The Black Dog finding the aged care sector hasn’t acted on

You have done the hard thing. Pay went up after the Fair Work decision, and you funded it. And you are still losing registered nurses inside twelve to eighteen months. Mandatory 24/7 RN coverage gets harder to staff every quarter, not easier. And your own workforce survey keeps returning the same line in slightly different words — people feel stretched too thin to deliver the care they would like to give. If pay were the lever, the curve would have bent by now. It hasn't, and that is worth sitting with rather than explaining away.

The reason is not a mystery, and as of March 2026 it is no longer an opinion. It has been measured.

What did the Black Dog data actually find?

The Black Dog Institute surveyed 1,085 residential aged care workers and published the results in npj Mental Health Research (Black Dog Institute, March 2026). The headline numbers are blunt: 56% reported elevated burnout, 35% reported anxiety symptoms, and 24% met the threshold for probable depression. Those are not figures you would tolerate in any other operational risk register. If a quarter of your workforce had a measurable physical injury rate, it would be a board agenda item by Friday.

But the headline numbers are the symptom. The finding that matters for an operator is the one underneath: lower perceived supervisor support was among the strongest predictors of all three outcomes (Black Dog Institute, 2026). Not workload alone. Not pay. Not the residents. The variable most tightly bound to whether a worker is burnt out, anxious or depressed is whether they experience their immediate supervisor as someone who has their back.

This is what makes the finding useful rather than merely alarming. Supervisor support is modifiable. You can build it, measure it and hold it. You cannot do that with the resident acuity mix or the structural turnover rate. The data has handed you the one lever that is actually inside your operating model — and it is the layer most providers invest in least.

Why didn't the pay rise move the retention numbers?

The $2.6 billion Fair Work aged care pay rise (ANMF, 2024) did something important and overdue. It recognised the value of the work in the one language the labour market understands without translation. Recognition matters, and underpaying people is its own hazard.

But pay is a recognition signal, not an experience signal. It changes how a worker feels about the sector on payday. It does not change the Tuesday afternoon when they are two staff short, an agency nurse is on the floor who does not know the residents, a family is escalating, and the person rostered to support them is a team leader managing forty direct reports across two shifts. Pay does not touch any of that. Supervisor support does.

Consider the structural context. There are roughly 450,000 aged care workers in Australia; about 30% were born overseas, the median age is 46, and turnover sits structurally at around one third of the workforce (Brightstar Nursing Australia, 2025). You are running a large, mobile, multilingual workforce through a churn rate that means a third of your supervisory relationships reset every year. The relationship the data says protects people most is also the one your turnover rate dismantles fastest. That is the gap.

Is this a wellbeing problem or a risk problem?

It is a risk problem, and framing it as anything softer will cost you. Under the model WHS laws, psychosocial hazards — including high job demands, low support and poor supervisory relationships — are hazards you have a positive duty to identify and control so far as is reasonably practicable. The same hierarchy of controls you already apply to manual handling applies here. Officer due diligence does not pause at the office door because the harm is psychological rather than physical.

The Royal Commission into Aged Care Quality and Safety named the workforce strain in plain terms. The Black Dog study has now quantified the mechanism behind it. Put together, you have public-record testimony and peer-reviewed data pointing at the same control point. A regulator, an insurer, or a plaintiff's lawyer can read both. The question they will ask is not "did your people feel supported" — it is "what reasonably practicable steps did you take once the hazard was known". After March 2026, "we lifted pay and refreshed the EAP" is a weaker answer than it was last year.

This is the difference between the top of the cliff and the bottom of it. Your EAP is the ambulance — necessary, but it activates after harm has already occurred and after the worker recognises they need it and chooses to call. The supervisory layer is the fence at the top. Investing only in the EAP while the strongest modifiable predictor of harm sits unaddressed is an identifiable gap in your controls, and it is now a documented one.

Why won't another resilience module fix it?

Because resilience training puts the burden of the hazard onto the person exposed to it. It is the equivalent of responding to a manual-handling injury rate by teaching people to lift more carefully while leaving the load unchanged. The data does not say workers lack resilience. It says they lack support from the layer directly above them.

That layer is usually the least-equipped group in the building. Team leaders and clinical coordinators are promoted for being excellent on the floor, then handed people-leadership responsibility with no structure underneath it. They become accidental counsellors — absorbing distress they were never resourced to hold, with no defined point at which they hand it on. Spread across a span of control far wider than any one person can hold a real relationship across, the support the Black Dog data says is protective simply cannot form. You cannot support forty people the way the data requires.

The fix is structural prevention, not a content drop. It means building supervisory capability deliberately — defining what support looks like in practice, setting spans of control that make it possible, giving supervisors a clear route to escalate rather than absorb, and treating wellbeing infrastructure as part of the operating model rather than an annual event. That is the work that sits pre-EAP, in the gap, and it is the work that both reduces escalations and routes the people who do need clinical support there earlier.

How would you know if this is your problem?

You already have the signals; they are just filed under the wrong heading. Look at where in the worker lifecycle your RNs leave — the twelve-to-eighteen-month cliff is a supervisory-relationship signal, not a pay signal. Look at how exit interviews describe leaving, stripped of the polite reasons. Look at the variance in turnover between sites with the same pay, same acuity and same agency reliance — that variance is usually a supervisory variance, and it is measurable.

Making the invisible visible is the first move. Psychosocial risk is invisible right up until it appears as turnover, an incident, a claim or a regulator's letter — at which point it is no longer a prevention question. A proper risk assessment puts the hazard on the risk register where it belongs, alongside the controls you have actually implemented and the ones you have not. That is the document that answers the "reasonably practicable" question before someone else asks it.

What does acting on it actually look like?

It looks like treating the supervisory layer as a core control in your risk framework and building it on purpose. Define the support behaviours the data rewards. Right-size spans of control toward something a human can actually hold. Give supervisors structured backing so they are not absorbing distress alone. Use wellbeing champions and a clear escalation pathway so the system catches people before the EAP has to. And measure it — supervisor support is a number you can track, the same way you track falls or medication errors.

The Black Dog study did the expensive part for the whole sector. It identified the lever. The providers who act on it will hold their RNs through the eighteen-month cliff, staff their 24/7 coverage with less agency dependence, and answer the due-diligence question with evidence rather than intention. The ones who file it under "wellbeing" and run another module will be having this same conversation next quarter, with the same survey line, and a sharper paper trail working against them.