Upper extremity musculoskeletal disorders (MSDs) have been linked to many features of video display unit (VDU) operation, including duration and intensity of use.¹ Neck pain has been less studied than the shoulder and the distal hand-wrist regions; the study by Wahlström and colleagues, published in the June 2004 issue of this journal,² is a new and noteworthy exception.

Perhaps the lower level of attention to date is because neck pain is less serious and has less potential to cause disability. How long lasting and important are these symptoms? Do people recover readily, or do they progress to having more frequent and chronic pain? Do the disorders interfere with work or have other social or economic costs? And—especially important for occupational health and safety practitioners and clinicians—do the answers to these questions depend on whether the individuals are continuing to work in the same conditions under which the symptoms first developed?

A few investigations have addressed these questions. For example, among employed persons with neck pain, the proportion with lost work time is similar to those with back pain, which is notorious for high absenteeism costs.^3,4 Manufacturing workers with neck pain lost, on average, about 14 days from work in one year because of their neck problems. Among nurses, more frequent and severe neck pain was associated with a higher probability of work absenteeism as well as having to modify or restrict work activities.⁵ Other outcomes included inadequate sleep, reduced participation in non-work activities and recreation, seeking more medical attention, and greater use of pain medications.

Among people with neck pain severe enough to seek medical care, pain severity predicted decreased mental wellbeing as well as limitations in ability to perform activities of daily living and related functional capacities.⁶ The longer the episode had lasted, the worse the expected outcome; a substantial minority were still suffering one year after seeking medical attention.

While healthcare costs are not as high for neck as for low back pain, they may be substantial in patients with neuropathies affecting the neck.⁷ Non-surgical treatment of chronic radicular pain is of limited efficacy.⁸ These data, although very limited, should nevertheless motivate us to seek effective primary and secondary interventions for preventing work related neck pain and for limiting the duration and progression of those episodes that do occur. There is a need for more prospective data on predictive factors of adverse clinical and employment outcomes. It is to be hoped that populations such as the one described in the current study will be followed up in the future for new information on the natural history of neck disorders and how they can most effectively be addressed in the workplace.

Despite the large literature on work related MSDs, one theme in the recurring debate centres on the possible confounding of physical exposures by psychosocial strain in that literature. While these two types of exposures are often correlated with each other,⁹ and while each has physiological effects with biological plausibility for MSD risk, nevertheless “confounding” may not be the correct way to frame the issue. Independent variables that share a common causal pathway cannot confound each other, although there may be interaction if, for example, the effect of one is conditional on the other’s presence or absence. In addition, if the two variables are measured in such a way that they overlap operationally (if not conceptually), then any attempt to separate their effects may lead to spurious conclusions.

This has important practical implications for the way that MSDs are studied. Given their acknowledged multifactorial aetiology, we need to employ multivariable analytical methods that are consistent with our hypothesised biological models. Standard regression approaches, with main effect terms only, imply control of confounding but do not examine effect modification or other causal pathway structures. At a minimum, stratified analyses or interaction terms are required to examine potential effect modification, and more complex modelling approaches are often justified.

One important domain of psychosocial strain, assessed through any of the commonly used instruments, is job “demands”, usually representing a combination of time pressure, actual work pace, physical effort, and/or competing demands from multiple supervisors or job responsibilities. Conversely, any operational measure of physical work pace is likely to imply a psychological experience of time pressure. Job “control”, another important psychosocial feature, generally has a strong inverse relation with the performance of physically repetitive work. Clearly these measurement constructs have both psychological and physical dimensions. Under these conditions, the attempt to partition the risk of MSDs between physical and psychosocial exposures is inherently flawed. The dilemma is especially pronounced in the case of VDU operators because physical force levels are generally low, leaving work pace and repetitive hand and finger motions as predominant physical stressors.

Again, the best approach to resolving this problem is to define clearly the causal pathway in advance of undertaking any data analysis. Identifying factors at different stages along that pathway can provide valuable information about whether or not different exposures have common mechanisms of effect, provided that the statistical analysis is carried out so as to illuminate rather than obscure those mechanisms. In the current study, “physical exposure” was assessed in terms of precision and repetition of manual activity, while “psychosocial strain” was defined as the combination of high demands and low control. The same questionnaire obtained “perceived muscular tension” (PMT)—a likely intermediate variable for both exposures and possibly also a reflection of non-occupational stressors. PMT was treated in the analyses as a risk factor only, even though the authors called attention to the need to determine its position on the pathway(s) of effect. Since we cannot determine that position from the analyses presented, the cause(s) of the tension remain obscure. However, it is evident that if PMT represents the effects of either occupational exposure (physical and/or psychosocial), then those effects are underestimated in the analyses that are adjusted for PMT. Complicating the matter even further, if either physical or psychosocial exposures were assessed incompletely or with error, PMT might be serving here as a proxy for that unmeasured load or strain. Thus, despite the other strengths of this investigation, the results provide less information than we might wish regarding which interventions would decrease that muscle tension and the resulting neck pain.