The Unseen Shift.

Six findings

1. 1

   The evidence base is one study deep. The largest UK study of security officer mental health surveyed 754 operatives in 2020 (University of Portsmouth). No large UK study has repeated it since. Industry press calls it the only published research on the question.

2. 2

   Exposure defines the work. In that study, 64.6% of operatives took verbal abuse at least monthly, 43% were threatened with violence monthly, and over 30% were assaulted within a year. Attacks on lone workers rose 132% in the three years to 2024 (SoloProtect). One security guard, Gary Stanley, was killed on duty in Milton Keynes in February 2026.

3. 3

   Screening put PTSD symptoms at four in ten. 39.3% of surveyed operatives showed PTSD symptoms on the PCL-C screening tool. The sample was self-selected, so the figure is an alarm bell and a screen result. It still stands far above every comparable frontline baseline.

4. 4

   The suicide data is old and partial. The only official rate analysis (ONS, England, 2011 to 2015) put male security guards at the national average, inside the male elementary occupations group that carries 44% elevated risk. Since 2015 the ONS has published counts only: roughly 25 to 35 male security guard suicides a year in England. The HSE does not record work-related suicide at all.

5. 5

   The workforce concentrates the highest-risk profile. SIA licence data (June 2026): 88.9% male, modal age 26, concentrated in door supervision, paid near the wage floor, on 12-hour shifts and nights, often alone. Men are about three quarters of UK suicides, and only 36% of NHS talking-therapy referrals are men.

6. 6

   Nothing in the sector has been evaluated. Proven and promising interventions exist in rail, police, ambulance and construction. No evaluated mental health programme exists anywhere in UK private security. The gap is the opportunity.

Method Peer-reviewed studies, official statistics and industry data, gathered and verified August 2026. ONS suicide-by-occupation data files and the SIA June 2026 licence dataset were parsed directly. Every figure carries its source in the sources section. Claims that could not be traced to source were excluded, including one widely repeated industry violence figure.

Hundreds of thousands of licensed officers, young and overwhelmingly male

Where they are. London holds 127,659 licence holders, 28% of the national total. Manchester (31,630) and Birmingham (24,821) follow. 53.4% of holders are British nationals; the largest other nationalities are Pakistani (71,541), Nigerian (29,283) and Indian (21,718). The SIA publishes nationality and not ethnicity.

Why the profile matters. Private security outnumbers the police roughly three to one. The workforce skews young, male and migrant, spread across dispersed sites and subcontracting chains. Every one of those traits weakens the reach of conventional workplace health provision, and the same traits define the groups least likely to seek help.

Source: SIA licence holder demographics, June 2026 dataset (gov.uk), parsed from the official ODS file.

Violence and abuse are routine, and the trend is up

64.6%of operatives suffered verbal abuse at least monthly; half of those weeklyPortsmouth · 2020

43%threatened with violence at least monthly; one in ten dailyPortsmouth · 2020

30%+physically assaulted at work within a yearPortsmouth · 2020

+132%physical attacks on lone workers over three years; weapon incidents up 104%SoloProtect Impact Report 2024

2,000+daily incidents of violence and abuse against retail staff at the 2025 peak; ~1,600 a day in 2026, four times the 2020 levelBRC Crime Surveys 2025 / 2026

Later surveys repeat the pattern. A 2021 industry survey of over 1,000 security professionals found 51% face verbal abuse every shift, 36% are physically attacked monthly, and 48% suffer flashbacks after incidents. A 2024 GMB survey of over 900 Job Centre guards found 83% verbally abused within the year.

The law now protects the person beside them, and not them. The Crime and Policing Act 2026 created a standalone offence of assaulting a retail worker. Security officers are outside it. A Security Industry Federation petition for parity passed 10,000 signatures and drew a government response in May 2026.

The cost is occasionally fatal. Gary Stanley, 52, was stabbed to death on duty at Centre:MK, Milton Keynes, on 28 February 2026.

Sources: Talas & Button, Policing and Society (2020) · SoloProtect (Feb 2025) · BRC (2025, 2026) · TwinFM/Working the Doors (2021) · GMB (2024) · gov.uk · petition.parliament.uk.

One landmark study, six years old, and nothing since

The anchor. Talas, Button, Doyle and Das, University of Portsmouth: 754 operatives surveyed with SIA and GMB support, published in Policing and Society (2020). It used validated instruments (WEMWBS wellbeing scale, PCL-C trauma screen, CAGE-AID substance screen) plus 15 interviews. It remains the largest UK study of security operative mental health.

The headline.39.3% of respondents showed symptoms of PTSD. The study also found managers “unable or unwilling to accept that some employees were suffering from poor mental health” and a lack of employer provision for mental health support.

The caveat, stated plainly. The sample was self-selected and the PCL-C is a screen, and a diagnosis needs a clinician. The figure below is a comparison of screen results across different tools and samples. It sets the scale of the problem; a probability sample would set the true rate.

Since 2020: silence. Security Journal UK (2024) describes the Portsmouth paper as effectively the only published UK research on security industry mental health and calls for it to be repeated. Searches in August 2026 found no peer-reviewed UK follow-up, no SIA prevalence study and no BSIA survey. Smaller polls fill the space: useful for direction, unusable for rates. Internationally the picture is the same; the largest recent study (Taiwan, 2024, 1,032 guards) confirms violence exposure predicts worse mental health.

Sources: Portsmouth (2020) · Mind Blue Light (2019) · Police Care UK/Cambridge (2019) · Petrie et al. (2018) · Stevelink et al., PLOS ONE (2020) · Security Journal UK (2024) · Frontiers in Public Health (2024).

The state measured the risk once, eleven years ago

The one rate analysis.The ONS has published suicide rates by occupation once: England, deaths registered 2011 to 2015, ages 20 to 64. Male security guards sat at the national average (118 deaths, SMR 95). The small residual group “elementary security occupations n.e.c.” was significantly elevated (SMR 174). The wider group male security guards belong to, elementary occupations, carried 44% elevated risk (SMR 144), against SMR 48 for managers and directors.

Since then: counts only. Every later ONS occupation release publishes counts and states that counts cannot establish risk. The counts run at roughly 25 to 35 male security guard suicides a year in England (249 male deaths 2011 to 2021; 34 in 2023; 27 provisionally in 2024). No security-specific rate has been published since 2015.

The context. Men are about three quarters of the 6,190 suicides registered in England and Wales in 2024; the 2023 rate was the highest since 1999. The Hazards Campaign estimates around 650 UK suicides a year, roughly one in ten, may be work-related.

The blind spot.The HSE does not record or investigate work-related suicide; RIDDOR excludes it. A workforce of hundreds of thousands therefore has no current occupation-level suicide rate, no regulator recording work-linked deaths, and one rate analysis that predates Martyn’s Law, the pandemic and the retail crime wave.

Sources: ONS suicide by occupation (2017 analysis; 2011–2021 and 2023–2024 count datasets, parsed directly) · ONS suicide registrations 2023, 2024 · Hazards Campaign (2022) · CIEH (2019).

Five features of the work compound the risk

Together, these factors describe a workforce with elevated exposure, depleted buffers and structural barriers to help. Each factor is individually evidenced; their combination in security has never been studied as a system. That combination is the study worth commissioning.

Sources: HSE INDG73 · Security Matters/SoloProtect (2025) · Lee et al. (2017) · Torquati et al. (2019) · Indeed/Glassdoor (2025–26) · GMB (2023–24) · Breakroom (2026) · Eversheds Sutherland (2026) · SIA (2026) · Priory · Counselling Directory.

The interventions are proven elsewhere and untried here

The design lesson from the evidence. Fix the work before treating the worker: organisational measures first (rosters, hours, supervision, incident response), targeted support second, awareness last. Programmes that only train awareness or hand out an app do not move outcomes. Peer-delivered trauma support and gatekeeper skills work best inside a system that also controls exposure and measures results.

The precedent that maps cleanest. Rail: a dispersed, shift-based, public-facing workforce, trained at scale, with results counted year on year. Security officers hold the same bystander position in public space, at three times the police headcount, and no one has built them the equivalent.

Sources: Zalsman et al., Lancet Psychiatry (2016) · Nuij et al., BJPsych (2021) · Joyce et al. (2016) · NICE NG212 · Greenberg et al. (2010) · Network Rail/Samaritans (2025) · Health Promotion Int. (2023) · Lighthouse (2024) · HSE RR1135 · Fleming, IRJ (2024) · Oscar Kilo · Mind Blue Light (2019).

Seven things nobody can currently answer

1. 1

   The prevalence of mental ill-health in the workforce today. The last large study closed in 2020.

2. 2

   The current suicide rate for security occupations. The last published rate covers 2011 to 2015.

3. 3

   How many deaths are work-related. The HSE does not record work-related suicide; RIDDOR excludes it.

4. 4

   How many officers work alone, and for how long. No official figure exists.

5. 5

   What proportion of employers provide any mental health support. No open figure exists.

6. 6

   Whether any sector initiative works. None has been evaluated.

7. 7

   What the combination of exposure, nights, low pay, precarity and male silence does as a system. Never studied.

What credible research must do

Sample from the register. The SIA licence register is a complete sampling frame of the workforce. A probability sample drawn from it would replace self-selected surveys and set true prevalence for the first time.

Use validated instruments, longitudinally. Repeat WEMWBS and a current PTSD screen (PCL-5), add suicidal ideation measures, and follow the cohort so change can be attributed.

Link the deaths. Commission occupation-coded analysis of current suicide registrations for security SOC codes, and press the case for recording work-related suicide.

Trial at organisation level. Test rostering, supervision, post-incident response and peer support as controlled organisational interventions, and publish the results whatever they show.

Sources

OnSiren Research · August 2026 · Compiled and verified from primary sources; confidence ratings and data files retained.

More from OnSiren Research →