Five documented risks, and the evidence behind them
Graylands doesn't publish a public incident record. This page is what can be assembled without one — every documented incident and every official statistic we could find, in one place.
Four things show up when they're read together. The people held here under custody orders have been charged with Western Australia's most serious offences, including homicide and child sex offences. Absconding is common — 153 recorded events in a single year, seven times the next-highest WA facility — in the same document that separately records the hospital's dedicated secure forensic unit, Frankland Centre, at fewer than five. Judgements about who is well enough to leave, and when, have gone wrong. And a small but non-zero number of those failures have ended in serious harm to people outside the fence, repeatedly, across three decades.
None of this is an argument that these patients don't deserve treatment, or that this facility shouldn't exist. It is the basis for the five risk factors below. The plan triples forensic capacity here, removes the general hospital, and builds in a legislated path to supervised community access — within about a kilometre of five schools. Whether that specific model is safe at this specific site is a question nobody has publicly answered. We can't prove it isn't. The government hasn't shown it is. Closing that gap is what we're asking for.
Five risk factors, and what grounds each one
Every claim below links directly to the source it comes from — a log entry, an official dataset, or a comparable case elsewhere — so none of it asks to be taken on faith.
Absconding is frequent here, and the public record shows what can follow it
What it isPatients leaving Graylands without authorisation — whether a physical escape or simply not returning from approved leave by the required time.
What it looks likeIn the one year for which a full facility breakdown has ever been published, Graylands recorded 153 absconding events — seven times the next-highest WA hospital (Bentley Adolescent, 22) — while the same document lists Frankland Centre, the hospital's dedicated secure forensic unit, separately at under 5. That doesn't make the 153 a general-patient figure: civil involuntary patients have their own legally separate leave-of-absence and absence-without-leave provisions under the Mental Health Act, entirely apart from forensic custody-order leave, so a large number outside Frankland doesn't require a forensic explanation — but it doesn't rule one out either. Government doesn't publish what became of the other 152 individually, or how many were held under any form of custody order. What the split does establish, and what's directly relevant to Stage 1: Frankland's low number reflects its own tightly controlled maximum-security model — it isn't evidence about how the new 32-bed sub-acute unit will perform, since that unit is explicitly a lower security classification, built around structured community access rather than Frankland's model. See the full facility breakdown →
Why it mattersThe public only learns what an absconding event led to when it becomes serious enough to make the news or reach a court. What's visible is a repeating pattern going back to at least the 1980s — documented in Parliament in 1990, before the purpose-built secure unit even existed: a sexual assault after a patient wandered into Perth, a maximum-security escape from a unit an Attorney-General called escape-proof, a nurse stabbed by a recently released patient, a homicide two days after an absconding. What isn't visible is everything that didn't become a headline.
Documented examplesTransitions between security levels are where documented failures cluster
What it isThe point where a patient moves from one security tier to another — locked ward to open ward, Frankland Centre to a step-down facility — rather than staying in one static placement.
What it looks likeTwo of the clearest documented failures in this record happened in the same year, 2014, and both happened specifically during a transition — not while a patient was static on one ward. Romily House, the lower-security step-down facility used for exactly this kind of move, is 2.3km from Graylands — a short trip, on paper — and someone still went missing making it.
Why it mattersNo published document explains who decides a patient is ready to move security tiers, what happens when that judgement is wrong, or how the CLMI Act's push toward staged community reintegration (Risk Factor 5) affects how often that decision now has to be made.
This isn't a new theoryColin Barnett made this exact argument to the Legislative Assembly on 29 November 1990, before the Frankland Centre was built: moving offenders down through security tiers carries a risk that can't be engineered away, he said, and failures serious enough to harm someone outside the system were inevitable, not hypothetical. He asked for a physical buffer around the point of first transition. Government did not build one, then or since. Read what else he warned Parliament, in full →
Documented examplesRelease and diagnostic judgements have gone wrong, with fatal consequences
What it isA clinical decision — a diagnosis, a security downgrade, a judgement that someone is well enough for more freedom — made in good faith, that later turns out to be wrong.
What it looks likeGraylands diagnosed Jesse de Beaux with drug-induced psychosis and treated him on that basis. The court-appointed psychiatrist later testified the diagnosis was wrong — de Beaux had genuine psychosis, unrelated to substance use, and "was not in control of his actions." Two days after leaving Graylands' care, he killed a Munster woman he had never met, in her own home.
Why it mattersThis is not an argument that clinicians acted with anything but good faith — diagnosing psychosis is genuinely difficult. It is the argument that when this specific kind of judgement goes wrong, the cost falls on someone outside the facility who had no part in the decision, and no document has been published showing how that risk is managed for a campus of this scale, this close to five schools.
Documented examplesSome leave models depend entirely on a patient's own compliance
What it isA leave arrangement — day release is the clearest example — where the only thing standing between supervised custody and unsupervised community access is the patient choosing, unmonitored, to keep taking medication and to come back.
What it looks likeDavid Lockyer Harman, a convicted rapist, spent weeks as the only patient in a secure six-bed ward before his condition was assessed as improved. That assessment triggered a security downgrade — and the downgrade came bundled with day release: free movement in the community by day, on the condition he returned to Graylands each night. He failed to return. Police said he could become violent specifically because he wasn't carrying the medication he needed, and warned the public not to approach him.
Why it mattersThe safeguard in this model isn't a fence or a monitor — it's the patient's own choice, unsupervised, to keep taking medication and come back. While that holds, there's no external check on it at all. The public only learns a window like this existed when it fails and becomes a police alert.
Documented examplesThe law now pushes toward community access on a legislated timetable, not just clinical discretion
What it isThe Criminal Law (Mental Impairment) Act 2023, which replaced indefinite custody with mandatory limiting terms and made supervised community access a standard outcome — a legal requirement, not a discretionary clinical call.
What it looks likeWhen the Act commenced, everyone previously held indefinitely had to be brought back before a review process for a limiting term to be set "as soon as practicable." Some of those backdated terms had already expired by the time they were set, producing what's now described as "unplanned discharges" — immediate release because the term was already served.
Why it mattersWe're not arguing against the reform — moving away from indefinite detention is broadly regarded as appropriate, and people under these orders are entitled to the least restrictive setting consistent with safety. We're asking a narrower question: given a legislated push toward more community access, for a cohort whose offence profile includes 12 wilful murders and 8 child sex offences on the last published count, has anyone assessed what that means for a site within about a kilometre of five schools?
Documented examplesThese five risk factors sit alongside — and are separate from — the governance and consultation failures set out on the Concerns page: no site-specific risk assessment, no published notification protocol, and no consultation with residents or schools before Stage 1 was funded. Read together, the pattern is the same one: decisions have been made without the evidence being shown to the people who have to live with the outcome.
Why "we'll review it if something goes wrong" isn't a good enough answer
The likely government response to this page is a version of what happened at Sydney's Cumberland Hospital: an incident, a promised review, additional security. That sequence has already been tried, recently, at a comparable facility — and it hasn't worked. NSW Premier Chris Minns ordered an urgent review within days of two absconding-linked deaths in February 2026; the health district promised bolstered security and independent oversight. Six months later, NSW Health's own data confirms at least 14 further absconding incidents, and the state's shadow mental health minister says the review didn't produce a major reduction. Cumberland's absconding rate remains three times the state average — the same multiple recorded before the review began.
Bennett Brook, WA's own comparable facility, shows the same shape at a smaller scale: government acted only after the third escape, not the first, and the underlying confidence problem has never fully resolved — official reports confirm it remains under-utilised nine years on. Full detail below.
The argument here isn't that reactive review is worthless. It's that a proactive, published assessment — before an irreversible $698 million construction commitment — is available now and hasn't been done, while the reactive alternative has a recent, repeated, documented record of not solving the problem it exists to solve.
What happens elsewhere when this isn't managed — and what happens after
Every case below follows the same shape: an incident, a government review promised in response, and — where enough time has now passed to check — a review that didn't fix it. That sequence is the whole argument for asking the question before construction, not after an incident forces it.
Two escapes, three deaths — then a review that still hasn't worked
On 7 February 2026, patient Luke Peter Francis overpowered a nurse, took a security access card and left Cumberland Hospital; about a week later he allegedly crashed a stolen vehicle in Camden, killing two women aged 60 and 84. The next day, patient Setefano Mooniai Leaaetoa, 25, escaped during a transfer to another facility; ten days later he allegedly stabbed three people in a busy Merrylands shopping strip, killing one. NSW Health's own data showed Cumberland's absconding rate was already more than three times the state average before any of this. Premier Chris Minns said something had "gone badly wrong" and ordered an urgent review.
That was six months ago. NSW Health's own data confirms at least 14 more absconding incidents at Cumberland since the review began — despite an urgent review ordered at Premier level and a promised security overhaul. The state's shadow mental health minister says the review "did not result in a major reduction in absconding." Cumberland's rate remains three times the state average — the same multiple recorded before the review started.
WA's own precedent — and its unresolved aftermath
Bennett Brook is WA's only other "declared place" for mentally impaired accused. Two residents escaped on New Year's Eve 2015. Lockridge residents said what Mount Claremont residents are saying now: the centre was too close to homes and a primary school, and they deserved to be told about escapes. Government did nothing after the first two escapes. It took a third, in May 2016, before Minister Helen Morton — the same Minister a Mount Claremont resident wrote to in February 2013, and appears never to have heard back from — commissioned an independent review and funded $640,000 in security upgrades.
Even that hasn't settled it. The Mental Health Advocacy Service's own 2024–25 report, tabled in Parliament this year, records the centre as still "under-utilised" — the identical finding Hon. Alison Xamon MLC made in 2020, when only three residents passed through a 10-bed facility in an entire year. Two official records, five years apart, say the same thing. This is the closest WA has come to an adequate response to exactly this kind of community concern — and it still took repeated harm to get there, and still hasn't resolved the underlying confidence problem nine years on.
More structure than Graylands has disclosed, and patients still don't return
Thomas Embling operates under considerably more oversight than anything published for Graylands — clinical assessment before leave, graduated escalation from escorted to unescorted, a Forensic Leave Panel chaired by a sitting judge, and a Chief Psychiatrist who can suspend leave immediately. Even so, patients on approved leave have failed to return on the public record in 2013, 2015, and October 2025, each triggering a Victoria Police alert. Three weeks after the October 2025 incident, a Victorian MLC raised on the record that Forensicare had acknowledged security issues with illicit substances inside the hospital, alongside staffing cuts to specialist roles proceeding regardless.
Other states don't put this in a residential suburb
NSW's Forensic Hospital at Malabar sits inside the Long Bay Correctional Complex — co-located with corrections, not standing alone in a general hospital campus in a residential suburb. Victoria's Ravenhall forensic service follows the same pattern at the Ravenhall Correctional Centre. Two states have proven this model operationally for at least part of their forensic capacity. WA has Hakea and Casuarina as equivalent sites, 22km from Mount Claremont. No independent assessment of that alternative for Graylands has been disclosed.
This pattern isn't confined to two states. Queensland's own Chief Psychiatrist annual reports have separately documented dozens of absconds a year from Gold Coast mental health units, including patients police described as posing a serious risk. Different state, different facility, same underlying shape.
The log
Every claim above is grounded in what follows. Each row separates two things on purpose: what happened, sourced directly, and why it matters, which risk factor it demonstrates and how. That split exists so the fact and the argument are never blurred together — you can check one against the other. Sourced from WA Parliament, mainstream media, coronial findings, and residents (see the Source column for each). Full citations for every non-resident entry are on the Sources page.
| Date | What happened | Why it matters | Source |
|---|---|---|---|
| 1987 | A Graylands patient in low-security care left the hospital and, in Perth's CBD, sexually assaulted a young girl. Colin Barnett — then the newly elected member for Cottesloe — put this on the record in Parliament three years later as an example the government already knew about. Full record → | Risk 1The earliest documented absconding-related harm on this site now on the public record — six years before the purpose-built Frankland Centre existed. | WA Parliament Hansard ↗ |
| 1990 | Barnett told the Legislative Assembly that three people he considered dangerous had absconded from Graylands that year alone, one of them having recently committed a serious stabbing. Full record → | Risk 1Multiple absconding incidents in a single year, read into the parliamentary record while government was still deciding whether to build a secure unit at all. | WA Parliament Hansard ↗ |
| Late 1990s–2000 | A pattern of inpatient deaths at Graylands. The ratio of suicides to admissions rose from ~2 per 1,000 (1987) to 6.6 per 1,000 (1999), with multiple cases showing failures of perimeter security and observation. | BaselinePerimeter and observation failures at this site predate every incident below — this isn't a recent problem. | WA Parliament Hansard ↗ |
| Sept 2004 | Two Frankland Centre patients, one with convictions for violent offences, smashed through a wall and jumped the security fence of a unit the Attorney-General had called "escape-proof" — it had reportedly been tested by the SAS. Police called them "extremely dangerous" and ran land and helicopter searches. | Risk 1Undercuts the premise that "maximum security" classification alone is sufficient — a unit built and tested to be escape-proof was defeated. | ABC News ↗ |
| Sept 2007 | Manuel Francisco, released from Graylands five days earlier, attacked mental-health nurse Christine Bruce with a sharpened screwdriver as she walked to work, inflicting 24 stab wounds. He was convicted of attempted murder. | Risk 3A release-timing judgement proved wrong within five days, with the harm falling on someone with no connection to the decision. | ABC News ↗ |
| 2010 | Gregory Hunt, a long-term involuntary patient, was moved between locked and open wards repeatedly, including two unauthorised overnight absences — the first ended when police returned him the next evening; the second when he returned on his own. | Risk 2The clearest example on record: each absence coincided with a transition between security tiers, not a static placement. | WA Coroner’s Court ↗ |
| July 2012 | David Charles Batty held his 79-year-old father hostage in a car at Kings Park for more than four hours, threatening police with a gas cylinder and lighter. Found unfit to stand trial, he was detained at the Frankland Centre — from where he would later abscond twice. | BackgroundShows how someone enters this system — and previews Risk 2: he'd go on to abscond during a security-tier transition, twice. | Newcastle Herald (AAP) ↗ |
| Dec 2012 | A patient walked out undetected. Two days later, police brought in an unrelated homeless man matching a general description, and staff administered Clozapine — an antipsychotic with a risk of fatal reaction — without confirming his identity. A subsequent audit found 12 patient bedrooms and 16 clinical rooms with no CCTV coverage. | Risk 1The response to an absconding event caused serious harm to an uninvolved member of the public — a consequence no published protocol addresses. | The West Australian ↗ |
| 15 Feb 2013 | A Mount Claremont resident reported a man — identified to him on the scene by police and hospital security as a Graylands patient — attempting to force entry into a neighbouring home, then moving through several more properties before police arrived. Full account below. | Risk 1A direct, named, local example of what an absconding event looks like from inside a resident's own street. | Read the full account ↓ |
| 2013 | Leaked clinical minutes revealed acute-care overflow patients being placed in rehabilitation beds, compromising security protocols — while the incidents either side of this row were happening. | ContextIndependently corroborates systemic strain at the exact time these other incidents were occurring — not an isolated lapse. | The West Australian ↗ |
| Apr 2013 | A patient in the secure male-only ward set a mattress on fire. About 40 patients and staff were evacuated. Damage exceeded $40,000. The Health Department investigated. | ContextDocumented failures at this site aren't confined to absconding — the same facility had a serious on-site security lapse the same year. | ABC News ↗ |
| 19 June 2014 | Batty absconded from the Frankland Centre during a transition between security levels, last seen at Claremont Quarter. His absence went unnoticed for about a week; the community wasn’t told. Then-Opposition mental health spokesperson Stephen Dawson — now a senior Minister in the Cook Cabinet — called it "an outrage." | Risk 2The clearest single example of Risk 2 — the failure happened specifically during a security-tier transition, and the notification gap it exposed remains undocumented as resolved twelve years on. | Newcastle Herald (AAP) ↗ |
| Oct 2014 | Enoch Walsh — held under a custody order after cutting his mother’s throat with a sword and being found not of sound mind, on a supervised leave arrangement between Frankland Centre and Romily House in Claremont — failed to return on 5 October 2014. Graylands wasn’t told until the next morning; police weren’t notified for almost 24 hours. | Risk 2Romily House is 2.3km from Graylands — a short trip on paper — and someone still went missing making it. Then-Opposition Leader Mark McGowan publicly criticised the notification delay. The Premier at the time was Colin Barnett — the same man who, as a backbencher in 1990, had told Parliament transition failures like this one were inevitable. His response as Premier was that the process had been handled properly, adding, "I'm sure he'll either turn up or be found." The full 1990-to-2014 story → | ABC News ↗ |
| July 2015 | Thirteen months on the run, Batty resurfaced in Bunbury, taking laundromat manager Peter Williams hostage for 12 hours. He threatened Williams with a speargun, put a chain around his neck, and falsely claimed to have explosives strapped to himself. He was sentenced to four years and nine months. | Risk 1Shows the outer bound of the timeframe: the consequence of a 2014 absconding wasn't resolved until a hostage crisis 13 months later, in a different city. | ABC News ↗ |
| Feb 2017 | David Lockyer Harman, a convicted rapist with a history of violent offending, spent several weeks as the only patient in a secure six-bed ward, monitored by security guards 24 hours a day. When his condition was assessed as improved, his security classification was downgraded and he was moved to a less secure part of the campus — a decision that also granted him day release, letting him move freely in the community on the condition he returned to Graylands each night. He failed to return. Police said he could become violent because he was not carrying the medication he needed, and warned the public not to approach him. He was located more than 24 hours later. | Risks 3 & 4The clearest illustration of the actual mechanism: a clinical downgrade decision unlocks community access that then depends entirely on the patient's own, self-enforced medication compliance — with no external check described anywhere in the public record. | PerthNow ↗ |
| Aug 2017 | Mark Thomas Morton, 40, went missing from Graylands Hospital. Police said he required regular medication and the public should not approach him. | Risk 4A second, independent case the same year: police again tied the danger he posed explicitly to his medication regimen, not to any assessment of character or history. | PerthNow ↗ |
| May 2018 | Mark Bradley Ribbons, 46, was last seen at Graylands Hospital before going missing. Wembley Police issued a public appeal; the public was told not to approach him and to call Police immediately. | Risk 1Risk 1 recurring at scale — a further public police alert connected to this site within the same few years. | 97.3 Coast FM ↗ |
| Dec 2018 | Graylands diagnosed Jesse de Beaux with drug-induced psychosis and treated him on that basis. The court-appointed psychiatrist later testified the diagnosis was wrong. He absconded two days after leaving Graylands’ care and fatally stabbed a Munster woman, a stranger to him, in her own home. The WA Supreme Court found him not guilty by reason of unsound mind; Justice Fiannaca pointed to "lost opportunities" to properly diagnose him. | Risk 3The most severe outcome in this entire record, and the plainest demonstration of Risk 3: a good-faith clinical judgement, later found wrong, with the cost falling on someone with no part in making it. | ABC News ↗ · what changed under CLMI ↗ |
| Oct 2022 | Frankland Centre beds sat empty — not from lack of demand, but because there wasn’t enough staff to safely admit another acute patient — while up to 20 mentally ill people waited in custody for a bed. WA’s Inspector of Custodial Services confirmed an average waitlist of 10–15 through the year. | ContextStaffing, not just security design, is a live constraint on how safely this population can be managed — relevant to any expansion that assumes current resourcing scales. | ABC News ↗ |
| 27 May 2023 | A patient went unaccounted for from Graylands, with allegations of threatening behaviour that day. Parliament asked what happened, how, and for how long before anyone noticed. The government’s only answer, fourteen months later, was that the word "escape" was "not correct" — with no account of what actually happened. | Risk 5The pattern extending to the present: asked directly what happened, the government's only response disputed the terminology rather than publishing an account — the transparency gap Risk 5 warns is structurally likely to recur. | WA Parliament, QoN 1121 ↗ |
| 2025 | A Mount Claremont parent, whose children walk past Graylands to and from John XXIII College, reported that a patient left the facility "in only a sheet." Full account below. | Risk 1The most recent entry in this log — confirming the pattern above hasn't stopped, it's simply mostly unreported. | Read the full account ↓ |
If you're a resident or parent, tell us about your experience with Graylands →
Absconding, by the numbers
Graylands doesn't publish an ongoing public count of patients who leave without authorisation. The most complete official breakdown we've found is a 2013 Parliamentary answer, obtained by the then-Member for Mining and Pastoral via a Department of Health FOI release, covering 2012–13. We searched for anything more recent published since — nothing has been. Read the original Hansard record ↗
| Facility | Reported absconders, 2012–13 |
|---|---|
| Graylands Hospital | 153 |
| Frankland Centre (secure forensic unit, part of Graylands) | fewer than 5 |
| Bentley Adolescent Hospital | 22 |
| Swan Hospital | 13 |
| Royal Perth Hospital | 7 |
| Albany Hospital | 6 |
| Princess Margaret Hospital | 6 |
| Armadale, Bentley, Joondalup, Kalgoorlie, Katanning, Nickol Bay, Peel Rockingham Kwinana Hospitals | fewer than 5 each |
| Six metropolitan Emergency Departments (combined) | fewer than 5 each |
| NGO and hostel accommodation (combined) | fewer than 5 each |
| Total, all WA facilities | 240 |
What "absconder" means here: the government's own answer defines it broadly — it includes patients on approved leave who simply didn't return by the specified time, not only unauthorised escapes. That caveat applies to every number in this table, Graylands included. Worth knowing when reading the comparison: Graylands is also WA's largest standalone psychiatric hospital, so some gap in raw numbers reflects its larger patient throughput, not necessarily a higher rate per patient — the data available doesn't let us calculate a rate. Even accounting for that, 153 out of a state total of 240 — 64%, from one hospital — is a lot to explain by size alone. Which part of Graylands this describes: not Frankland Centre, the existing secure forensic unit, which is broken out separately in the same 2013 answer at fewer than 5. Beyond that, the source doesn't say — "Graylands Hospital" here could mean the general wards, could include forensic-status patients held outside the Frankland building, or both, and nothing published lets anyone split it further. Civil involuntary patients have their own leave-of-absence and absence-without-leave provisions under the Mental Health Act, separate from forensic custody-order leave, so a large non-Frankland number doesn't require a forensic explanation — but it doesn't rule one out either. What isn't in question: the new 32-bed sub-acute unit Stage 1 adds is a lower security tier than Frankland, built around structured community access rather than Frankland's tighter model — so Frankland's own low number isn't evidence for how the new unit will perform.
Assaults on nurses, by the numbers
The most complete official breakdown we've found for this is a 2009 Parliamentary answer covering 2008–09 — also dated, and we couldn't find anything more recent published since either. Read the original Hansard record ↗
| Facility | Reported incidents of alleged assault on nurses, 2008–09 |
|---|---|
| Graylands Hospital | 386 |
| Royal Perth Hospital (incl. Shenton Park) | 163 |
| Fremantle Hospital and Health Service | 109 |
| Swan Kalamunda Health Service | 101 |
| Rockingham General Hospital | 100 |
| Sir Charles Gairdner Hospital | 72 |
| Princess Margaret Hospital | 61 |
| Bentley Health Service | 20 |
| Armadale Health Service | 6 |
| Osborne Park Hospital | 2 |
| King Edward Memorial Hospital | 1 |
What's counted: the government's answer says these figures include verbal assault, except at Sir Charles Gairdner, which reported physical incidents only. The Minister's answer separately listed the nature of assaults recorded across WA hospitals in this period: punching, kicking, pushing and shoving, slapping, spitting, biting, scratching, elbowing, assault with equipment or a weapon, stabbing with an object, grabbing, injury during restraint, hair pulling, pinching, and being smeared with faeces or urinated on. Worth knowing when reading the comparison: unlike the absconder numbers, size cuts the other way here — Royal Perth and Sir Charles Gairdner are both far larger hospitals than Graylands' 109 beds, treating a much higher volume of patients overall, and both recorded well under half of Graylands' incident count.
Who is held under a custody order, by the numbers
Graylands and the Frankland Centre don't publish a current breakdown of the offences behind the custody orders they hold. The most complete official one we've found is the Mentally Impaired Accused Review Board's own statutory annual report — required by law, tabled in Parliament. MIARB Annual Report 2014–15 ↗
| Offence for which a custody order was issued | Number of offences |
|---|---|
| Wilful murder | 12 |
| Murder | 4 |
| Attempted murder | 10 |
| Manslaughter | 2 |
| Unlawful killing | 1 |
| Sexual penetration of a child under 13 | 3 |
| Sexual penetration of a child under 16 | 8 |
| Indecent dealings with a child under 16 | 3 |
| Indecent dealings with a child, lineal relative | 3 |
| Grievous bodily harm | 3 |
| Unlawful wounding | 3 |
| Assault occasioning bodily harm | 9 |
| Aggravated armed robbery | 2 |
| Arson | 1 |
| Assault on a person working in a hospital | 1 |
| All other offence types recorded | 25 |
How current this still is: WA's own 2024–25 forensic mental health data shows 42 of 49 people currently under a custody order were carried over from the pre-2023 regime — so this 2014–15 profile still substantially describes today's cohort, even though it's over a decade old. Graylands and Frankland are confirmed in the same statutory report as WA's only authorised hospitals for this population. A parliamentary answer given 15 September 2026 puts a current number on part of that cohort: on 11 August 2026, 13 people in the Dryandra Ward were subject to CLMI custody orders, out of its 15-bed capacity WA Parliament, QoN 3174.
Separately, a parliamentary answer on the operation of the Criminal Law (Mental Impairment) Act 2023 itself — not specific to Graylands, but describing the same legal pathway this cohort moves through — gives the first public throughput figures since the Act commenced: of 12 Special Hearings held to date, 9 resulted in a finding that the accused committed the offence charged, 3 resulted in a not-guilty finding on account of mental impairment, and none resulted in acquittal. Of three District Court prosecutions discontinued before reaching a Special Hearing, two involved sexual offences and one involved stealing; all three were discontinued after an unfitness-to-stand-trial finding, for evidentiary or public-interest reasons, and the victim was consulted in every case WA Parliament, LC QoN 1916, 15 Sep 2026. This describes the legal pathway statewide, not this site specifically — but it's the most current public data available on how the process this cohort is held under actually resolves.
Two residents' accounts, in full
Names and street addresses are withheld; we hold them privately to confirm each contributor is a real, local resident. Everything else below is exactly what they reported.
A Mount Claremont resident wrote the same day to the Minister for Mental Health, the Director General of Health, and the Chief Psychiatrist. In his account: a man — identified to him on the scene by police and hospital security as a Graylands patient — attempted to force entry through his next-door neighbour's laundry door while the neighbour's wife was home alone with her 13-year-old daughter. He went to help; the man ran through several more properties before police arrived. Police confirmed the same man was believed responsible for other offences in the area that day. His own two daughters, aged 6 and 9, had recurring nightmares afterwards that the intruder would come back. He wrote that his family now had to take on the cost and burden of extra home security because of it, and asked the hospital to review its security procedures and tell residents the outcome.
No response to that letter has been located in the record we hold. The same duty-of-care and communication questions this campaign is asking in 2026 were put to government, in writing, thirteen years earlier — and appear to have gone unanswered.
"Our children walk to and from school past Graylands... there was an incident last year where someone escaped in only a sheet. I am concerned with the proximity of the facility and the school. Mostly Graylands is fairly quiet currently, but increasing the intake and size will increase the risk for the kids who walk past Graylands every day. Why increase the size of this facility in this location? It is a built-up area with schools and families. It would be better to build a purpose built facility outside of the residential area and use Graylands as an outpatient facility."
This is the point she's making: the facility is currently quiet enough that an incident like this is a story residents tell each other rather than a published record — and the plan on the table triples the size of the population next to that school.
Share your own experience → · See every source → · Read the full research paper →