Zak FentonWorking notes

· Note

The employee assistance programme is sold on a return nobody measured

Half of employers acting on stress use it as a method. Three to eight per cent of staff ring it. No trial has compared it with not having one.

Evidence review (in progress)The employee assistance programme: what the evidence supports

The trade body’s 2023 report says that for every £1 spent on an employee assistance programme, employers have seen an average return of £10.85, and that average usage has reached 12.14 per cent.1 The government’s Keep Britain Working review repeats the trade body’s figures: 75 per cent of UK employees have access, and £8 back for every £1.2 Deloitte’s 2022 report on mental health and employers counted the trade body’s return report as one of four new studies.3

I conducted an evidence review this week to find out who measured that.

Nobody did. The EAP is the benefit most large employers have, few staff use, and nobody has tested against not having one. The return it’s bought on comes from a form.

Where the figures come from

The trade body runs an online calculator. HR people type in their headcount, what the EAP costs, how many cases it handled and, where they don’t have the data, a default for how much absence the EAP is assumed to have reduced.4 The report averages what comes out. It says itself that the calculator’s users are “a self-selected group” with “an active interest in the performance of their EAP”, and that there’s no standard definition of usage, so most people enter the number of cases.15

That’s the £10.85. An average of what interested buyers typed into a form, built on an absence reduction they assumed. Deloitte counted the report of it as a study. By the time it reached the government’s review, it was a finding.

It matters because of where the EAP sits. In my experience it’s the line at the bottom of the stress policy, the absence policy and the bereavement policy, the thing HR points people to. The numbers below say how many employers lean on it, how few staff use it, and what’s known about what it does for them.

How many use it

It’s the benefit most large employers have: 78 per cent of large private sector organisations in the CIPD’s 2025 survey.6 Across all employers in Great Britain it’s 13 per cent.7

The government has asked employees three times whether they use it. In 2009, 40 per cent said their organisation provided access to counselling or an EAP, and 8 per cent of those who had it had used it in the last 12 months.8 In 2014 it was the least used policy asked about, at 3 per cent of those with access.9 In 2024 to 2025, 7 per cent of workers with occupational health available had used an EAP in the year.10 Among NHS staff in the pandemic, 5.3 per cent.11 The trade body’s own provider data, published separately from the calculator, give 3.34 per cent.12 The government’s modelling for its new support package assumes 5 per cent.13

Three to eight per cent of the people who have it use it in a year, from surveys with no stake in the answer. Whether the rest know they have it, nobody has measured nationally.

The thing the policy points to

The CIPD’s 2025 survey puts a number on the signposting. Among organisations taking steps on stress, the EAP was the second most used method, at 52 per cent, behind only flexible working. Risk assessments or stress audits came in at 38 per cent, a written stress policy at 30 per cent, and HSE’s Management Standards at 20 per cent.6 The CIPD’s own reading is that employers’ activity tends to be more reactive than proactive.6 HSE says employers have a legal duty to assess the risk of stress at work and act on it.14

So in more organisations than do the risk assessment, the EAP is the stress response. A service 3 to 8 per cent of staff use is carrying half of employers’ answer to stress. Nobody has studied whether pointing someone to it changes anything for them, or whether having something to point to is what stops the risk assessment getting done.

What it does for the people who use it

No randomised trial has compared an EAP with no EAP. A Cochrane review of return to work after adjustment disorders listed EAPs as eligible and found none.15 The one trial that involves an EAP tested telephone CBT added to one, and its authors called the EAP-only arm a weak control because almost nobody used it.16

What exists is matched comparisons, people who chose to use the service set against people who didn’t, matched on what could be measured. NICE graded the main one, a US study of state employees, for its 2022 workplace guideline. Absence was 0.23 standard deviations lower among users, with an interval from 0.44 to 0.01 lower, which NICE calls a benefit. Distress and presenteeism showed no difference. All of it very low certainty, after downgrades for the sample, for 35 to 45 per cent of people dropping out, for self-report and for confounding.17 The study’s own abstract reports depression and anxiety reduced over two to twelve months.18 A German matched study found self-rated productivity improved and absence didn’t change.19 A Canadian one found less distress at six months.20

NICE’s committee summed it up as very low quality evidence from a single study, which may reduce absenteeism, with no effect on presenteeism or distress. It made a “consider” recommendation because of the lack of evidence, and asked for research on the long-term impact and cost effectiveness.1721

Short counselling does something for a while. In UK primary care trials, counselling beat usual care by 0.28 standard deviations at one to six months and by nothing detectable at seven to twelve.22 In routine UK workplace counselling data, 6,459 clients averaged 6.9 sessions and 62.5 per cent improved, with no comparison group to say how many would have improved anyway.23

So people who use it often feel better afterwards. Whether they’d have felt better anyway, whether it lasts, and what it does for absence, is very low certainty and points both ways. For getting people back from long-term sick leave, there is no study at all. In the latest government survey, 6 per cent of employees who’d had a long-term absence had talked, or would talk, to an EAP adviser about getting back.10

All of that is about counselling. The EAP also sells legal advice, financial advice and manager support, and nobody has studied any of those.

The return

NICE found no published cost-effectiveness study for the organisational interventions it reviewed, EAPs included.17 The one independent controlled evaluation I could find, a 400-employee Australian firm in the 1990s, found the programme broke even. Benefit cost ratio 1 to 1.0.24 A US provider’s 2025 model reports $5.39 per $1 and says in its own limitations that there was no control group.25

That’s the whole independent economic evidence. One break-even study, abroad, from the 1990s.

What I got wrong

Two weeks ago I published a grid of 32 benefits against ten outcomes. The EAP sat in it at low certainty, improves, for mental health and for productivity. The mental health cell leaned on the CBT trial, which doesn’t test the EAP. The productivity cell didn’t downgrade the matched study NICE grades very low. Both now read very low, mixed. Return to work moves to none found, and take-up moves up to moderate, for the narrow statement that use is low. The corrections are dated on the grid.

I’d also have told you the use figures all come from providers. Half right. The 12 per cent does. The government’s figures exist, and they’re lower.

What I’d do with this

I wouldn’t cancel it. A confidential line that a few per cent of staff ring in a bad year is cheap, and the people who ring it are mostly already unwell.26

I’d stop writing absence, productivity and return on investment into the business case, because nobody has measured them for an EAP in the UK. If a figure of £8 or £10 per £1 is in your renewal pack, ask who measured it and against what. It will be the calculator.

Ask the provider how many unique employees used it in the year, and compare that with the 3 to 8 per cent above. Cases and calls count something else. If it’s lower, nobody knows it’s there.

Find out what happens at the edges. What’s escalated when someone discloses risk, what follows the session limit, and how long people wait. None of that has been studied either.

And keep the stress risk assessment, the adjustments and occupational health running alongside. The EAP is a phone line. The plan is the rest.

What this can’t tell you

The evidence review behind this grades eight questions over 176 sources and shows 53 claims, each with the passage it rests on, where it sits in the source, and what that source can’t tell you. I read every one of those passages against its source and graded the evidence with GRADE logic. AI agents did the searching and the first pass, and a random 30 per cent of the sources were reopened by separate verifiers. Where a study was behind a paywall the ledger says so, and that includes the one NICE graded, so I couldn’t check who paid for it. Three figures from the latest government survey are read off its charts, because it publishes no table for them. I sell employee benefits for a living, and this is one of them.

What would change my mind

A randomised trial of EAP access against usual care in UK workplaces, independent of providers, with symptoms and absence at six and twelve months. NICE asked for exactly that in 2022. A government survey question that asks every employee whether they have an EAP and whether they used it. If either comes back the other way, this note gets a dated correction, like the grid.

Footnotes

  1. UK Employee Assistance Professionals Association (2023). Financial return on EAPs 2023: managing the rise of complexity and employees at risk. eapa.org.uk. ↩ ↩2

  2. Mayfield, C. (2025). Keep Britain Working: final report. Independent review for the Department for Work and Pensions. gov.uk. ↩

  3. Hampson, E., Henry, J., Assal, V., Polner, A. et al. (2022). Mental health and employers: the case for investment, pandemic and beyond. Deloitte. deloitte.com. ↩

  4. Bevan, S. and Bajorek, Z. (2019). Designing and testing a Return on Investment tool for EAPs: final report. Institute for Employment Studies, Report 515, for the UK Employee Assistance Professionals Association. eapa.org.uk. ↩

  5. UK Employee Assistance Professionals Association (2022). Financial return on EAPs 2022. eapa.org.uk. ↩

  6. Sinclair, A. and Suff, R. (2025). Health and wellbeing at work 2025. CIPD. cipd.org. Figure 10 and Appendix Table 2. ↩ ↩2 ↩3

  7. Department for Work and Pensions (2023). DWP Employer Survey 2022: research report. gov.uk. Table 3.4. ↩

  8. Young, V. and Bhaumik, C. (2011). Health and well-being at work: a survey of employees. DWP Research Report 751. gov.uk. Table 2.3, fieldwork October to December 2009. ↩

  9. Steadman, K., Wood, M. and Silvester, H. (2015). Health and wellbeing at work: a survey of employees, 2014. DWP Research Report 901. gov.uk. Table 3.8 and Figure 3.3. ↩

  10. IFF Research (2026). Survey of Employees and Self-Employed Workers, 2024 to 2025. DWP Research Report 1117. gov.uk. Figures 7.6 and 6.4, read from the charts. ↩ ↩2

  11. Croak, B., Lamb, D., Stevelink, S.A.M. et al. (2026). Inequalities in NHS staff support among those from ethnic minority and migrant groups during the COVID-19 pandemic. Occupational and Environmental Medicine. doi:10.1136/oemed-2025-110203. ↩

  12. UK Employee Assistance Professionals Association (2025). Focusing on the job: EAPs and keeping Britain working. eapa.org.uk. ↩

  13. Department for Work and Pensions (2025). Keep Britain Working: technical note. gov.uk. ↩

  14. Health and Safety Executive. Work-related stress: stress risk assessment. hse.gov.uk. Read 27 October 2026. ↩

  15. Arends, I., Bruinvels, D.J., Rebergen, D.S. et al. (2012). Interventions to facilitate return to work in adults with adjustment disorders. Cochrane Database of Systematic Reviews, CD006389. doi:10.1002/14651858.CD006389.pub2. Abstract. ↩

  16. Furukawa, T.A., Horikoshi, M., Kawakami, N. et al. (2012). Telephone cognitive-behavioral therapy for subthreshold depression and presenteeism in workplace: a randomized controlled trial. PLoS One, 7(4), e35330. doi:10.1371/journal.pone.0035330. ↩

  17. National Institute for Health and Care Excellence (2022). Mental wellbeing at work (NG212). Evidence review A: organisational universal-level approaches. nice.org.uk. The graded study is Richmond, M.K., Pampel, F.C., Wood, R.C. and Nunes, A.P. (2017), Journal of Occupational Health Psychology, 22(2), 170-179. ↩ ↩2 ↩3

  18. Richmond, M.K., Pampel, F.C., Wood, R.C. and Nunes, A.P. (2016). Impact of employee assistance services on depression, anxiety, and risky alcohol use: a quasi-experimental study. Journal of Occupational and Environmental Medicine, 58(7), 641-650. doi:10.1097/jom.0000000000000744. Abstract. ↩

  19. Zieringer, R.C. and Zapf, D. (2024). The effects of an employee assistance program on productivity at work, workability, absenteeism, and smartphone measures of heart rate and heart rate variability. Journal of Occupational Health Psychology. doi:10.1037/ocp0000380. Abstract. ↩

  20. Milot, M. (2019). The impact of a Canadian external Employee Assistance Program on mental health and workplace functioning. Journal of Workplace Behavioral Health, 34(3), 167-191. doi:10.1080/15555240.2019.1609978. Abstract. ↩

  21. National Institute for Health and Care Excellence (2022). Mental wellbeing at work (NG212). Recommendation 1.4.6 and recommendations for research. nice.org.uk. ↩

  22. Bower, P., Knowles, S., Coventry, P.A. and Rowland, N. (2011). Counselling for mental health and psychosocial problems in primary care. Cochrane Database of Systematic Reviews, CD001025. doi:10.1002/14651858.CD001025.pub3. Abstract. ↩

  23. Stiles, W.B., Barkham, M. and Wheeler, S. (2015). Duration of psychological therapy: relation to recovery and improvement rates in UK routine practice. British Journal of Psychiatry, 207(2), 115-122. doi:10.1192/bjp.bp.114.145565. ↩

  24. Blaze-Temple, D. and Howat, P.A. (1997). Cost benefit of an Australian EAP. Employee Assistance Quarterly, 12(3), 1-24. doi:10.1300/j022v12n03_01. Abstract. ↩

  25. Attridge, M. and Pawlowski, D. (2025). Return on investment for employee assistance programs: conceptual model and outcomes from 166,116 employee cases. International Journal of Scientific and Research Publications, 15(5). doi:10.29322/ijsrp.15.05.2025.p16113. ↩

  26. Arthur, A.R. (2002). Mental health problems and British workers: a survey of mental health problems in employees who receive counselling from Employee Assistance Programmes. Stress and Health, 18(2), 69-74. doi:10.1002/smi.930. Abstract. ↩

The evidence behind this note (in progress)The employee assistance programme: what the evidence supports