Training mental health first aiders might tick a box intended to sit in a different column, but will not change workload, rosters, communication, role clarity, management styles, leadership, or other factors that we might roll into psychosocial risk.
‘Number of people who attended MHFA’, or ‘20% completed, 35% scheduled within six months’, should never, repeat never, make it into Board papers as a metric for addressing psychosocial risk. Ignoring for a moment the chasm between ‘attended’ or ‘completed’ and ‘this was useful material that created reaction, learning, behaviour change and measurable results (thanks Kirkpatrick) it is just plain wrong. I get that attendance or completion are tidy things to measure and thus easy to hold up as markers of effort or success, but they really are way off the mark.
What we know
Let’s start with what the research says, which seems consistent. MHFA changes what people think. Across 18 controlled trials involving nearly 6,000 people, trainees came out knowing more about mental illness, feeling more able to help, and saying they intended to help. Those are genuine gains.
What it doesn’t seem to do, is reliably change what people then do.
The founding trial shows perhaps the clearest evidence. This is work by Betty Kitchener and Anthony Jorm in 2004, who put 301 Australian government employees through the course, with a randomly chosen control group. Kitchener wrote the material and taught every session herself. She notes this as a limitation, along with the fact the trial tested ideal conditions rather than real ones. By that, we mean that there was significant consistency in delivery and material quality, people were given time off to do it, and the set up was ideal. Kitchener herself notes that in the real world it might be different, and specifically noted research is needed on how the course performs when other people deliver it.
In the ideal conditions, confidence in helping rose by 20 percentage points in the trained group. The proportion who actually went on to give someone ‘some’ or ‘a lot’ of help rose only 2 percentage points. That can be explained by chance.
Twenty points of confidence, but two points of behaviour, in an ideal scenario
There’s an obvious objection
Here it is. Perhaps those people simply never came across someone who needed help?
Except that Kitchener’s data show seven in ten people had contact with someone experiencing a mental health concern during the study. Fewer than four in ten gave any real help. The opportunity existed for most people to do something, and most of them didn’t take it.
This may also tell us the obstacle, and it isn’t knowledge. Kitchener didn’t measure why, but let’s make an educated guess. Raising someone’s mental health with them is still awkward, even though we’re generally more aware. It might not be my business. I might be wrong. Maybe I don’t feel like it’s my job. I might make it worse. Mental health can still feel really hard. Am I really equipped to handle it? And maybe others. But several hours of training made only two percentage points of difference.
The good news
One behavioural result did hold, and it’s a good one. The trained group went up slightly on advising someone to see a professional. At the same time, the untrained group fell by ten points over the same timeframe. So the finding is mostly the control group getting worse, not the trained group getting better. However, from a different perspective, preventing a decline is a VERY positive outcome, and we should bank that.
Two decades later, research hasn’t really closed the gap. Very few studies have even looked at people on the receiving end of help. Those that did showed effects the reviewers called “less convincing”, with the numbers too imprecise to lean on.
Here’s the thing. The major issue is that what’s measured is about the trainee. I feel I know more. I feel more confident. I think I’ll help more. Or it might measure whether the person they helped may feel better afterwards. They tell you how an organisation might respond to distress that already exists. Nothing in there tells you if less distress was produced, which is what a risk control measure might do. Happy for you to show me an MHFA trial that measured whether anybody’s exposure to a psychosocial risk went down.
What MHFA is for
None of this makes MHFA a bad programme. It makes it a bystander programme, per its design.
The trials can’t show much beyond routine helping which is common and largely low-stakes. It’s much more blind to rare, or edge cases. If a trained colleague is involved with one suicidal person every few years, it won’t show up in a trial this size, but it might be the single most important thing they, or anyone in the building does.
There’s also what Kitchener found by accident. The trained group’s own mental health improved relative to the control, and she had no explanation for it. Whatever it was in the training, it appeared to do them some good.
So, as I said, the issue isn’t MHFA, the issue is the column we put it in.
What a first aider can’t do
They can direct more people to professional help, but they can’t reduce a workload, remove a bullying team leader, fix a roster, manage sleep or shift work, improve role clarity, and so on. Where the distress is generated by the work, even the most compassionate conversation in the world still returns the person to the exposure.
The compliance challenge
A psychosocial hazard arises from the design of work, or the social context, or the organisation of work, any of which can cause psychological or physical harm, and organisations must address these. WorkSafe also specifically calls out the tendency for organisations to treat psychological health as an individual-level problem, rather than a problem with the work system itself. Aas I said earlier, ‘attended a course’ is easier to measure than ‘changed work structures to prevent burnout or fatigue’ or ‘dealt with a longstanding culture of middle management bullying we’ve previously been too scared of handling’.
Under the Health and Safety at Work Act, duty runs to eliminating risks to health so far as is practicable, and minimising them only where elimination isn’t. Health includes mental health. WorkSafe is also abundantly clear that efforts should go on eliminating or preventing risks, and supporting individual wellbeing “does not replace your legal duty to manage psychosocial risks”.
Consequently, a first aid kit isn’t the answer for working at heights. Barriers are. So are harnesses. And training. And clear and realistic expectations. And the right equipment in good repair. You get it, and you can pick your own example. The kit is there for the time the controls fail.
Which is where MHFA sits. It is an after-event response, staffed essentially by volunteers, dependent on someone disclosing to them or them asking, and most likely being in no authority over the thing contributing to, or making, them unwell.
Australia has moved on this with new regulations and code of practice. ISO 45003 does the same with international guidance. WorkSafe signals the same shift here in Aotearoa NZ without the regulation behind it. Yet. In my view we’ll close that gap in the next few years.
Some broader context
If we step back and take a broader view, there’s more information available to us. Fleming’s 2024 study of more than 45,000 workers across 223 organisations didn’t cover MHFA, but it did cover adjacent programmes, including mindfulness, resilience training, stress management, and wellness apps. Conclusion: people who took part were no better off than matched colleagues who didn’t.
What they did report was feeling supported. Participants were more likely to agree that their organisation helps them manage stress, even though the actual wellbeing scores showed no improvement. The survey item itself was worded “my organisation supports me to manage stress at work”. Fleming points out, correctly, the emphasis on the individual, opining it leaves room to be read as “I’m supported as a person” when in fact the employer has done nothing about conditions.
In any event, it’s a snapshot, not a before-and-after. Fleming says as much, and his conclusion is what the Board should hear. He says organisational change, (management practices, staffing, job design, and scheduling) “appear more beneficial for improving wellbeing”.
A better strategy
MHFA isn’t the strategy.
Instead, identify hazards upstream, and at the level of the job, rather than the person. Which roles carry high demand, and how do you know? Seek exposure data, and provide the Board information you’d give a regulator. Then redesign the work, right down to the behaviour of specific managers. Then verify that the exposure to risk changed, and you didn’t just run a wellbeing programme.
Your first aiders sit downstream, doing the job they’re good at. They just aren’t in the column where the control measure goes.
A closing question
If MHFA sits on your psychosocial risk register as a control, what is it controlling?
Photo by Diana Polekhina on Unsplash
References
WorkSafe New Zealand. (2025). Managing psychosocial risks at work. WorkSafe New Zealand.