Stress prevention at work : intervention effectiveness and implementation process evaluation

Abstract

Background: Work-related stress is a prevalent condition, which is costly for individuals, organizations and society. It is a complex phenomenon that involves and is influenced by factors at many levels. It is consequently only by means of intervening on several levels that preventive work can have a broad impact. There is mixed evidence about the efficacy of organizational-level interventions for stress prevention and more research is needed. Furthermore, process evaluations of trials in this field of research are still scars.

Aims: The overall aim of this thesis was to contribute to the evaluation of organizational level interventions for the primary prevention of stress and its consequences for workers mental health. More specifically, the aim of Study I was to explore whether a participatory, organizational intervention that has the theoretical potential to reduce work-related risk factors for mental ill-health can be effective in preventing stress. Study II was a process evaluation of the intervention implementation in Study I. It aimed to describe the implementation process, examine the influence of contextual factors and explore the participants' experience of working with the method. Study III examined the psychometric properties of the single-item stress question (SISQ) used in Study IV. The aims of Study IV were to describe the trajectories of stress experience in the study population, to examine the association between the subjective stress experience and the objective organizational measures of workload, and to examine the intra-individual variability in stress experience.

Methods: Several designs and methods were used: cross-sectional and longitudinal designs; quantitative and qualitative methods, and multiple data collection methods. Study I was a randomized controlled trial including 57 employees in the intervention group and 61 employees in the comparison group at the baseline. Questionnaire data was collected at the baseline and at 6- and 12-month follow ups. This was the main data collection. Study II had a mixed method design and used cross sectional data from the intervention group: checklists, administrative data, one focus group, three purposefully sampled semi-structured interviews and a process evaluation questionnaire (N=49, 73 %). In Study III, cross-sectional data was used to examine the convergent validity of the SISQ (N=118). To examine the SISQ’ s predictive validity, the stress experience of employees with no sick-leave and no signs of depression or exhaustion at the baseline was examined. Eighty-three employees were included in this analysis. The reliability of the SISQ was analyzed by the test–retest procedure using a separate convenience sample including 108 individuals. Study IV had a longitudinal design. The association between the stress experience and the objective, quantitative workload was examined using administrative data gathered monthly and the data from two time series of weekly administered SISQ. Intra-individual variability was examined by means of standard deviation of a time series and by analyzing the speed of change in stress experience.

Results: No statistically significant differences between the intervention and the control group regarding the primary outcome (job strain), or the secondary outcomes (effort-reward imbalance, exhaustion, sleep and recovery) was found. However, at the 12-month follow up, the perception of effort decreased, and work was experienced as more rewarding in the intervention group by those employees who showed no signs of exhaustion at the baseline. Employees with signs of exhaustion at baseline continued to deteriorate with time regardless off group. The objective data showed that the intervention group had significantly less time per task, more administration per hour worked and more telephone calls answered per hours worked during the entire trial. The reach was satisfying, with high proportion of employees participating in the intervention. Employees had a positive experience of the intervention regarding several of its aspects. However, not all the components of the intervention were introduced in all occupational subgroups. This was mostly due to difficulties obtaining productivity data needed for work with the intervention, and because of staff shortage. The validity of the (SISQ) was supported by its convergence with relevant scales measuring work-related stress and mental ill-health. Furthermore, the SISQ could predict sick leave, exhaustion and depression at 12-month follow up. The stress experience in this sample of primary health care employees was highest in mid-October, mid-November and before Christmas. There was a significant association on a group level, between the quantitative monthly objective measures of workload and the experience of stress. In addition, the fewer the measured tasks were accomplished per hour worked, the higher the experience of stress. The association between the administrative tasks and stress was stronger than between stress and number of patient visits. The analysis of the intra-individual variability showed that the employees scoring high on exhaustion scale have higher rate of change (changing faster) in their stress experience then other employees

Conclusions: This thesis evaluated an organizational intervention in a primary health care context. Statistically significant support for its effects on job strain, effort-reward imbalance, exhaustion, sleep or recovery could not be found. The process evaluation showed that the intervention was not fully implemented, due to several contextual barriers such as staff shortage. No definitive conclusions can therefore be drawn about its effects. However, during the project, several lessons were learned. One of them was that employees already showing signs of exhaustion need special consideration when designing an intervention. Furthermore, these employees showed more rapid changes in their experience of stress from week to week, possibly demonstrating reduced resilience to additional stress. Another lesson learned was that the employees’ experience of stress was associated with objectively measured quantitative workload. Administrative tasks seemed to be more strongly associated with stress than patient related tasks. Managing to complete fewer tasks per hour worked was associated with increased stress. Finally, the single-item stress measure (SISQ), administered by weekly SMS messages, is a valid and reliable measure of experienced stress, in a Swedish population of predominantly female primary health care employees. It can be administered by SMS messages in a work context and be useful in predicting sick-leave and exhaustion one year later.