Losing Sleep Over Work? Does It Matter?

THE GENERAL PUBLIC VIEWS THE ASSOCIATION OF VARIOUS LIFE STRESSORS WITH POOR SLEEP AS A GIVEN FACT, AND RESEARCHERS HAVE DONE THEIR best to prove the apparently obvious. The majority of the working population reports that work has a significant impact upon their sleep,¹ with more than 75% of respondents to a media survey stating they have lost sleep over work issues. Individuals with insomnia frequently attribute their sleep problems to work-related stress or job dissatisfaction.² One of the more stressful experiences at work must be that of being bullied—the media, medicolegal offices, and the Internet are full of cases and anecdotes, often driven by unions, litigants, and whistleblowers with good reason. In this issue of SLEEP, Niedhammer and her colleagues³ in France add more evidence as to the effects of bullying.

Their cross-sectional survey of nearly 8000 employees recruited from a nonrandom network of occupational physicians in southeast France shows that current bullying is associated with a 4- to 5-fold increase in reported sleep disturbance. Frequency and duration of bullying, and whether or not the respondent also observed the bullying of others, had sex-specific effects upon this association, which are reported (but not shown) to be little affected by adjusting demographic, occupation, working hours, and adverse physical exposures.³ Adjusting for the potential confounding of self-reported health and depression dramatically reduced these associations, although they were still statistically significant, and, for many associations, the odds ratios were in the range of 1.6 to 2.4. Of interest is that the association between experiencing bullying and sleep disturbance disappears after these adjustments among those who didn't report seeing others bullied. This may suggest that the effect is stronger in workplaces where such behaviour is seen as being more widespread, as is often described.⁴ This may implicate broader measures, such as perceived organizational injustice, and organizational downsizing, which is receiving more attention, especially in the current economic climate.

There are self-acknowledged limitations in the report by Neidhammer and colleagues,³ including caution about generalizing from the 40% response rate, interpreting the direction of causality, and the lack of detail arising from the requirements of large surveys and understanding that measuring sleep was never the focus of the study. One point that is sidestepped is the use of a dichotomous measure for adjustment of depression. By doing so, the authors may have lost important information that may have reduced the observed association even further. Additionally, the effect of bullying upon anxiety has been reported as stronger than on depression,⁵ but anxiety was not measured. The issue of statistical overadjustment or underadjustment is complex. Do you treat sleep disturbance as independent from depression and anxiety (as in this study) or, at least partly, as attributable to these conditions and, thus, subject to overadjustment?

Depending upon the definition, the prevalence of bullying, or “mobbing” as it is known in some countries, ranges from 5% to over 50%.⁴ A strength of this paper³ is the triangulated determination of bullying, which gives a figure of around 10% and thus presumably defines a group who have a more severe exposure than in many other studies. The assessment of duration and frequency of bullying is welcome, as previous work has shown not only that the majority of bullying is transient,^6,7 but that chronic minor stressors, such as financial problems,⁸ appear to have an effect upon the development of more chronic sleep disturbance, which, in turn, has been suggested to have a greater impact on other health outcomes. In this context, the replication of chronic bullying being shown to have a stronger effect on sleep disturbance adds to this knowledge.

Addressing study design and definitions in such an area is vital. This is particularly so in areas in which the “exposure” may be influenced by “outcome.” In the area of bullying, it has been shown that certain people are more likely to perceive and later report bullying.^6,7 These include those people with less assertive neurotic personalities, the obese, and those with depression, anxiety, and chronic physical conditions^4,7—all conditions that have been associated with sleep disturbance. To our knowledge, there exist only 2 cohort studies^6,9 evaluating the risk of bullying on later psychological and physical health problems, both showing a negative effect.

There exists a wealth of data evaluating the effect of workplace stressors upon later health outcomes. In Niedhammer et al³ the discussion alludes to (but results are not presented) the association of bullying and sleep disturbance being consistent after adjusting for more widely recognized workplace stressors such as demands, decision latitude, and social support. Recent systematic reviews^10,11 have concurred that there are consistent effects of these upon downstream psychological health, although inadequate study methodology, particularly in adjusting for confounding, and obvious publication biases have been identified.

It is not only psychological illness that results from workplace stressors. Chronic, but not acute, job strain has been reported to predict recurrent cardiovascular events.¹² Identifying the pathways by which psychosocial stressors impact health is a fundamental task in risk modification.¹³ The biologic mechanisms linking work stressors and cardiovascular events include proinflammatory markers and other psycho-neuro-immunologic mechanisms.¹⁴ Given the effects of sleep disturbance upon these markers,¹⁵ evaluating the role of sleep disturbance as a potential mediator of this pathway seems warranted. This leads to a requirement to establish adequately powered cohort studies, with reliable and valid determination of workplace stressors such as job strain, bullying, and injustice over a period of time; equally valid measurement of sleep disturbance rather than reliance upon 1 or 2 questions; baseline and ongoing evaluation of the many potential confounders identified in the latest systematic review¹¹ (demographics, life events and demands outside of work, personality and previous psychiatric disorder, physical health risk factors and disease, and family history); and appropriate multiple-response analysis techniques such as generalized estimating equations. Although costly and difficult logistically, given the amount of money spent by governments and organizations on tackling workplace stressors such as bullying, this seems not out of place.

DISCLOSURE STATEMENT

Drs. Glozier and Grunstein have indicated no financial conflicts of interest.