The Workplace Wellbeing Survey
A 240-bed hospital trust runs a staff wellbeing survey. Of 3,100 staff, 620 respond. Respondents are disproportionately from administrative and managerial roles; clinical staff on wards are under-represented. The headline result is that 68% report high stress. The director of nursing wants to cite the figure in a board paper as evidence that 'two-thirds of staff are highly stressed' and to commission a resilience training programme for all staff at a cost of £45,000. A colleague points out that a previous resilience programme had poor attendance among ward staff.
1. Critically appraise the survey evidence and assess whether it supports the director's claim and proposed action.
Read the worked answer
Response rate and representativeness. 620 of 3,100 is a 20% response rate. That is low, and low response is a recognised source of non-response bias: those who respond tend to differ systematically from those who do not - often those with stronger views, more time, or different working patterns. Here the bias is visible, not hypothetical: administrative and managerial staff are over-represented and ward clinical staff under-represented. Since shift-working ward staff are likely to have different stress experiences and are the group the intervention most needs to reach, the estimate is likely to be biased in an unknown direction.
The claim. 'Two-thirds of staff are highly stressed' overstates the evidence. The accurate statement is that 68% of the 620 respondents - 20% of staff - reported high stress, and that the sample is not representative. The board paper should report the response rate and the occupational breakdown alongside the figure, and should not present it as a trust-wide prevalence estimate.
The proposed action. Commissioning trust-wide resilience training on this evidence is weak. First, the evidence base is inadequate to establish need or target. Second, resilience training locates the problem in individuals, whereas staff stress is substantially driven by organisational factors - staffing levels, shift patterns, workload, and the ability to take breaks. Third, the colleague's observation that ward staff did not attend the previous programme is a strong signal: if the people with the highest need do not attend, the programme will show poor reach and may widen inequity between staff groups. The £45,000 would be better preceded by better diagnosis.
2. Propose a stronger evidence-gathering and intervention strategy, including how you would evaluate it.
Read the worked answer
Better evidence first. Improve the survey before spending: extend the fieldwork window, use shift-aware distribution (paper in staff rooms, links via ward managers, time protected to complete), and offer a short version to boost completion. Report results by staff group, shift pattern and department, with response rates shown for each. Supplement the survey with qualitative work - focus groups or short interviews with ward staff - and with available objective indicators: sickness absence, turnover, vacancy rates, bank and agency use, incident reports and exit interview themes. Triangulating these gives a far stronger case than one survey.
Intervention design. Address the drivers the evidence identifies rather than assuming resilience is the problem. If workload and staffing are the main themes, the intervention should include rostering and break-cover improvements alongside any individual support. If an individual-level component is retained, co-design it with ward staff and deliver it in ways that fit shifts (in-person at handover, short modules, protected time).
Evaluation. Specify outcomes and a comparison in advance. A pragmatic design: identify wards or directorates and introduce the intervention in a phased way, comparing early-implementation and later-implementation groups. Measure reach (proportion of each staff group attending), acceptability, and outcomes such as stress and wellbeing scores, sickness absence and turnover, at baseline, post-intervention and follow-up. Include process evaluation - did staff attend, and why not - because the previous programme's failure was an attendance problem. Pre-register the analysis plan and the success criteria so results are not interpreted after the fact, and report the findings to staff as well as the board.