Summary of evidence
The heterogeneous study designs and experimental setups did not allow pooling of data or any direct comparison of results across the studies. In addition, the poor overall documentation of the experimental setups, particularly with regards to participant demographics and technical aspects such as sampling frequency and cut-off frequency, impaired full interpretation. However, despite the great variability there was enough consistency in results to show that patients suffering from NSNP and WAD exhibit an increased COP sway compared to healthy individuals, especially in AP direction. Unfortunately, the magnitude of these differences in postural sway cannot be summarily expressed in terms of specific percentages or values. As a result, only a general trend is noted.
As we outlined in a previous systematic review , the reliability of COP measurements is primarily determined by the six main factors (Table 4). Although only two of the included studies fulfilled more than half of the recommended reliability criteria [18, 23], it is worth bearing in mind that studies considering less than all six criteria may still present fairly reliable results.
While a general trend towards decreased postural stability can be observed irrespective of the origin of the pain, the altered sway pattern appears to correlate with the associated degree of proprioceptive impairment. This is signified by the generally greater COP excursions in WAD cases [21–23] where damage to proprioceptive structures and neck musculature due to the sustained trauma may be expected. In addition, higher pain intensities or the underlying neurological or vestibular impairments observed in several studies [21, 25] may be the determining factor in the reported highly significant differences in sway pattern compared to healthy controls. The lack of comparable data does not allow the interpretation of previous pain duration or associated perceived disability in this context. While some WAD patients may have also been included in NSNP studies, it appears unlikely that this affected the overall results.
We have decided to include studies using induced pain in our review. While this cannot be considered similar to (chronic) NSNP, it may nevertheless mimic many alterations in sensorimotor functions documented in acute clinical pain conditions, although it should be noted that it does not replicate any potential long term neurological adaptation. Both experiments resulted in significantly altered sway pattern which may underline the role of acute "pain inhibition"  in the observed postural response. However, the COP sway area measured was nevertheless smaller than reported in people with WAD  which may underline the likely role of proprioceptive impairment associated with the pain in the development of COP excursions of larger magnitude.
Visual deprivation caused an increase in postural sway in numerous studies of healthy participants [29–32] and has shown to be a major challenge to the balance systems in studies investigating the effect of non-specific low back pain on postural stability [29, 33, 34]. Nevertheless, statistically significant differences were not found in a number of NSNP studies (Table 2). In addition to issues arising from the experimental setups and the generally small sample sizes of seven  to thirty  symptomatic participants, the variations in the perceived pain intensities may offer an explanation.
Pain severity has shown to be a determining factor in non-specific low back pain cases  where a significant, linear increase in postural sway was observed beginning at a NRS-11 score of 5. If this can be applied to NSNP patients as well, low pain intensities at the time of recording such as those reported by Field et al.  may well explain the fact that no significant differences could be identified, while patients suffering from more severe pain exhibited significantly increased postural sway compared to healthy controls .
If rather small differences in COP measures between the groups can be anticipated, the choice of appropriate sway parameters is important. However, only Vuillerme et al.  and Endo et al.  used mean velocity (mVel), a parameter that has shown both consistently high reliability  and discriminative value in pain conditions . Despite a small sample size and low scores for the reliability of the experimental setup, they found highly significant differences with eyes open  and under both visual conditions .
The effect of ageing can be observed when comparing the studies by Field et al.  and Poole et al. . Although the methodologies are very similar, varying results were reported. This may be explained by the fact that the latter enrolled elderly patients (65-82 years compared to 27-30 years). Older individuals exhibit increased COP excursions  and any pre-existing deficits in proprioception associated with ageing may add to the alterations caused by the neck pain.
Overall, the lack of data available, no conclusions can be drawn regarding a possible relationship between postural stability and perceived pain or disability levels. For the same reason, no conclusion about the effect of impairments in cervical ROM is possible.
At this point, there are several important limitations to the application of COP measures in the assessment of postural sway in a clinical setting:
Although the results tempt us to hypothesize a correlation between the magnitude of COP excursions and the extent of damage to proprioceptive structures, the causative factor for the altered postural sway pattern remains largely unclear in people with WAD and NSNP. The question still remains whether the increased COP excursions are predominantly related to the previously described physiological changes due to chronic pain perception, acute or chronic damage to proprioceptive structures in the neck or acute "pain inhibition" . If the latter mechanism is mainly responsible or if the proprioceptive impairment is of acute and reversible nature, monitoring neck pain patients during their treatment and rehabilitation process may aid as an objective tool in assessing the patient's progress. If long-term neuro-physiological changes are primarily involved, individually varying recovery time frames may render such measurements less useful.
Finally, the data available is insufficient to determine whether some form of correlation between the neck pain intensity, its duration or the perceived disability and the magnitude of postural sway exists. As a linear relationship between pain intensity and COP sway velocity has been demonstrated in patients with non-specific low back pain , further research is necessary to investigate whether this also applies to people with neck pain. If this can be established COP may have a clinical role as an instrument of measurement for neck pain patients.