The Luck of the Draw
What thirty years of injured workers told me, and what the research finally say.
I’ve worked in rehabilitation for more than thirty years, and in that time I’ve sat across from hundreds of injured workers. I can count on one hand the number who told me their case manager was a good one, someone who genuinely worked with them rather than to their own agenda. Rarer still is the person who had the same case manager the whole way through. Some of my clients have had nine, ten, eleven case managers across the life of a single injury, and each new one arrives with the file but not the history, so the worker starts again. You explain how you were hurt. You explain what you can’t do anymore. You explain why you’re still not back at work, to someone who’s meeting you for the first time and has already formed a view from the paperwork. Then a few months later you do it all again.
I’ve watched people become smaller with every retelling, and for most of my career the only thing I could say about it was that I’d seen it happen a lot. What I find genuinely encouraging now is that we finally have the research to say it properly, with numbers instead of the anecdotes I’ve been carrying around for three decades. The work that’s come out over the past two years makes an argument I think many of us have half known for years, which is that a good deal of what we record as the injured person’s problem is something the system did to them.
Why does luck keep appearing in the return to work research?
Start with a Norwegian interview study by Amundsen and colleagues, published online last October. They interviewed 29 people with persistent pain who were unemployed and receiving benefit payments, and analysed the transcripts using the Framework method. Twelve had gone back to work in some capacity and seventeen hadn’t. The obstacle themes were the ones you’d expect, covering the welfare administration, healthcare, psychological factors, pain, and how people saw work itself. The facilitator themes are where it gets interesting, because alongside the predictable ones, two more came out of the data. One was the disability pension. The other was luck.
Nobody sets out to measure luck. It appeared because participant after participant described a chance encounter with a particular staff member who happened to take an interest, happened to make a call, happened to understand. When luck turns up as a category in qualitative data it usually means something real is operating that people can see clearly but can’t name, and what these participants could see was that the system produced different outcomes depending on who inside it they happened to meet. Which is what my clients have been telling me for thirty years, in almost those words.
The barriers we record are partly something the system did
A second Norwegian paper, this one from Myhrer and colleagues in February, goes further. They’ve proposed that the barriers to returning to work should be understood as iatrogenic effects of the sickness absence itself, so not characteristics the person brought with them but side effects of the process they’ve been put through, producing what the authors call a lock-in effect and an exclusion trap. They’ve built a questionnaire to measure it, tested at a Norwegian sickness absence clinic, and they’re candid that it still needs revision. The most common barriers their participants reported were beliefs about health, so beliefs about what their bodies could tolerate, about the risk of doing more damage, about whether they were still the sort of person who could hold down a job.
Where do those beliefs come from? Some come from the injury. Rather more come from eighteen months of medical certificates that describe capacity in the negative, independent examinations that carry an air of doubt, and a claims process where every improvement has to be reported and every setback explained. And some of them come from telling your story to your eleventh case manager.
This isn't a problem of case managers trying harder
None of this is an argument that case managers are the problem, and it would be a poor reading of the evidence to conclude that better people would fix it. A German meta-synthesis by Baasner and colleagues examined sixteen qualitative studies of what makes return to work counselling effective, drawing on research from Australia, North America, the Nordic countries and Germany, and their model has four levels with the individual counsellor sitting at only one of them. Above it sits the counselling relationship, the team and stakeholder structures, and what they call the extended organisational level, which is whether counsellors have enough time and resources to do the work at all. Their conclusion is worth quoting to anyone who sets caseload targets, because what they found is that effectiveness in counselling is a collective outcome rather than an individual one.
A Danish scoping review from Poulsen and colleagues adds to this. They screened more than eight thousand records to find 22 studies of collaboration between general practice and sickness benefits offices, and the barriers recurring across two decades were stereotyping, differing priorities, and an over-reliance on written communication. Face to face contact helped. Written communication did both, in that high-quality written communication helped and dependence on it hindered. That’s an uncomfortable finding given the direction nearly every scheme in the developed world has travelled, towards portals and standardised forms and certificates that turn up without a conversation attached.
We can't measure what the field hasn't agreed to name
There’s a reason none of this has yet turned into practice change. Popovic and colleagues went looking for how injured worker stigma is described in the literature and found a hundred papers describing it, using 271 different terms between them, with fewer than a quarter of the papers that used the word stigma using it consistently. You can’t design a system to reduce something the field hasn’t agreed to name, and you certainly can’t measure it.
Two conversations, and 16 fewer sick days per person
So what do we actually do? The Norwegian MI-NAV trial, published in the Scandinavian Journal of Work, Environment and Health last September, randomised 509 workers on sick leave with musculoskeletal disorders to usual case management, usual case management plus two sessions of motivational interviewing delivered by their existing caseworkers, or usual case management plus a stratified vocational advice intervention delivered by a physiotherapist. Both additions beat usual care, with roughly 16 and 18 fewer sick leave days per person over twelve months, and net benefits of around €5,200 and €7,200 per person.
Two conversations, conducted differently, by the same people who were already having conversations. That’s what moved the outcome by more than two weeks of work and paid for itself several times over. Whatever it is that makes someone one of the good ones, it isn’t a fixed personal quality handed out by chance, it’s a practice, and a practice can be taught and supervised and resourced and measured. Which means that leaving it to chance is itself a choice, and we keep making it.
The accommodation we're least likely to provide is the one that works best
There’s a second choice like that hiding in a German study published last month. Sikora and colleagues looked at two longitudinal datasets of employees with common mental disorders returning after long term absence, and between 91 and 99 per cent said they needed a work accommodation, 64 per cent received at least one, and only 13 to 16 per cent had all their needs met. The gap isn’t random either, because what gets delivered is the administratively easy material like graded hours and a reduced load, while what goes undelivered is any reorganisation of how the work is actually done. That matters more than it looks, because an Australian led meta-analysis by Sanatkar and colleagues found that work focused interventions built around modifying the work outperformed those built around manager cooperation. The accommodation we’re least likely to provide is the one the evidence most supports.
Three things this should change in practice
If that’s right, three things follow for practice, and none of them need a new program or a new funding submission.
The first is that our assessments should record what the system has done to someone, and not only what that person presents with. If barriers to return to work are partly iatrogenic, then the number of certificates issued, the number of independent examinations attended, the number of case managers the file has passed through, and the months since anyone sat down with that person face to face, are all clinical variables. Most of us can pull those numbers in ten minutes. Almost none of us put them in the report.
The second is that we should be treating an unmet accommodation need as a clinical finding. Sikora’s data show that receiving even one accommodation predicted lower depressive symptoms and better work ability afterwards, so a file note recording that the employer declined a duties redesign is, on that evidence, documentation of a deterioration risk, and it deserves the same escalation we’d give a worsening symptom score.
The third is harder, and it’s about what we count. A meta-analysis by Backes and colleagues pooled 23 studies of occupational health interventions and found no meaningful effect on sick days at all, at minus 0.18 days with a confidence interval straddling zero, while return on investment trended positive. Their explanation is that the benefit is showing up in presenteeism, which almost nobody measures. If they’re right, a lot of us have spent years reporting a number that was never going to move, and defending it to funders who could see perfectly well that it wasn’t moving.
We measure the person and leave the system alone
I don’t think our systems are broken, which is both unhelpful to say and not quite accurate. I think we’ve become very good at measuring the injured person and rather casual about managing the system around them, and this year’s evidence suggests the returns sit in doing the reverse. It also suggests the fix is smaller and cheaper than we’ve been telling ourselves, which is either encouraging or uncomfortable depending on how long you’ve been arguing for the budget.
And the handful of clients who told me they’d had a good one? They were right about their case manager. They were only wrong to think they’d been lucky.
Evidence referred to in this article
Dates given are when each paper first appeared online, which is often well before its issue year.
Amundsen, P. A., Irgens, P. M. S., Burton, K., Malmberg-Heimonen, I., & Froud, R. (2025). Obstacles and facilitators of return to work among people with persistent pain who receive benefit payments: an in-depth interview study. BMC Public Health, 25(1), Article 3532. https://doi.org/10.1186/s12889-025-24747-0 [online 21 October 2025]
Baasner, A.-L., Petrak, S., Albersmann, L., Gröhl, S., Lemke, S., & Bethge, M. (2025). A meta-synthesis of qualitative research on effective return to work counseling for individuals with work participation restrictions — a systematic review. Journal of Occupational Rehabilitation, 35(4), 725–740. https://doi.org/10.1007/s10926-024-10250-7 [online 8 November 2024]
Backes, J., Mueller, S. I., Geissler, A., & Ehlig, D. (2026). Occupational health interventions’ impact on absenteeism and economic returns: A systematic review and meta-analysis. Scandinavian Journal of Work, Environment & Health, 52(2), 79–97. https://doi.org/10.5271/sjweh.4265 [online 17 December 2025]
Myhrer, M., Brinchmann, B., Trichet, L., Aars, N. A., & Mykletun, A. (2026). Barriers for return to work as an iatrogenic effect of sickness absence: a proposed conceptual framework and questionnaire based on a cross-sectional study. BMC Public Health, 26(1), Article 920. https://doi.org/10.1186/s12889-026-26584-1 [online 12 February 2026]
Øiestad, B. E., Maas, E., Aanesen, F., Tingulstad, A., Rysstad, T., van Tulder, M., Tveter, A. T., Hagen, M., Berg, R. C., Foster, N. E., Wynne-Jones, G., Sowden, G., Bagøien, G., Hagen, R., Storheim, K., & Grotle, M. (2025). Effectiveness of two vocational interventions on sickness absence and costs for people with musculoskeletal disorders: 12 months results from the MI-NAV multi-arm randomized trial. Scandinavian Journal of Work, Environment & Health, 51(6), 505–515. https://doi.org/10.5271/sjweh.4248 [online 3 September 2025]
Popovic, M., Reynolds, L., Noël, C., Cooper, L., & Maranzan, K. A. (2026). Conceptualizing stigma in the injured worker literature: A scoping review. Journal of Occupational Rehabilitation, 36(1), 43–56. https://doi.org/10.1007/s10926-025-10280-9 [online 18 March 2025]
Poulsen, A. G., van Meerkerk, I., Nielsen, C. P., Rolving, N., & Jensen, L. G. (2025). Facilitators and barriers to collaboration between general practice and sickness benefits office in return to work processes — a scoping review. Journal of Occupational Rehabilitation. Advance online publication. https://doi.org/10.1007/s10926-025-10340-0 [online 1 November 2025]
Sanatkar, S., Lipscomb, R., Petrie, K., Collins, D., Arena, A., Xu, M., Counson, I., Dalgaard, V. L., Mykletun, A., LaMontagne, A. D., Harvey, S. B., & Deady, M. (2025). A systematic review and meta-analysis of the effectiveness of work-focused interventions for employees with symptoms of depression, anxiety, and psychological distress. International Archives of Occupational and Environmental Health, 98(9–10), 859–882. https://doi.org/10.1007/s00420-025-02181-4 [online 11 November 2025]
Sikora, A., Seufert, L., Starke, F., Stegmann, R., Wegewitz, U., & Bültmann, U. (2026). Needed and received work accommodations during return to work among employees with common mental disorders and relationships with depressive symptoms and work ability. Journal of Occupational Rehabilitation. Advance online publication. https://doi.org/10.1007/s10926-026-10427-2 [online 13 July 2026]
Join the conversation
Get real-time psychosocial risk strategies, legislative compliance breakdowns, and mature workforce insights delivered straight to your professional feed by Dr. Caroline Howe.