In the present study we evaluated whether or not a web-based tool, designed for health promotion and stress management, reduces stress and increases physiological markers and psychological ratings of health, recovery and general well-being. At the end of the 6-month intervention period, the intervention group had improved significantly as compared to the reference group on self-ratings of perceived ability to manage stress, sleep quality, mental energy, concentration ability and social support. SRH increased significantly in both groups, with no differences between the groups.
A striking finding is that that ratings of sleep quality improve in the intervention group vs. reference group together with related systematic findings in biological markers and other self-ratings. There is emerging evidence suggesting that sleep alterations can modulate the stress-health relationship. Acute and chronic stressors are associated with subjective and objective measures of sleep disturbances . Thus, improvements in sleep quality might mediate some of the stress protective and health promoting effects found in the intervention group.
Since all participants received some kind of intervention, some beneficial changes were expected in both groups. As a matter of fact, there were several health-related statistically significant improvements for both groups over time (time effect). To mention some, ratings of SRH, eating habits, memory, physical activity, self-esteem and work joy improved as well as levels of cortisol and cholesterol that decreased. However, as the groups did not differ (time × group effect was ns) it is not certain that these effects can be attributed to the web-based tool although they might have resulted from it. To draw such a conclusion a third, passive reference group would have been needed, which was not possible for budget reasons.
The findings of the present study are in line with previous computer-based intervention studies with cognitive exercises that have shown beneficial effects on affective states, such as depression and anxiety [11–13], stress management  and insomnia . Results of the present study are further confirmed in a prospective non-controlled study, in which a web-based intervention was found to decrease ratings of loneliness and depression, whereas perceived social support and self-esteem increased .
In the present study, DHEA-S decreased significantly in the reference group but remained unchanged in the intervention group. DHEA-S is a steroid hormone that has anabolic as well as neuroprotective effects. DHEA-S has also been found to counteract the effects of corticosteroids, such as cortisol, and to be inversely related to both stress and cortisol [32, 33]. Thus, the DHEA-S decrease in the reference group may indeed be a consequence of physiological stress caused by the turbulence that occurred in connection with the study period. This indicates that the intervention program might be protective against stress and facilitate recovery, since DHEA-S remained unaltered in the same stressful period in the intervention group. Furthermore, a number of studies have suggested that DHEA-S can have beneficial effects on cognition, metabolism, wellbeing, and vascular and immune function [32–34]. Considering such prior knowledge, it is of interest that we in the present study found concurrent improvements in DHEA-S and a range of cognitive functions, such as improved concentration ability and increased mental energy in the intervention group.
NPY increased significantly in the intervention group as compared to the reference group. NPY is a hormone that has been reported to have a soothing, anxiolytic as well as antidepressive effect in the central nervous system . The anxiolytic effects of NPY are probably mediated by Y1 receptors in the amygdala and involve inhibition of corticotrophin-releasing hormone (CRH). Moreover, NPY inhibits hypothalamus-pituitary-adrenal (HPA) activity and is thereby effective in reducing secretion of CRH, adrenocorticotropic hormone (ACTH) and cortisol. Finally, NPY has been found to promote and improve sleep . Consequently, the increase in NPY found in the present study may partly explain the beneficial effects, including sleep improvement, found in the intervention group. Some of the findings of the previous literature however, are hard to apply to the present study since they are based upon pharmacological doses of NPY.
Chromogranin A (CgA) decreased in both groups, but significantly more in the intervention group as compared to the reference group. CgA is stored in the core of catecholamine vesicles and is often, but not always co-released with catecholamines. Secretion occurs only during marked activation of the sympathochromaffin system and only stimuli strong enough to induce catecholamine secretion are associated with CgA release. However, CgA also shows ultradian variation, which does not appear to be linked to modifications of catecholamine release [36, 37]. It has been suggested that in situations of mild mental stress CgA is stable and slow to respond . The decrease in CgA in the intervention group might indicate a lesser activation of the HPA-axis and a higher activation in the reference group, perhaps combined with a reduction in activity related to seasonal variation. Cortisol production, for instance, usually declines during the autumn. This might explain why CgA decreased more in the intervention group.
One of the major inflammatory cytokines, TNFα, decreased significantly in the reference group compared to the intervention group. TNFα is one of many markers of the immune system, and the production increases during immunological, inflammatory and stress responses . It has been suggested that cytokines are involved in the regulation of HPA-axis activity . For example, TNFα increases the secretion of CRH (corticotrophin releasing hormone) from the hypothalamus, which in turn results in an increased secretion of ACTH. In turn, ACTH stimulates the release of glucocorticoids from the adrenal cortex. During chronic stress, however, there seems to be a poor relationship between ACTH plasma concentrations and the release of glucocorticoids [39, 41]. Thus, at present there is insufficient information concerning the relative effects of acute and chronic stressors on cytokine activity . It has however been shown that TNFα as well as other hormones, such as cortisol and DHEA-S, exhibit seasonal variation in production. The production of these biological markers seems to be elevated in the spring time and reduced during the autumn [19, 43]. This seasonal variation, i.e. reduction in the autumn, in combination with long-term stress might explain why TNFα decreases in the reference group compared to the intervention group. This possibly stress-related reduction might partly have been counteracted in the intervention group that showed improved self-rated as well as physiological stress management abilities. Consequently, since DHEA-S also remained stable in the intervention group as compared to the reference group that decreased this might be regarded as a systematic finding indicating better stress management and/or decreased stress level in the intervention group as compared to the reference group.
Everything was completely web-based from the start in the present study. It means that the stress management tool was utilized and assessed via the same medium that was used for collecting self-ratings and other relevant background data. This automated, interactive self-help approach differs from previous studies of web-based interventions. Most commonly, other studies have been more similar with face-to-face counseling, where in addition to a website an active counterpart, often a psychologist, issues assignments and evaluations via e-mail. Consequently, the results of the present study might not be completely comparable with other assessments of web-based intervention studies.
The intervention and reference group were treated in the same way concerning blood sampling, advice, web-based questionnaires, etc. The only thing that distinguished the groups was the addition of the interactive cognitive exercises and a chat for the intervention group, which indicate that the complete web-based health promotion and stress management system contributed to the beneficial effects on health, well-being and recovery.
A multitude of items (57) and physiological markers (30) were analyzed in the present study, which makes it relevant to discuss the possible problem of mass-significance. To clarify this issue, percentages of significant findings out of the total number of analyses are presented. Altogether 87 parametric analyses were conducted on relevant VAS items and physiological markers, out of which 11 (13%) significant results were obtained and 9 (10%) remained significant when non-parametric tests were used. Additionally, for several of the biological markers there may be systematic variations in levels during the sampling interval (7.15–11.30 am). For instance, serum cortisol may start decreasing. Most of this variation takes place between morning and evening however. The circadian variation during the morning hours may introduce additional random error in our results. Since there was no systematic difference in sampling hours between the two groups, no systematic bias is likely to have arisen due to this source of error.
The study period of six months might not be enough to cover long-term effects. Accordingly, the beneficial effects found in the intervention group compared to the reference group might be attenuated or continue to improve on a longer term perspective. Therefore, a post intervention follow-up was conducted six months after termination of the study, i.e. twelve months after initiation of the study. The result of this post intervention long-term follow-up will be presented in a future article. However, there are indications that some of the beneficial improvements found in the present study are attenuated after 12 months. Another aspect is that an intervention that focuses solely on individuals might have less ability to produce a lasting effect compared to interventions that also consider organizational aspects. Such multidimensional interventions could perhaps increase the possibilities for the participants to pursue beneficial effects of the individually focused intervention.
Finally, it was mentioned in the introduction that the study was conducted during a high stress period. Therefore, the general health status and occurrence of stress-related problems among the study participants might be discussed. However, apparently the participants were healthy enough to be at work and at baseline there were no participants on sick leave. Furthermore, in the case of participants going on sick-leave they could register these changes in the "profile" section on the website.
There were some weaknesses with this study, e.g. incorrect e-mail addresses to some participants complicated or made communication impossible. Furthermore, we have no exact number of potential participants in the study. This fact might bias the results considering that the sample might not be representative for all the employees. However, in general there was a great interest among the employees of the enrolling departments to participate, as for instance in one of the departments where we have the total number of potential participants, 95% enrolled. Similar participation rates are therefore estimated for the other departments. In any case, based on approximation of the total potential number of employees at each department, enrollment rate was most probably not less than 80% in the worst case scenario.
Implications and future directions
The results of the present study imply some short-term beneficial effects from a web-based tool for stress management. However, initial analyses from a long-term post intervention follow-up indicate a reversion for some of these beneficial effects. Future studies would benefit from pilot testing the web-based tools and thereto related functions to reduce risks of computer-based malfunctioning. Furthermore, logging of usage patterns may contribute with knowledge about how web-based interventions could be improved. More studies assessing psychological as well as physiological effects are needed before more firm conclusions could be drawn.