Community Safety & Crisis Response

1.1. The wrong responder, roughly every nine minutes

New South Wales has quietly made its police force the state's busiest front-line mental-health service. The NSW Police Force's own internal review of how it responds to mental-health incidents in the community found that, on average, a mental-health incident is attended or recorded by NSW police every nine minutes[i] — about 160 a day, carried right around the clock. The figure is the Force's own, drawn from its COPS records system. It is not a crime statistic, and the Force's own wording matters: some of those incidents are records created rather than crews dispatched.

· In 2022 the Force recorded 61,164 incidents involving a person experiencing a mental-health emergency where there was no associated criminal offence — up from about 43,000 in 2018[i]

· Releasing the review, the Government described more than 60,000 community mental-health incidents a year, rising about 10 per cent annually since 2018, and noted that in most of them no criminal offence had been committed, no violence was threatened and no weapon was involved[ii]

· NSW Health was still citing the nine-minute cadence in October 2025, listing it as evidence to validate its own proposed reform[iii]

· A Deputy State Coroner put the same finding into judicial findings in November 2024, alongside a finding that stretched community mental-health services had left police as the call of both first and last resort[iv]

The underlying count is 2022 data on a rising trend, so nine minutes is a conservative floor rather than a ceiling. We have not extrapolated it, because no agency has published a later count — and a policy page should not invent one.

Nobody designed this. It is what happens when every other service closes at five and the police radio does not.

1.2. When the only tool that arrives is a police tool

Police are trained to control a scene. Clinicians are trained to lower the temperature of one. Sending the first where the second is needed does not merely inconvenience the officer — it changes the odds for the person in crisis.

· The Law Enforcement Conduct Commission, reviewing five years of critical incidents, found that of 157 critical incident declarations between 2017 and 2022, 68 — 43 per cent — involved an interaction with a person in a mental-health crisis, and noted this may be an underestimate[v]

· The same Commission found police training on responding to someone in mental-health crisis "extremely limited": one day out of sixteen weeks of recruit training, with no role plays, and only 2,420 officers completed the four-day intervention course between February 2008 and September 2019 — fewer than the number of recruits attesting each year[v]

· In calendar 2025, 66 deaths were reported to the Coroner as occurring in custody or as a result of a police operation — the highest ever recorded in a single year — of which 27 followed a police operation, 7 of those reported as intentional self-harm and 5 arising in an operation to engage a person believed to be at risk of self-harm or in mental-health crisis[vi]

· Of the twelve police-operation inquests finalised in 2025, the Coroner recorded evidence of a diagnosis of mental illness in seven[vi]

The State Coroner's conclusion is not an advocacy position; it is a formal request. Reporting on those inquests, the Coroner wrote that "the need for a health led co-responder approach when Police encounter a person experiencing a mental health crisis has been raised in multiple inquests" and "remains a concerning issue in need of urgent attention", while adding the line that keeps this policy honest: a police response will almost always be required where there is a threat to public safety, but police are not always well placed to be the only first responders to severe distress[vi]. In the 2024 findings into the death of Todd McKenzie at Taree, the Deputy State Coroner recommended a full audit of the mental-health training given to officers of all ranks within two years[vii].

When the only service rostered on is the one with handcuffs, the outcome is decided before anyone knocks on the door.

1.3. Hours that never reach a crime

The second cost lands on the other side of the ledger. Every hour a general-duties crew spends on a welfare check, a hospital escort or a corridor handover is an hour not spent on break-and-enters, retail crime, domestic-violence compliance checks or highway patrol.

· The Auditor-General found NSW police were engaged in 265,430 welfare and mental-health support call-outs in 2023–24, up from 216,430 in 2019–20[viii]

· That is a deliberately broader count than the mental-health incidents behind the nine-minute figure — it includes welfare checks — and the two should never be treated as the same measure[viii]

· Police told the audit team they "spend many hours at hospitals with patients in mental health crises and this is removing them from core duties", and that escorting people to hospital, supporting NSW Ambulance and prison transport "is not part of the core police function"[viii]

· Of roughly 16,000 sworn officers serving 8.5 million people, only about 12,000 are frontline first responders — so 265,430 call-outs is on the order of 22 per frontline officer per year[viii]

· There were 130,040 mental-health-related emergency department presentations in NSW in 2023–24 — the destination of that handover queue[iii]

No published NSW figure exists for the average officer-hours lost per emergency-department handover. The Auditor-General says "many hours" and stops there[viii], and we would not manufacture a number the state does not collect. What can be said is that the Productivity Commission's national mental-health inquiry cites evaluations in which transport to emergency departments was between 27 and 63 per cent lower under co-response models, and an Allen Consulting Group estimate — Victorian data, from 2012 — that 2.8 hours of police time per case could fall to 0.7[ix]. The Commission also records that in 2018 NSW police completed about 14,700 orders under section 22 of the Mental Health Act 2007, and that such interactions had been estimated, as long ago as 2012, at about 10 per cent of police time[ix]. Queensland offers the only hard Australian arithmetic: its police service told a parliamentary committee that 16,246 mental-health calls in 2021 consumed 50,884 police hours — an average of 3.13 hours and 3.37 officers per call — and conceded it could not isolate the time spent waiting at hospitals[x].

An officer parked in an emergency-department corridor is not a mental-health service and is not a police service. It is the most expensive way New South Wales has found to do neither job.

1.4. The trials worked. The coverage never came.

The Police, Ambulance and Clinician Early Response program embeds mental-health clinicians with police. It is the model this state already knows works — and the model this state has never finished building.

· In July 2024 the responsible Minister confirmed PACER operated across 20 of the state's 57 police area commands and districts, of which exactly one — Lismore — is a regional command[xi]

· NSW Health's own submission to a parliamentary inquiry put the figure at 17 commands, within eight local health districts[xii]; the Auditor-General separately describes a program piloted in 2018 and still in a trial phase across 12 local commands, with "no confirmation that it will be expanded"[viii]

· The Ministry of Health does not maintain statistics on how many people PACER diverted from an emergency department or a police cell[xi]

· PACER cannot respond to incidents assessed as high-risk or involving a weapon[xii] — and in NSW it is structurally a police-led secondary response, unlike the ACT model where the officer's role is to keep the whole team safe[xii]

· The Minister for Mental Health told a budget estimates hearing that PACER "in its current delivery framework is not up to the task" and was "not funded beyond 2025"[xii]

The best NSW outcome data comes from a single site. South Eastern Sydney Local Health District reported that under the St George trial police time on scene fell to an average of 45 minutes, that mental-health presentations to the St George Hospital emergency department dropped by almost 10 per cent, and that of more than 1,500 clinician contacts only about 500 required hospital-based assessment or treatment[xiii]. Those are the health district's own figures for its own program, published alongside an expansion announcement rather than as an independent evaluation — which is precisely the problem. As recently as February 2026 the state's regional PACER expansion amounted to a single clinician commencing with the Richmond Police Area Command[xiv], and the Mental Health Coordinating Council had reported months earlier that the Lismore service was inactive for want of staff[xv].

Meanwhile the Government has moved past PACER altogether. NSW Health's October 2025 consultation pack sets out an approved-for-consultation health-led model: a 24/7 virtual state-wide service inside NSW Ambulance staffed by mental-health professionals, plus six Mental Health Acute Assessment Teams — all six across metropolitan Sydney — with police attending where there is a serious risk of harm or a real and imminent risk to life[iii]. Asked in November 2025 whether PACER would be extended to the Hawkesbury, the Police Minister pointed to that model instead[xvi]. The direction is right, and the Democrats say so plainly: the Legislative Council recommended a health-led response and a secondary police role, and the Government answered "supported"[xvii]. But consultation closed in November 2025, and we could find no published final model, funded rollout or commencement date. Six in-person teams, all of them inside Sydney, is not a state-wide service.

A pilot that has run for eight years across a third of the state is not a pilot any more. It is a decision not to finish.

1.5. A force 2,507 officers below its own authorised strength

This is the half of the argument that gets least attention and deserves most. The shortfall is not an advocacy estimate — it is the Force's own published table.

· At 30 June 2025 the police officer headcount was 15,887 against 18,394 authorised positions — a gap of 2,507 officers, or 13.6 per cent. Four years earlier NSW was 182 officers over authorised strength[xviii]

· On the Productivity Commission's population-adjusted measure — a different count from headcount, and not to be mixed with it — NSW had 208 operational police per 100,000 people in 2024–25, the lowest of any state or territory, against a national average of 264, and has fallen every year from 247 in 2020–21[xix]

· NSW also spends less per person on police than the country does: $622.05 a head against a national average of $652.04, even though it runs the largest police force in Australia[xix][viii]

· The Force self-reports an overall vacancy rate of 11 per cent; the Audit Office's own analysis puts vacancies at 13 per cent and long-term sick leave at 7 per cent, meaning about one in five of the workforce — 20 per cent — was unavailable for duty in June 2024[viii]

· Some commands are operating with vacancies above 30 per cent[viii]

The regional picture is worse than the average, and the state has stopped publishing the number that would show it. In November 2024 the Orana Mid-Western police district had 47 of 243 authorised positions vacant; once long-term sick leave and other leave were counted, the district was running at 70 per cent operational capacity, with some smaller stations at half headcount and others with no police personnel at all — against 83 per cent in some outer-metropolitan commands. Shortages there are forcing single-person patrols and call-outs. Before 2019 the Force reported monthly against a 90 per cent operational capacity target; since 2019 it has not reported against that target at all, and has done no analysis of authorised strength by command in eight years — while commands running well below capacity are still expected to meet their performance targets[viii]. Asked at budget estimates in February 2026 how many regional stations were below authorised strength, the Deputy Commissioner replied that he "would be very surprised if not all of our regional police stations are operating below their authorised strength"; the formal answer that followed declined to provide a station-by-station breakdown, though it did disclose that as at 8 March 2026 Traffic and Highway Patrol Command had 1,234 officers against 1,437 authorised positions — 203 short[xx]. A separate answer recorded 1,508 officers classified as injured as at 30 April 2025 — a point-in-time snapshot, but close to one in ten of the headcount[xxi].

The state's answer to "how many officers is Bourke short?" is that it does not publish the number.

1.6. The exit door is a medical certificate

The strength gap will not close by recruiting harder, because the leak is at the other end of the pipe — and it is a health problem.

· In 2023–24 about 4.5 per cent of all sworn officers exited for medical reasons, and 93 per cent of medical exits between July 2019 and June 2024 were psychological. This form of attrition now outnumbers any other, including resignation and retirement[viii]

· Across those five years 2,495 officers medically separated, 2,348 of them for psychological injury — rising from 302 in 2019–20 to 756 in 2023–24, an average increase of 26 per cent a year, against zero growth in physical-injury exits[viii]

· There were 1,208 new psychological injury claims in 2023–24; psychological claims made up about 28 per cent of all police compensation claims but 74 per cent of the cost[viii]

· Compensation for police psychological injury cost approximately $1.75 billion over the five years to 2023–24 — against $34 million invested over roughly the same period in proactive workforce wellbeing services[viii]

· Time lost to unplanned absence rose from an average 148 hours per staff member in 2020–21 to 216 hours in 2024–25 — about eighteen twelve-hour shifts each — while officers carrying excess annual leave they cannot take rose from 463 to 880[xviii]

Pay was the one lever the state pulled hard, and it worked on the problem it was aimed at. The 2024 award delivered a 19 per cent base increase over four years and was accepted by 96 per cent of voting members on a record turnout[xxii]. In the year that followed, resignations fell from 575 to 230 — while separations on medical grounds barely moved, 769 to 761, and now account for more than two-thirds of every officer who leaves[xviii]. The award commenced part-way through that year, so the timing is a coincidence worth noticing rather than a proven cause; the Auditor-General attributes a separate fall in psychological injury notifications to the Enhanced Police Support Scheme[viii]. Recruitment has recovered too — 710 attestations in 2024–25[xviii], 991 across calendar 2025[xx], and 349 in a single class in May 2026, the largest in more than thirteen years[xxiii] — but against 1,111 separations, headcount still fell by 365 in 2024–25[xviii]. And for a force of nearly 16,000 with more than a thousand new psychological claims a year, there are 27 mental-health clinicians embedded across the entire organisation[xx], who exist to support officers' own mental health and are not the crisis-response clinicians who attend call-outs.

Money fixed the pay problem. It did not fix the injury problem — and the injury problem is the one emptying the stations.

1.7. The shift that never reaches the street

Every serious conversation about police workload runs into the same wall: nobody measures it. The Auditor-General's finding is blunt — "NSW Police managers do not calculate the time that frontline police spend on administrative tasks in relation to time spent on active call-out duties"[viii]. So no honest percentage exists, and any page that quotes one is inventing it.

· Frontline officers told auditors that administrative workloads are onerous and that tasks such as preparing and serving court documents would be better handled by unsworn administrative staff, freeing police to respond to call-outs[viii]

· At some busy metropolitan stations officers are expected to arrive an hour before and stay up to an hour after a twelve-hour shift without additional payment, leaving them unable to take the ten-hour break between shifts — described to auditors as a continuous cycle of fatigue[viii]

· A Transformation Office project set up in 2022 to reduce that administrative burden had seen limited action in three years as at February 2025[viii]

The technology that would fix it is late and over budget. In May 2026 the Auditor-General found the Force had not efficiently or effectively planned and sourced key components of its core policing technology program: about $155 million spent — 47 per cent of the approved $328 million capital allocation — with the supplier unable to deliver to contract, phase one pushed from 2027 to June 2029, and a further $493 million now estimated to finish the job. COPS itself, the Force's primary system, still runs on 1990s technology in an obsolete programming language[xxiv]. But the same audit contains the proof that the argument works: mCOPS, a modest mobile application, is estimated by the Force to have cut average time on scene for inspections from 24 minutes to 12 and saved the equivalent of 610 twelve-hour shifts in its first eighteen months, and the adapted BluLink responder app is estimated to have saved at least $6 million in search-and-rescue costs in its first year[xxiv].

Six hundred and ten shifts came back from one app. Imagine what a finished system would return.

1.8. The crime that waits

An honest page has to say this out loud: most crime in New South Wales is not rising. Over the two years to March 2026 most major offence categories were stable, robbery fell 8.2 per cent, and the increases were concentrated in a few places[xxv]. "Crime out of control" is not a claim the state's own statistical agency supports, and we would not make it. What the data does show is a set of specific pressures landing on a force that is short of people.

· Steal from retail store rose 9.1 per cent over the two years to March 2026 and other stealing offences 3.1 per cent, with theft of motorised bicycles up 27 per cent to 2,080 incidents[xxv]

· BOCSAR reported in September 2025 that the recorded rate of retail theft was the highest in twenty years and 36 per cent higher than in 2006[xxvi]

· Adult court proceedings rose 10 per cent over two years — 16,218 more court actions — and domestic-violence-related charges 13 per cent, which BOCSAR attributes substantially to increased enforcement and police proactivity rather than to more offending[xxv]

· The 90-day legal action rate for domestic violence assault rose from 66 to a record 71 per cent, producing nearly 2,000 additional court actions[xxv]

That last figure is the whole argument in miniature: proactive policing works when officers have the hours to do it. And the domestic-violence workload keeps climbing. In the twelve months to December 2025 there were 37,828 recorded domestic-violence assault incidents, 26,063 breach-AVO offences, a record 25,393 domestic-violence intimidation, stalking and harassment offences, 89,155 non-criminal domestic-violence episodes and 32 domestic-violence murders[xxvii]. The Far West and Orana region records the highest domestic-violence assault rate in the state — 1,502.4 per 100,000 people against a NSW rate of around 446[xxvii] — which is to say the highest-demand region in New South Wales is policed by the district the Auditor-General found running at 70 per cent operational capacity[viii].

The capacity to investigate has not kept pace either. At the Auditor-General's 2022 audit, frontline police reported domestic and family violence made up half or more of general-duties workload against approximately 140,000 domestic-violence calls a year, yet the state policy function for it was a centralised team of six, with about 280 specialists across 57 commands each carrying between 500 and more than 3,000 incidents a year[xxviii]. On cybercrime, BOCSAR examined 39,494 reports made by NSW-resident victims between July 2019 and June 2022, involving more than $404 million in reported losses and rising 42 per cent across the period — and found 71 per cent were closed by police with no further investigation undertaken[xxix].

A crime nobody has the hours to investigate is, from the victim's point of view, a crime nobody took seriously.

Our plan has two halves that only work together. Take the crisis work off the police and the police still have too few officers, too much paperwork and too little investigative capacity. Fund the police and do nothing else, and the fastest-growing part of their workload stays a job nobody trained them for. We would fund both sides of the ledger — and we would say so in the same breath, every time.

2.1. Clinician-led crisis teams, state-wide, funded as core infrastructure

We support a health-led response to mental-health emergencies, and we would fund it as permanent emergency infrastructure rather than a program line that has to justify itself every budget cycle.

· A state-wide network of clinician-led crisis teams — mental-health clinician and peer worker — available 24 hours a day and dispatchable through triple zero alongside police and ambulance

· Recurrent funding with a published multi-year envelope, so that a service which works cannot quietly lapse for want of a renewal

· Warm handover rather than a waiting room: teams able to refer directly into community mental-health care, safe havens and peer support

The Legislative Council has already recommended a health-led response informed by the South Australian co-responder program, the Western Sydney Mental Health Acute Assessment Team and PACER, and recommended that police be activated as a secondary response only where required to support the safety of primary responders. The Government recorded both as supported[xvii]. We would hold it to that — and note that "supported" in a government response carries no money with it[xv].

A crisis line that is answered is a health service. A crisis line that is funded year to year is a pilot pretending to be one.

2.2. Health-first dispatch — with a risk threshold that survives contact with reality

This is where well-meaning versions of this policy fall over, and we would rather name the problem than route around it.

· Triple-zero triage that adds a genuine health pathway, with trained call-takers routing mental-health emergencies to the crisis team and police looped in on assessed risk

· Co-response for high-risk calls, with de-escalation leading and force the last resort

· A published, clinically governed risk threshold — reviewed after every critical incident involving a person in mental-health crisis

The reason the threshold matters is on the coronial record. Examining whether PACER could have changed the outcome of a fatal police shooting, the Deputy State Coroner found that no PACER staff would have been on duty at that hour and that, under current policies, the high-risk nature of the incident would have precluded a PACER clinician attending at all[iv]. The parliamentary inquiry found the same structural limit: PACER cannot attend incidents deemed high-risk or involving a weapon[xii]. A co-response model designed to stand down at the exact moment of greatest danger is not a substitute for police, and we would not pretend otherwise. The answer is to redesign the threshold and resource a response capable of meeting it — not to promise clinicians instead of police. The Productivity Commission's position is the sensible middle: police should ideally attend mental-health incidents where there is a threat to public safety or an imminent risk to staff, and are not well placed to triage people into mental-health care[ix].

"Send a clinician, not a cop" is a slogan. "Send both, and decide which one leads" is a policy.

2.3. Built on the evidence — including the parts that did not work

We would rather quote the numbers that survive checking than the ones that circulate.

· Evidence given to the NSW Legislative Council inquiry recorded that Denver's Support Team Assisted Response answered 2,500 calls in its first six months, of which only 748 required police assistance[xii] — an American program, and evidence to a committee rather than a finding of one

· The widely repeated claim that Eugene's CAHOOTS program needed police backup on under 1 per cent of calls has been corrected by the host police force itself: the Eugene Police Department's crime analysis of 2019 data put the diversion rate at approximately 5 to 8 per cent of calls for service and the backup rate at 2 per cent overall[xxx]

· CAHOOTS stopped operating inside Eugene in April 2025, when neither the clinic nor the city could carry the cost — reported as a joint announcement rather than in a government document[xxxi]. The model worked and was defunded anyway, which is exactly why we would fund crisis response recurrently

· A Netherlands study of Amsterdam's psychiatric ambulance found 82 per cent of patients transported by the ambulance and transport by police vehicle reduced to 1 per cent, with less coercion and no rise in aggressive incidents — and, honestly, longer transport delays[xxxii]

· Stockholm's psychiatric emergency response team attended 1,254 cases in its first year, a third of which needed no further action after assessment and on-site crisis intervention — a small, descriptive, ten-year-old study, useful for feasibility rather than proof[xxxiii]

· The strongest Australian evaluation is Queensland's: across 2,394 incidents, two-thirds were managed in the community and more than 70 per cent of intended involuntary transport authorities were averted, with an average 41 minutes on scene — a single-service, in-house before-and-after study, in another state[xxxiv]

Victoria is the cautionary tale. Its co-response pilot was evaluated once, in 2012, and its Royal Commission's recommendation for a health-led crisis response is still in implementation years later[xxxv]. Good models die of never being finished.

We would rather publish an inconvenient evaluation than campaign on a flattering rumour.

2.4. Regional parity, not a phone line

A crisis in Bourke deserves the same response as a crisis in Bondi, and right now it demonstrably does not get one.

· The Government's own proposed model funds six in-person teams, every one of them in metropolitan Sydney, with the rest of New South Wales served by a virtual service[iii]

· PACER's 20 commands include exactly one regional command[xi], and the most recent regional expansion was a single clinician[xiv]

· The region with the highest domestic-violence assault rate in the state[xxvii] is policed by the district the Auditor-General found at 70 per cent operational capacity, with some stations at half headcount and some with none[viii]

We would fund in-person crisis teams based in regional hubs across every local health district, with telehealth clinician support as a supplement to those teams and never as a substitute for them. Where a region cannot staff a team, we would fund the recruitment, housing and retention measures to make it staffable — because the Lismore service went inactive for want of clinicians, not for want of a launch[xv].

Telehealth is a tool. It is not a response. Nobody has ever been talked down by a phone that arrives on its own.

2.5. Rebuild the force to its own authorised strength

We are not asking for a smaller police force. We are asking for the one the state has already authorised and budgeted for, and we would measure the gap in public until it closes.

· Fill the 2,507-position gap between headcount and authorised strength[xviii], with priority to the commands running furthest below capacity[viii]

· Implement the Auditor-General's first recommendation — a responsive workforce allocation model matching police numbers to command-level workload, with staffing reviewed as workload changes. That recommendation carries a July 2026 deadline, which has now arrived, and the state should be asked to show its work[viii]

· Restore monthly public reporting against the 90 per cent operational capacity target abandoned in 2019[viii], and publish authorised strength against actual headcount command by command — a breakdown the Government declined to provide at budget estimates[xx]

· Keep the retention gains: the 2024 award is doing real work on resignations[xxii] and should be maintained and monitored, alongside the regional recruitment pathways the Force has already piloted[xviii]

· Treat recruitment as necessary but not sufficient — 991 attestations in calendar 2025[xx] did not stop headcount falling, because the medical exit door stayed open[xviii]

You cannot roster your way out of a shortage you refuse to publish.

2.6. Give the shift back to policing

Officer hours are the scarcest resource in this system, and New South Wales is spending them on the wrong things without even counting them.

· Measure it first: require the Force to calculate and publish frontline time on administrative tasks against time on active call-out duties, which the Auditor-General found it does not do[viii]

· Move the work that does not need a warrant card — including the preparation and serving of court documents — to unsworn administrative staff, as frontline officers themselves proposed[viii]

· Pay for handover time and protect the ten-hour break between shifts, so the fatigue cycle auditors documented stops being free to the employer[viii]

· Put the core policing technology program back on a published schedule with milestone reporting to Parliament, given $155 million spent, a four-year delay and a further $493 million now estimated[xxiv]

· Ship the small things while the big thing is built: one mobile app halved time on scene for inspections and returned the equivalent of 610 twelve-hour shifts in eighteen months on the Force's own estimate[xxiv]

Every hour returned to the frontline is cheaper than every hour recruited to replace it.

2.7. Investigative capacity where the crime actually is

Returned hours only matter if there is capacity to use them. We would direct that capacity at the crime the evidence identifies rather than the crime that headlines.

· Domestic and family violence: a properly resourced specialist function, against a 2022 audit finding that DFV made up half or more of general-duties workload and roughly 140,000 calls a year while the state policy team numbered six[xxviii], and a record 25,393 intimidation, stalking and harassment offences in the year to December 2025[xxvii]

· Fund the compliance work that prevents the next assault: a single four-day state-wide operation in May 2026 conducted 14,318 apprehended domestic violence order compliance checks and 1,847 bail compliance checks, identifying 391 alleged breaches[xxxvi] — surge capacity that ought to be ordinary capacity

· Retail and organised theft: investigative and analytical capacity aimed at repeat and organised offending, with retail theft at its highest recorded rate in twenty years[xxvi]

· Cybercrime: a real investigative pathway for volume fraud and online offences, so that 71 per cent of reports do not close without investigation[xxix], and public reporting of what happens to a NSW victim's report after they lodge it

Proactive policing works — the record domestic-violence action rate proves it. It just needs someone free to do it.

2.8. Look after the people who do the job

The state is already paying for police psychological injury. It is simply paying at the most expensive end, after the damage is done.

· Roughly $1.75 billion in compensation costs over five years against $34 million invested in prevention is not a saving — it is a deferred bill[viii]

· Implement the Auditor-General's remaining recommendations in full: report on the causal factors behind psychological injury claims, investigate and report on role overload and burnout, and link wellbeing programs to evidence-based psychological risk factors with a published evaluation framework[viii]

· Expand and evaluate the embedded clinician program — 27 clinicians for a force of nearly 16,000 with more than a thousand new psychological claims a year is not a scheme, it is a gesture[xx][viii]

· Reform the workers' compensation pathway so early intervention is easier than medical separation, when more than two-thirds of officers now leave the Force on medical grounds[xviii]

· Recognise that the single biggest preventable driver auditors heard about was workload made worse by staffing shortages[viii] — which is why sections 2.5 and 2.6 are officer-wellbeing policy as much as they are resourcing policy

A force that loses more people to injury than to resignation has a workplace problem, not a recruitment problem.

2.9. Honest measurement, published

We would rather argue about a published number than about duelling anecdotes — including when the number is inconvenient to us.

· Publish quarterly, district by district: mental-health call-outs by responder type, the share resolved on scene, the share ending in an emergency-department presentation, the share ending in involuntary detention or a use of force, and the officer-hours returned to general duties

· Fill the gaps the state currently has no answer for — the Ministry of Health does not keep diversion statistics for PACER[xi], and no NSW figure exists for hours lost per emergency-department handover[viii]

· Evaluate before expanding: NSW has never published an independent state-wide evaluation of PACER, and stakeholders asked the parliamentary inquiry for exactly that[xii]

· Name the gaming risk rather than bury it — "resolved on scene" can be recorded generously, and a target for fewer emergency-department presentations creates pressure to leave someone at home who should have gone to hospital

· Fund what the data supports and stop funding what it does not, including this policy, if it does not perform

These are the commitments the Australian Democrats would take to the NSW Parliament and press on whichever party forms government: funded crisis teams in every health district, a health-first dispatch protocol with a risk threshold worth the name, a police force restored to the strength the state has already authorised, and transparent public reporting so the whole thing can be checked.

Send the responder the call actually needs — and fund the police properly enough that they are free to answer the calls only police can.

Every number, legal citation and attributed finding on this page is referenced below, so you can check our work — the same standard we ask of government. Where a figure comes from an agency describing its own performance, or from another state or country, we say so.

[i] NSW Police Force: Summary Internal Review of the NSW Police Force response to mental health incidents in the community — internal review, released September 2024 (data year 2022)

[ii] NSW Government: "NSW Police Force mental health review released" — media release, 12 September 2024

[iii] NSW Ministry of Health: Stakeholder consultation pack — An Overview of the Proposed Health-Led Model for Consultation (October 2025)

[iv] Coroners Court of New South Wales: Inquest into the death of Andrew James Chee Quee — findings of Deputy State Coroner Harriet Grahame, 26 November 2024

[v] Law Enforcement Conduct Commission: Five Years (2017–2022) of Independent Monitoring of NSW Police Force Critical Incident Investigations — oversight report (2023)

[vi] Office of the NSW State Coroner: Deaths in Custody / Police Operations Report 2025 — report to Parliament under s 37(1) of the Coroners Act 2009 (NSW), covering calendar year 2025 (2026)

[vii] Coroners Court of New South Wales: Inquest into the death of Todd McKenzie — findings and recommendations of Deputy State Coroner Harriet Grahame, 5 April 2024

[viii] Audit Office of New South Wales: The mental health and wellbeing of NSW police — Auditor-General's performance audit report to Parliament, tabled 11 June 2025

[ix] Australian Government Productivity Commission: Mental Health, Inquiry Report No. 95, Volume 3 ("Justice") — national inquiry report (2020), citing Allen Consulting Group (2012) and the NSW Law Reform Commission (2012)

[x] Queensland Police Service: answers to questions taken on notice, Mental Health Select Committee public hearing of 12 April 2022 — Queensland Parliament (2022)

[xi] Parliament of New South Wales, Legislative Assembly: Question on Notice 3223 — PACER Program, answered 25 July 2024

[xii] Parliament of New South Wales, Legislative Council, Portfolio Committee No. 2 – Health: Report 64 — Equity, accessibility and appropriate delivery of outpatient and community mental health care in New South Wales — committee report, 4 June 2024 (including evidence of the Minister for Mental Health, 1 November 2023)

[xiii] South Eastern Sydney Local Health District: "St George PACER program success prompts government expansion" — local health district communications item reporting the program it operates (2020)

[xiv] NSW Government: "New mental health support teams hit the ground in Lismore" — ministerial media release, 20 February 2026

[xv] Mental Health Coordinating Council: one-year-on tracking of the NSW Government response to Report 64 — peak body monitoring report (September 2025)

[xvi] Parliament of New South Wales, Legislative Assembly: Question and Answer 6232 — Police Ambulance Clinical Early Response (PACER) Program, answered 20 November 2025

[xvii] NSW Government: response to Legislative Council Portfolio Committee No. 2 – Health, Report 64 — government response to recommendations (2024)

[xviii] NSW Police Force: Annual Report 2024–25 — "Flexible workforce" table, p. 31 (tabled in the NSW Parliament on 20 November 2025 as "Report of the NSW Police Force for the year ended 30 June 2025"). The published headcount line excludes paid student police officers, of whom there were 358 at 30 June 2025

[xix] Productivity Commission: Report on Government Services 2026, Part C Section 6 (Police services), data tables 6A.1 and 6A.2 — released 3 June 2026, covering 2024–25

[xx] Portfolio Committee No. 5 – Justice and Communities (NSW Legislative Council): Budget Estimates 2025–2026, Minister for Police and Counter-terrorism — hearing of 27 February 2026 and answers to questions taken on notice (2026)

[xxi] Parliament of New South Wales, Legislative Assembly: Question and Answer 5092 — Officer Numbers, asked 8 May 2025, answered 12 June 2025 (figures as at 30 April 2025)

[xxii] NSW Government: "Historic award agreement accepted by NSW Police officers" — media release, 26 November 2024

[xxiii] NSW Government: "Rebuilding our Force — largest police class in more than 13 years sworn in to keep NSW safe" — ministerial media release, 1 May 2026

[xxiv] Audit Office of New South Wales: Upgrades to core policing technology — Auditor-General's performance audit report to Parliament, 11 May 2026

[xxv] NSW Bureau of Crime Statistics and Research: NSW Recorded Crime Statistics quarterly update, March 2026 — released 18 June 2026 (24 months to March 2026)

[xxvi] NSW Bureau of Crime Statistics and Research: NSW Recorded Crime Statistics quarterly update, June 2025 — media release, 16 September 2025

[xxvii] NSW Bureau of Crime Statistics and Research: NSW Trends in Domestic & Family Violence — quarterly report, data to December 2025 (published March 2026)

[xxviii] Audit Office of New South Wales: Police responses to domestic and family violence — Auditor-General's performance audit report to Parliament, 4 April 2022

[xxix] NSW Bureau of Crime Statistics and Research: Trends in and characteristics of cybercrime in NSW — Bureau Brief No. 165 (2023), covering 1 July 2019 to 30 June 2022

[xxx] Eugene Police Department Crime Analysis Unit: CAHOOTS Program Analysis — City of Eugene, Oregon, United States (2020, analysing 2019 data)

[xxxi] Oregon Public Broadcasting and KLCC: reporting of the joint White Bird Clinic and City of Eugene announcement ending CAHOOTS services within Eugene, April 2025 — news reporting, United States

[xxxii] Zoeteman JB, de Wit MAS, de Haas HJ, Borkent KM, Peen J, Mulder CL and Dekker J: "Coercion During Psychiatric Ambulance Versus Police Transport in Mental Health Crises: A Pre- and Postimplementation Study" — Psychiatric Services 75(11):1144–1150 (2024), Netherlands

[xxxiii] Bouveng O, Bengtsson FA and Carlborg A: "First-year follow-up of the Psychiatric Emergency Response Team (PAM) in Stockholm County, Sweden: A descriptive study" — International Journal of Mental Health 46(2):65–73 (2017), Sweden

[xxxiv] Wyder M and Powell S: Metro South Addiction and Mental Health Services QPS and QAS Co-Responder Evaluation — Metro South Addiction and Mental Health Services, Brisbane (2022), covering March 2019 to February 2021, Queensland

[xxxv] Department of Health & Human Services Victoria: Police, Ambulance and Clinical Early Response (PACER) Evaluation Report (2012); and Victorian Department of Health: Royal Commission into Victoria's Mental Health System, Recommendation 10 implementation status

[xxxvi] NSW Police Force: "Police charge more than 900 people during Operation Amarok XIII" — media release, operation of 13–16 May 2026