which I suspect will resonate with student and teachers alike.

Error-catching Centaurs

This issue of Pluralistic by Cory Doctorow, on AI, is a great read, and relevant to many of us in the safety, performance and quality fields: Humans are not perfectly vigilant. Cory describes the idea of a “centaur”: a human head with a powerful robot body, which echoes the TPS concept of Jidoka; automation with a human touch. Centaurs, and the concept of Jidoka, make use of the strengths of humans to solve tricky problems and deal with complexity and uncertainty, whilst machines can do what humans are not good at – doing something the same way every time. This is also why we address automation and human factors in our psychological safety work – to foster cultures where we can identify and learn from mistakes, we must also implement systems that catch the errors that fallible humans inevitably make. 

However, Cory rightfully points out that with AI (well, LLMs really), it’s humans that have to be the error-detecting and correcting component. AIs are so prone to mistakes and hallucinations that it’s us humans that need to check it for accuracy. Instead of centaurs, we’ve created reverse centaurs that require us humans to check it for mistakes and errors, which is not something humans are generally good at. The solution to this, according to the AI firms currently burning through cash, is to add vastly more training data. But a lot of that training data (content on the internet) has, already, been created by AI’s (including their resultant mistakes and hallucinations), and this, I fear, is creating a feedback loop that will ultimately make AI tools only ever more confidently wrong.

Authoritarianism and Psychological Safety

Tim Harford’s Cautionary Tales might be the podcast I share most often because it’s not only excellent but covers stories and ideas that are core to this newsletter – learning from mistakes and the complex causes of disaster. This episode gets into salami-slicing, hypochondria and authoritarianism, and psychological safety, using the example of Paul O’Neill at Alcoa, in contrast to the authoritarian dictatorships of Stalin, and others.

Psychological Safety, DevOps and Agile

A while back, I had a wonderful conversation with John Ruppel for DASA and this is a write up of our conversation about psychological safety, DevOps & Agile. We covered loads of stuff, from unintended consequences to personal user manuals, and highlighted Weak Signals: “The concept of amplifying “weak signals” is a way to identify subtle signs of failure. Weak signals on their own may appear irrelevant or insignificant, but can actually signal much larger issues lurking – so creating the conditions and capabilities to detect and act on weak signals is key to success: and this requires psychological safety.”

This week’s poem:

Jerez at Easter, by Robert Bly

Please tell me why the lamb is in love with the wolf
And why the child’s finger calls the hammer down
And why at dusk Alexander walks toward his enemies.

Tell me why the gazelle grazes so close to the lion
And why the rat makes up games on the snake’s tail
And why the student bends his head when he’s attacked.

One meadow in the redwoods can contain a thousand ferns.
By this we deduce we are living in the serpent’s home.
Each curly fern is his tongue unfolding.

The poet makes a meadow from each leaf.
Each curve of language turns into a lamb’s ear,
Because a genius is a child in the house of suffering.

None of us is free from a certain bend in the knee.
The caws from the oak-bound ravens in the trees
Around our house guide Alexander toward the night.

The old man’s voice breaks as he sings at Easter.
In between the clapping, there’s always a voice breaking.
Last night in Jerez some people lived, some people died.