A chief executive in Vancouver has set aside a budget for employee mental health this year and wants to know where to spend it. Three proposals are on the desk: an app, a two-day workshop, and an annual consulting contract. All three vendors have brought numbers and charts. The published evidence says something rather different.
How effective are workplace mental health programmes?
Workplace mental health programmes reduce burnout reliably only when they change how work is organised, not merely how employees feel about it. A 2025 systematic review examined 14 studies, comprising 9 randomised controlled trials and 5 quasi-experimental studies, with 3,572 participants in total, covering occupational burnout across 6 healthcare, 5 corporate and 3 public sector settings. Participatory organisational interventions sustained reduced burnout for 12 months or longer. Digital tools showed short-term benefit but attrition reached 42 percent. Brief workshops had no sustained effect beyond 3 months.
That pattern is the single most useful fact a budget holder can carry into a planning meeting. The interventions that lasted were the ones that altered workload distribution, decision latitude, role clarity or team-level process. The interventions that faded were the ones delivered to individuals and disconnected from the conditions that produced the exhaustion in the first place.
The review carries an explicit limitation that matters for interpretation: moderate-to-high heterogeneity, seven different burnout scales, and effects varying by implementation quality. A programme that works in one hospital unit may not transfer intact to a trading floor or a construction site. Effectiveness is therefore not a property of the programme label. It is a property of fit between the intervention, the setting and the quality of delivery.
Why does the evidence point in two opposite directions?
The evidence points in two opposite directions because individual-level and organisational-level interventions are measured against different baselines and produce different magnitudes. A separate 2024 meta-analysis of 33 studies, with 2,536 participants, compared these two approaches in resident physicians: 25 individual studies and 8 organisational studies. Individual intervention versus control produced an emotional exhaustion Cohen’s d of -0.25, with a 95 percent confidence interval of -0.40 to -0.11, and a depersonalisation Cohen’s d of -0.17, with a 95 percent confidence interval of -0.32 to -0.03.
Organisational interventions in that same meta-analysis showed no significant association in any domain. This is the opposite of what the 2025 review found, and the contradiction is instructive rather than embarrassing. The 2024 meta-analysis covered resident physicians only, the data were inconsistent, risk of bias was high, and the authors concluded the effects ranged from none to small. Small individual effects can be real and still insufficient to move a workforce.
A multi-level cluster randomised controlled trial has been designed across 8 European countries and Australia in construction, healthcare and information technology. This is a published protocol only and has no results yet. Until it reports, the field is arguing from partial evidence, and any vendor claiming certainty is ahead of the science.
Before buying any programme, one number is needed: over the past six months, how many people left and how many took sick leave? — Image: AI-generated / Gemini 3.1 Flash via OpenRouter
What happens to one-day workshops and wellbeing apps?
One-day workshops and wellbeing apps typically produce measurable short-term change and no durable reduction in burnout. In the 2025 review, brief workshops had no sustained effect beyond 3 months, while digital tools showed short-term benefit but attrition reached 42 percent. Nearly half of the people given access to a digital tool stopped using it. A tool nobody opens cannot lower emotional exhaustion.
“When a company calls me, someone has usually already broken. And what they ask for is almost always a workshop. My first question is this: over the past six months, how many people from that same team resigned, and how many took sick leave? A leader who does not have that number is buying in the dark.”
The market for organisational mental health consulting is full of guaranteed off-the-shelf packages while the evidence itself still points in two opposite directions, so the only thing a leader can do with confidence is measure before and after each intervention with one single consistent scale, rather than trusting a vendor’s promise.
Burnout does not stop at five o’clock; the relationship pays part of the bill. — Image: AI-generated / Gemini 3.1 Flash via OpenRouter
Where should a leader start?
A leader should start by measuring the baseline before spending, then fund participatory organisational change rather than individual content delivery. The 2025 review found that participatory organisational interventions sustained reduced burnout for 12 months or longer, which is the only durability finding in the brief that survives a full year. That is where the marginal dollar has the strongest published support.
[Needs further verification: the evidence base is drawn mainly from clinical and Western settings, and it is not yet clear that the same pattern holds in Iranian companies or in immigrant family businesses, where obligation, family reputation and informal hierarchy may shape how burnout is expressed and how organisational change is received.]
The hopeful part is that the mechanism is not mysterious. Workload, control, clarity and fairness are measurable, and they are the levers that held their effect for a year or more. A leader who measures honestly, funds the organisational layer first, and refuses to buy a promise without a before-and-after number will spend less and change more. That is not a slogan. It is the only conclusion the current evidence actually supports.
Tiredness resolves with rest. Burnout has three layers: emotional depletion, cold detachment from the work and the people, and a drop in the sense of competence. The second and third layers do not resolve with a holiday.
Are wellbeing apps useful for employees?
In the 2025 review, digital tools showed short-term benefit but attrition reached 42 percent. The tool is not the problem; on its own, without any change to how the work is structured, it does not last.
Why does a one-day workshop fail?
In that same review, brief workshops had no sustained effect beyond 3 months. A workshop transfers knowledge, but it does not change workload or decision latitude — and burnout comes from those two.
With a limited budget, what is worth doing first?
Measurement. Before any purchase, record the resignation and sick-leave rate for that team and run one burnout scale once. Without a baseline number, no intervention can be judged.
Conclusion
Two bodies of evidence point in two directions. The 2025 systematic review of 14 studies and 3,572 participants finds that participatory organisational interventions hold their effect for 12 months or longer. The 2024 meta-analysis of 33 studies and 2,536 resident physicians finds that only individual interventions reached significance, and the effect was small. That contradiction should not be hidden; it should be read as evidence that no off-the-shelf package is guaranteed.
What a leader can do this week is concrete: put the past six months of resignations and sick leave for that team on paper, run one burnout scale once, and sign nothing until that baseline number exists. Measurement is not expensive. Not having it is.
Looking for professional support with your relationship or mental health?
1. Bagasi A, Al Harbi EK, Alabbasi SM, Alqaedi RO, Alharbi BA, Alhomaid TA. Effectiveness of Workplace Mental Health Programs in Reducing Occupational Burnout: A Systematic Review. Cureus. 2025;17(7):e88715. PMID 40861700. DOI: 10.7759/cureus.88715
2. Kiratipaisarl W, Surawattanasakul V, Sirikul W. Individual and organizational interventions to reduce burnout in resident physicians: a systematic review and meta-analysis. BMC Medical Education. 2024;24(1):1234. PMID 39478552. DOI: 10.1186/s12909-024-06195-3
3. Arensman E, et al. Implementation and evaluation of a multi-level mental health promotion intervention for the workplace (MENTUPP): study protocol for a cluster randomised controlled trial. Trials. 2023;24(1):621. PMID 37777787. DOI: 10.1186/s13063-023-07537-0 — this reference is a study protocol only and has published no results; the article says so explicitly.
Sources retrieved from PubMed.
Image Credits
All three images in this article are AI-generated and none depicts a real person — Images: AI-generated / Gemini 3.1 Flash via OpenRouter.
