In March 2025, three Canberra Health Services workers — nurses and allied-health professionals doing home visits as part of the Hospital in the Home and Belconnen Recovery Service programs — were assaulted at the doors of the people they had come to care for. Two separate visits. Severe enough that CHS was required by law to notify WorkSafe ACT. Severe enough that the watchdog, in its prohibition notice, used language that is rare for a workplace regulator:
"Both situations resulted in physical and psychological harm to the affected workers and had serious potential to result in the death of a person, or the serious injury or illness of a person."
A subsequent investigation issued six improvement notices and one prohibition notice. WorkSafe found that the risk assessments CHS used for home visits were ineffective, the alarm systems unsafe, and the duress protocols inadequate. The notices were eventually lifted — but only after CHS implemented a specific set of fixes: strengthened risk-assessment tools, mandatory pre-visit training, properly working duress devices, and — critically — "ensuring staff have a duress device and that their manager always knew where they were."
That last sentence is the entire post.
What the union has been saying for two years
CPSU ACT Regional Secretary Maddy Northam told the ABC, in language that should make every operations director in community health pause:
"We've had members who have had the front door opened and the patient or a family member may be holding a machete or a knife or an axe or chain. There are firearms present, guns, bows and arrows. CPSU members at Canberra Health Services have been raising pretty serious safety concerns about home visits for more than two years now."
This is not industrial hyperbole. The four assaults in March 2025 were specific, documented, and severe enough to require medical assessment and occupational-violence leave. The CPSU's two-year warning was on the record, in writing, before the assaults happened. CHS — like community-health services across most of Australia — had a duress system, had training, had policies. What it didn't have, until WorkSafe required it, was a real-time picture of where its workers were when they were inside someone's home and the situation went wrong.
The Jigalong case — same gap, different geometry
Fourteen months before the ACT assaults, in January 2024, a group of young people allegedly attempted to enter the home of three nurses at Jigalong — a remote Aboriginal community 160 kilometres east of Newman in the Pilbara — by scaling the security fence. The nurses, all employees of Puntukurnu Aboriginal Medical Service (PAMS), were quickly relocated to Newman.
For most of the next week, Jigalong's only health service — for a community of roughly 333 people, many of them elders dependent on regular medication — was staffed by a single nurse.
The Jigalong gap is the inverse of the ACT gap. ACT Canberra Health Services has an alarm system, a duress device, a manager nearby, and a city's emergency-services geography to escalate into. What it lacked was the discipline of knowing where the worker is right now. Jigalong has the opposite shape: no nearby emergency-services geography to lean on, a community where the nurses are themselves the emergency response, and an isolation profile that makes any incident-of-doubt automatically escalate to "evacuate the workforce." What it lacks is the visibility that would let an organisation distinguish between this nurse is in the kitchen with a stable patient and this nurse has not responded to the planned check-in for thirty-five minutes and her last known location is half a kilometre off the visit address.
Both gaps have the same shape at the level of the operations layer. The lone worker is across the threshold of a door that the organisation doesn't fully control. The organisation either knows where she is and what her situation looks like, in real time, or it doesn't.
What WorkSafe ACT actually demanded
The fix list CHS reported to WorkSafe, after the notices, has four ingredients. They are the same four ingredients the Reg 48 audit framework asks for in mining services, the same four the post-Helene mutual-aid review asks for in utilities, and the same four the post-Sigma/Chemist Warehouse integration window asks for in pharma distribution. They are, evidently, the same four ingredients every distributed-workforce operator now needs to produce:
- Risk assessment — a structured pre-visit assessment that flags the specific risk of the specific home before the worker leaves the office.
- Communications cadence + duress equipment — the worker is reachable, has a working duress device, and there is a documented expected check-in.
- Manager-knows-location — the worker's location and check-in status is visible to a named, on-shift manager throughout the visit, not retrospectively.
- Escalation chain — when a check-in is missed or a duress signal is raised, there is a time-stamped chain of who was alerted, what they did, and what the resolution was.
These four ingredients are what allow a service to do what WorkSafe required: ensure staff have a duress device and that their manager always knew where they were.
Most community-health services in Australia have a partial version of this. Almost none have the integrated, audit-ready version. That gap is what 2025 caught, and it is what 2026 audits — across ACT Health, NSW Health Local Health Districts, Queensland Hospital and Health Services, Victorian community-health providers, and the major NDIS operators — will keep catching until the gap closes.
The geography of the gap
Community nursing in 2026 looks very different from community nursing in 2010.
The Royal District Nursing Service of the Victorian era — a single nurse with a leather bag walking a known beat through a known neighbourhood — has been transformed by Hospital in the Home, by NDIS in-home supports, by aged-care reform, by the post-COVID surge in community-based mental-health care, and by the systemic shift toward keeping patients out of acute beds whenever possible.
What that means in practice is that any urban community-health team is, on any given weekday, deploying somewhere between dozens and hundreds of clinicians to addresses across a metropolitan area. Each visit has its own risk profile. Each patient has a clinical record, but the operational picture — who is at which address, when did they arrive, when was their last check-in, who is responsible for them if something goes wrong — frequently lives in a roster spreadsheet, a clinical-system schedule, and a series of WhatsApp groups among supervisors.
That is not, by 2026 regulator standards, a system of work that includes effective communication with the worker. It is a roster that hopes communication won't be needed.
A roster is not a system. A roster is what's left when a system is absent.
What changes between 2025 and the next assault
The hard part of community-health operations is that the very nature of the work makes the next incident almost impossible to predict. Most home visits are uneventful. Most patients are vulnerable people who need care and are grateful for the visit. The four CHS assaults in March 2025 were outliers — not in their existence, but in their cluster. The CPSU's two-year warning is what the operational layer needs to catch before an outlier becomes a fatality.
What "catches it before" looks like, operationally:
- A pre-visit risk score that draws on the clinical record (history of violence, history of substance issues, history of police involvement), the geographic context (known difficult addresses, known social context), and the worker's own situational assessment (visit is going ahead alone, or with a pair-worker, or with a duress-acknowledged escalation path).
- A live check-in window that the worker satisfies on arrival, mid-visit (longer visits), and on departure. The duress device is a separate channel; the check-in is the positive signal.
- A manager dashboard that shows the team's current state at a glance: who is at which address, whose check-in is overdue, whose duress device is currently triggering an alert.
- A logged escalation chain that captures, for every alert: time raised, time received, contact attempts, outcome, resolution.
This is, again, the same four-category framework the Reg 48 audit applies. It is the same one that survived contact with Hurricane Helene in our previous post. Healthcare is not, in this respect, a unique vertical. It is just one where the consequences of getting it wrong involve a nurse at a stranger's door rather than a lineman in a forested ravine.
The Health Minister's controversial response
In the immediate aftermath of the WorkSafe notices being publicised, ACT Health Minister Rachel Stephen-Smith made remarks that drew sharp pushback from healthcare unions: dealing with upset patients or "people who have little control over their own behaviour" was, she suggested, "unavoidable when you deliver a health service."
The CPSU response was direct: "Occupational violence in hospitals is not unavoidable. It is foreseeable, and this government needs to do more to manage this risk."
The minister and the union are arguing about the wrong axis. The question is not whether violence is unavoidable — some baseline of risk is, in this work, irreducible. The question is whether the operational layer that wraps around the worker turns a risk into a manageable incident or into an open question that ends in an injury or a death. That layer is engineerable. Foreseeability is the operations director's burden.
A specific scenario, written down
A registered nurse from a Hospital-in-the-Home service is dispatched to a suburban address at 2:15pm on a Tuesday to administer an intravenous antibiotic course to a patient discharged from acute care the previous evening. Her workflow, as it stands at most services in 2026:
- 10:00am: clinical handover note in the EMR; the patient's family-violence history is not flagged because the previous admission's safety screen wasn't completed.
- 1:45pm: she leaves the office in her work vehicle; the duress device on her lanyard is charged.
- 2:14pm: she arrives at the address. The front door is opened by a family member she does not recognise; the patient is not visible.
- 2:20pm: her manager, back at the office, glances at the roster, sees she's on a 30-minute visit, expects her back around 3pm.
- 2:32pm: the situation in the home escalates. The nurse retreats; the duress device alerts. The alert lands in a queue that her manager checks at 2:40pm.
- 2:42pm: the manager attempts to phone. The nurse — who has retreated to her vehicle — answers. She is shaken but unharmed. She drives back to the office.
That scenario, as written, ends well. It is also the scenario in which an incident that could have ended in a fatality is logged, retrospectively, as a "near miss" that gets discussed at the next team meeting and is then forgotten until — somewhere across the next thousand visits — the same conditions assemble themselves around a worker who is not as quick to retreat.
The version of that scenario that ends consistently well looks slightly different:
- 10:00am: handover note in the EMR. The risk-assessment tool, integrated, flags the patient's prior family-violence history. A "pair visit" recommendation is appended to the scheduler.
- 1:45pm: the nurse leaves the office. The system has her itinerary.
- 2:14pm: she arrives. The system records the arrival.
- 2:20pm: her manager's dashboard shows her green: arrived, in-visit, check-in due 2:35pm.
- 2:32pm: situation escalates. She retreats. The duress signal is raised. The manager's dashboard turns red within 30 seconds.
- 2:33pm: manager has a worker on-screen, address visible, last-known location accurate to 5 metres, contact attempt initiated.
- 2:35pm: contact established; debrief; vehicle confirmed mobile; emergency services notified by protocol; the entire incident log is captured for the after-action review.
The difference between the two scenarios is not equipment. Both nurses had the same duress device. The difference is the operational layer — the risk integration on the front end, the manager dashboard in the middle, the audit trail on the back end. The layer is buildable. Most services haven't built it yet.
What we'd recommend looking at
If you are a community-health operations director, an NDIS provider's chief operating officer, or an aged-care home-services manager reading this from inside an organisation that produces ten or a hundred or a thousand home visits a day:
-
Run the four-question self-audit. For any home visit in the last 30 days: do you have, on demand, the pre-visit risk score, the check-in record, the manager-dashboard state at the time, and the escalation log? If you can't produce all four in 30 minutes, you have the gap the ACT WorkSafe investigation was about.
-
Test the dashboard against a real shift. Sit beside a duty manager during a peak community-services shift. Watch what they actually see, what they're inferring from absence-of-signal, and what they would do if the dashboard turned red. The honest answer is usually that the dashboard doesn't exist yet — there's a roster, a phone, a WhatsApp group, and institutional memory.
-
Map the worst-case workflow. For a visit that escalates: from the moment the worker first feels unsafe to the moment a named manager has eyes on the situation, how many minutes elapse and how many manual steps are involved? In 2025 ACT, the WorkSafe finding was that some workflows were instructing workers to use the duress device in a way that could further escalate the situation — because the device alert was visible to people in the room. The workflow design is the safety design.
-
Have the integration conversation. Most community-health services already own duress devices, scheduling systems, clinical EMRs, and risk-assessment tools. What they don't have is the layer that takes those signals and surfaces a real-time operational picture to a manager. Building that layer in-house has been attempted at most large LHDs; almost none have succeeded. The off-the-shelf product category exists; we're one of several.
Sources: ABC News (28 April 2025) "WorkSafe ACT investigation finds life-threatening work conditions for nurses after 'serious' assaults"; ABC News (11 January 2024) "Anger as youth crime drives health workers out of remote WA community of Jigalong"; ABC News (17 January 2024) "Pilbara elders call for new approach to youth crime in Jigalong, as nurses evacuated after break-in"; ABC News (26 November 2025) "Steady increase in occupational violence incidents against healthcare workers at Canberra Health Services"; The Canberra Times "Canberra staff face risks in home visits probe"; The Canberra Times "Violence not 'unavoidable part' of being healthcare worker"; Health Services Daily "WorkSafe ACT hits Canberra Health Services after home visit assaults".
