Zak FentonWorking notes

· Note

Private medical insurance is sold on absence. Nobody has tested that.

It probably gets people seen sooner, by an amount nobody has measured. No study anywhere has set it against sickness absence or the premium. I sell it.

Evidence review (in progress)Private medical insurance: what the evidence supports

On Monday a trade magazine ran an analysis of group private medical insurance’s role in Keeping Britain Working, with views from eleven people across insurers, brokers and consultancies. The line that stayed with me came from a broker, who said the sector will still need to prove that the benefit reduces absence.1

I sell this benefit. In most of the benefit spends I see it’s the biggest line. So this week I went looking for the proof.

There isn’t any, for or against. No independent study anywhere has set group PMI against sickness absence, return to work or the premium. What it probably does is get people seen sooner. Nobody has measured by how much.

What it’s bought for

The NHS side of the queue is measured every month. At the end of August, 7.4 million pathways in England were waiting to start treatment, the median wait was 12.2 weeks, and 65 per cent had waited under 18 weeks against a standard of 92.2

The insured side is not measured at all. I could find no UK study that compares how long someone with private medical insurance waits for a consultant, a scan or an operation with an NHS patient who has the same condition, and no insurer publishes its waiting times. The nearest evidence is NHS patients sent to private hospitals, who are not insured. For hip replacement they waited 59 days against 121 at NHS hospitals, and the authors put most of the gap down to private hospitals taking simpler cases.3 Across twenty years of hip and knee replacements, private-provider waits were half the NHS’s.4 In Ireland, more people without insurance waited over three months than people with it.5

One trial is the caution. PhysioDirect cut the wait for physiotherapy from 34 days to 7 by putting a telephone call first. The wait for an actual appointment didn’t change.6 Being seen sooner and being treated sooner are different things.

So that’s the product. A shorter queue, probably, of a length nobody has measured, for the people on the policy.

What it’s sold on

Absence. One insurer’s paper says PMI adds £2.5 billion a year to the economy and gives the average small business 22 fewer sick days and £14,700 more output.7 The insurers’ trade body says health and protection insurance delivered £6.1 billion of benefits in 2021.8 A second insurer, adapting the trade body’s model, gets £750 million from a 25 per cent rise in take-up.9 A provider says it has proven that its plans pay off, at £2.87 back for every £1.10

None of these measured anything. The trade body’s model takes return-to-work effects from studies of other programmes, vocational rehabilitation and occupational health, and halves them to be cautious.8 The £2.5 billion paper publishes no inputs for the days it assumes are saved, and says in its own executive summary that the benefits remain under-researched.7 The £2.87 is arithmetic on an assumed absence saving for a model employee.10 No comparison group anywhere. No independent check of any of them.

Three models, one set of assumptions, a press release each. By the time the figures reach a renewal pack they’ve become findings.

The one study

One study has set employer-paid health insurance against sickness absence, in Denmark, where the cover was then tax free. Two surveys from 2009 and 2010, run through regression and matching. The main analysis found no significant effect on absence. The matching found none in one dataset and some sign of one in the other, where the matching was weaker.11 The paper also says that nationally and internationally there was no evidence on the question. There still isn’t. That’s the whole literature on the thing the benefit is sold on.

The waiting literature doesn’t rescue it. In Norwegian records a quarter of employees off sick were also waiting for specialist care, and how long they waited made almost no difference to how long they were off.12 A second Norwegian study found the opposite for one group: long waits for orthopaedic surgery pushed people who were already off sick into longer absence and disability benefit.13 A randomised trial of fast assessment is recruiting there now.14

Does sooner mean better

For a short wait, the trials say little changes. PhysioDirect found the same physical health at six months, and its authors wrote that the results challenge the assumption that faster access reduces time off work.15 Immediate physiotherapy for arm pain gave full recovery in 36 per cent at 26 weeks against 39 per cent after the usual six to eight week wait.16

For a long wait and a serious condition, the picture reverses. Every four weeks of delay to cancer surgery is linked to a 6 to 8 per cent higher risk of death, across 34 studies.17 Nobody doubts that.

So the honest version of the mechanism is narrow. For someone already off sick and facing a long NHS wait, getting them seen might shorten the absence. That’s plausible, it’s where the Norwegian study points, and it has never been tested with insurance.

Who it covers

The last government figures are from 2009 and 2010. One in five employers provided it, 18 per cent of small firms and 61 per cent of large, 21 per cent in the private sector and 5 per cent in the public.18 Twelve per cent of employees were on a scheme.19 Nothing representative has been collected since. HMRC counted 2.4 million people with the benefit in 2016 to 2017, said payrolled cover was a large gap, and then stopped counting anything but company cars.2021 The trade body says 4.8 million people are on workplace policies now.22 In the CIPD’s 2025 survey, which leans to large employers, a quarter of organisations offer it to everyone and another quarter offer it by seniority.23

Policies don’t normally cover pre-existing conditions or the ongoing management of chronic ones.24 And the premium carries three taxes: it’s a benefit in kind for the employee, the employer pays Class 1A National Insurance at 15 per cent on it, and insurance premium tax takes 12 per cent.252627

Put those together. The people most likely to be off sick for a long time, older, lower paid, with a chronic condition, are the people least likely to be on the policy and most likely to find what they need excluded from it. The benefit sold on absence is built around the people least likely to be absent.

What Keep Britain Working says

The final report mentions private medical insurance once, to repeat the CIPD’s coverage figures, without a footnote.28 Its list of return-on-investment evidence has employee assistance programmes, mental health and musculoskeletal programmes and employer case studies. Nothing for PMI.28 When the September update talks about pooled insurance for small firms, the product it names is group income protection.29 The same update says that today a provider sells a service based on claims of efficacy, and that with better data providers could compete on published performance instead.29 For this benefit, the data to compete on would be waiting times and absence, and nobody publishes either.

What I got wrong

My grid had PMI at very low, mixed, for turnover and for take-up. The turnover reading came from American job-lock studies, where leaving your job means losing your only health insurance. That mechanism doesn’t exist next to the NHS. The take-up reading rested on a Danish question about total care use. Both cells now read direction unknown, and the corrections are dated on the grid.

I’d also have told you the Danish study was a controlled test that found no effect. It’s two surveys with regression and matching, and the matching came out both ways. Inconclusive, which is weaker than I was saying.

What I’d do with this

Buy it for the queue and say so. A shorter wait for the people on the policy is the one thing the evidence points to, and for a serious diagnosis it matters. Nothing else about it has been measured.

Take absence, productivity and return on investment out of the business case. If a figure is in the renewal pack, ask who measured it and against what. It will be one of the three models.

Ask the insurer for its own waiting times: median days from authorisation to first consultation, to diagnostic, to treatment, by specialty. They hold the data. Nobody publishes it, and a benefit bought for speed should come with a speed.

Lay the list of people on the policy next to the list of people who were off sick for more than four weeks last year. If they barely overlap, the policy is a perk for the people on it. That’s a legitimate thing to buy. Call it that, and buy absence management for the other list.

What this can’t tell you

The evidence review behind this grades nine questions over 215 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. AI agents did the searching and the first pass, and I read every passage in the ledger against its source and checked each by machine. Two tiers, absence and presenteeism, were still being searched when I wrote this, so the absence section rests on the sources opened for the rest, and the ledger says where a study was read at abstract level, which includes both Norwegian waiting studies. No insurer publishes waiting times, so the access grade rests on NHS patients in private hospitals and on Ireland. I sell this benefit for a living.

What would change my mind

Insurer claims data linked to NHS records, comparing referral-to-treatment times for insured and uninsured employees with the same condition. A staggered roll-out of PMI across employers with payroll absence records before and after. The Norwegian trial, when it reports. If any of them comes back showing the absence effect the models assume, this note gets a dated correction, like the grid.

Footnotes

  1. Health & Protection (2026). Group PMI’s vital role in Keeping Britain Working as demand and costs climb: analysis. 2 November 2026. healthcareandprotection.com. The contributors are not named here by choice. ↩

  2. NHS England (2026). Statistical press notice: referral to treatment waiting times, August 2026. england.nhs.uk. ↩

  3. Kasteridis, P., Siciliani, L. and Sivey, P. (2026). Waiting time prioritisation for public and private providers: evidence from hip replacement surgery. Health Economics. doi:10.1002/hec.70140. NHS-funded patients, 2018 to 2019. ↩

  4. Kirkwood, G. and Pollock, A.M. (2025). Outsourcing National Health Service surgery to the private sector: waiting time inequality and the making of a two-tier system for hip and knee replacement in England. International Journal of Social Determinants of Health and Health Services. doi:10.1177/27551938251336949. ↩

  5. Whyte, R., Connolly, S. and Wren, M.A. (2020). Insurance status and waiting times for hospital-based services in Ireland. Health Policy. doi:10.1016/j.healthpol.2020.07.001. Abstract. ↩

  6. Salisbury, C., Foster, N.E., Hopper, C. et al. (2013). A pragmatic randomised controlled trial of the effectiveness and cost-effectiveness of ‘PhysioDirect’ telephone assessment and advice services for physiotherapy. Health Technology Assessment, 17(2). doi:10.3310/hta17020. Table 13. ↩

  7. WPI Strategy for Bupa (2026). Healthy workforce, healthy economy: a health economics paper. April 2026. wpi-strategy.com. ↩ ↩2

  8. WPI Economics for the Association of British Insurers (2023). Futureproofing workplace health: scaling up the impact of protection and health insurance. abi.org.uk. Chapter 3 and Table 1. ↩ ↩2

  9. WPI Economics for AXA Health (2025). Health and work: improving the resilience of the UK workforce. axahealth.co.uk. ↩

  10. Equipsme (2026). ROI 2026-2027: employers recoup £2.87 for every £1 spent on Equipsme plans. Company blog. equipsme.com. ↩ ↩2

  11. Pedersen, K.M. (2011). Sickness absence and voluntary employer paid health insurance. Health Economics Papers 2011:1, COHERE, University of Southern Denmark. sdu.dk. ↩

  12. Dyrstad, K., Halvorsen, T., Hem, K.G. and Rohde, T. (2016). Sick of waiting: does waiting for elective treatment cause sickness absence? Health Policy, 120(12), 1383-1388. doi:10.1016/j.healthpol.2016.10.004. Abstract. ↩

  13. Godøy, A., Haaland, V.F., Huitfeldt, I. and Votruba, M.E. (2024). Hospital queues, patient health, and labor supply. American Economic Journal: Economic Policy, 16(2), 150-181. doi:10.1257/pol.20210399. Abstract. ↩

  14. Aars, N.A.P., Bardal, I., Terjesen, C. et al. (2026). Multicentre randomised controlled trial of the Norwegian health in work service: the Norwegian Sickness Absence Clinic Efficacy Study. BMJ Open, 16, e109083. doi:10.1136/bmjopen-2025-109083. Protocol. ↩

  15. Salisbury, C., Montgomery, A.A., Hollinghurst, S. et al. (2013). Effectiveness of PhysioDirect telephone assessment and advice services for patients with musculoskeletal problems: pragmatic randomised controlled trial. BMJ, 346, f43. doi:10.1136/bmj.f43. ↩

  16. Jones, G.T., Macfarlane, G.J., Walker-Bone, K. et al. (2019). Maintained physical activity and physiotherapy in the management of distal arm pain: a randomised controlled trial. RMD Open, 5(1), e000810. doi:10.1136/rmdopen-2018-000810. ↩

  17. Hanna, T.P., King, W.D., Thibodeau, S. et al. (2020). Mortality due to cancer treatment delay: systematic review and meta-analysis. BMJ, 371, m4087. doi:10.1136/bmj.m4087. ↩

  18. Young, V. and Bhaumik, C. (2011). Health and well-being at work: a survey of employers. DWP Research Report 750. gov.uk. Table 3.1, fieldwork 2010. ↩

  19. Young, V. and Bhaumik, C. (2011). Health and well-being at work: a survey of employees. DWP Research Report 751. gov.uk. Tables 2.1 and 2.3, fieldwork 2009. ↩

  20. HM Revenue and Customs (2019). Taxable benefits in kind: national statistics commentary 2019, data for 2016 to 2017. gov.uk. ↩

  21. HM Revenue and Customs (2026). Benefit in kind statistics commentary, July 2026. gov.uk. ↩

  22. Association of British Insurers (2026). Insurers process record £4bn across individual and workplace health schemes. Press release, 21 January 2026. abi.org.uk. ↩

  23. Sinclair, A. and Suff, R. (2025). Health and wellbeing at work 2025. CIPD, supported by Simplyhealth. cipd.org. Appendix Tables 1 and 2. ↩

  24. Private Healthcare Information Network (2026). Paying for treatment with private medical insurance. phin.org.uk. ↩

  25. HM Revenue and Customs. Expenses and benefits: medical or dental treatment and insurance. gov.uk. ↩

  26. HM Revenue and Customs (2026). Rates and thresholds for employers 2026 to 2027. gov.uk. ↩

  27. HM Revenue and Customs. Insurance Premium Tax rates. gov.uk. ↩

  28. Mayfield, C. (2025). Keep Britain Working: final report. Independent review for the Department for Work and Pensions. gov.uk. Paragraph 125 and Appendix A. ↩ ↩2

  29. Mayfield, C. (2026). Keep Britain Working: a national growth opportunity. Review update, September 2026. gov.uk. Chapters four and five. ↩ ↩2

The evidence behind this note (in progress)Private medical insurance: what the evidence supports