Any preventive intervention should preferably be based on an understanding of the phenomena underlying the disorder or at least a theory founded on evidence with regard to this issue. In the absence of such understanding, any intervention may target irrelevant issues and would most likely be ineffective. The intervention in this study was based on indications from a former study of the relation of symptoms to upper limb nerve afflictions at three locations (brachial plexus at chord level, posterior interosseous nerve and median nerve at elbow level) . Accordingly, a six month course of stretching was designed with the aim to mobilize the nerve segments at these locations.
The ability of the intervention to reduce symptoms in the studied sample of computer operators is encouraging. The decrease after the intervention of physical findings in the entire sample could not be related to the stretching exercises, however. This is not necessarily contrary to the favourable subjective improvement, but may be explained by statistical weaknesses and other potential sources of error.
First of all, the physical work environment of this small sample of computer operators was already optimized prior to the study. They also had fewer symptoms than reported in other studies on upper limb complaints in computer workers. These favourable circumstances in terms of health would tend to make it more difficult to demonstrate an effect of the intervention.
Secondly, 20 computer workers did not answer the questionnaire at start and 88 did not participate at the follow-up (Figures 1, 2). For the physical examinations the corresponding figures were 67 and 31 subjects, respectively. In fact only 62 subjects in the intervention group and 16 controls participated in all parts of the study. This small number especially of controls is a clear weakness, which together with the high number of limbs without symptoms and findings reduces the statistical power of the study.
The composition of the intervention group and the control group was comparable with respect to gender and age. We did not analyse for prior disorders and psychosocial factors but the two groups were of similar composition with respect to age, sex, and educational and social background. We consider bias due to differences in exposure or vulnerability in the two groups unlikely because of the almost identical content and organization of work, workstation ergonomics and psychosocial work environment. We did not analyse for other covariates such as prior disorders and psychosocial factors but the two groups were comparable with respect to these factors.
Complete blinding of the physical examination could not be achieved. This would demand randomization to the intervention which, however, was deliberately offered to staff in one division of an engineering company with controls in a geographically separated division in order to avoid mutual contacts between intervention and control subjects and thus to prevent the controls to also engage in the stretching exercises if the intervention subjects would perceive them as beneficial. Apart from this, subjectivity was reduced by performing all physical assessments blinded to any other information about the studied computer operators.
Findings were entered into patterns according to predefined algorithms. Still, it cannot be excluded that one finding, e.g. of weakness in a specific muscle, can bias other findings such as sensory deviations because all physical examinations were made by the same examiner. The execution of the examination may have changed slightly from baseline to follow-up. However, the intra-examiner reliability of the applied examination is likely to be good because the inter-examiner reliability has previously been found satisfactory [2, 3]. It may also be argued that a tendency of the study subjects towards more familiarity with the physical examination at the second examination could result in muscle function for example improving over time or nerve trunk soreness being perceived as less uncomfortable than previously. These potential sources of bias cannot be overcome.
A certain fluctuation of findings was noted in the intervention group as well as among controls. However, persistence for some time of physical neurological findings which appear to be characteristic for peripheral nerve afflictions would result in a reduced influence on physical findings of the intervention. Our clinical experiences with assumed neuropathic upper limb conditions suggest that patients reporting fewer symptoms at control visits after successful treatment may still present unchanged levels of pareses and sensory dysfunction. It therefore seems possible that a significant reduction of physical findings would require a follow-up of longer duration.
There is no indication that effective prevention can be accomplished by the isolated use of stretching exercises. However, as a supplement to ergonomic and organizational change stretching may well contribute to reduce the burden of computer-related upper limb disorders.
The outcome of this intervention may serve to further illustrate the character of computer-related upper limb disorder by indicating the relation of incident symptoms to neurological patterns suggesting nerve-afflictions with specific locations. Consequently, it contributes to a further validation of the employed physical examination [2–4].