Electronic measures of movement impairment, repositioning, and posture in people with and without neck pain—a systematic review

Background Neck pain is a major public health problem. Our objective was to describe differences in measures of movement and posture between people with and without neck pain. Methods PubMed and Embase were searched before 15 February 2019 for studies comparing people with neck pain with controls using electronic measurements of neck movement and/or posture. Data were extracted on participants, device, test methods, active range of motion (RoM) and quality of motion, joint positioning sense, and posture. Study quality was assessed using the quality assessment of studies of diagnostic accuracy included in systematic reviews (QUADAS) and Guidelines for Reporting Reliability and Agreement Studies (GRRAS) guidelines. Results Thirty-six studies were included: 24 studies included measurement of active RoM, 15 quality of motion, 12 joint positioning sense, and 5 cervical spine posture. Measurements and test methods were heterogeneous. The reporting of study populations and methods were poor, whereas devices and statistics were well described. All studies on RoM showed reduced active RoM in people with neck pain when compared with controls, 5 of 10 studies reported reduced movement speed for people with neck pain, and 5 of 9 studies reported significantly greater joint positioning error for people with neck pain compared with controls. Due to heterogeneous test parameters and methods, no conclusion regarding differences in conjunct motion, tracking a motion pattern, and measures of posture could be drawn. Conclusions People with neck pain appear to have reduced active RoM, movement speed, and head repositioning accuracy when compared with controls. However, quality of reviewed studies was low and better descriptions of participants and methods are required before firm conclusions can be drawn. Electronic supplementary material The online version of this article (10.1186/s13643-019-1125-2) contains supplementary material, which is available to authorized users.


Background
Neck pain is a common condition with a reported point prevalence of between 0.4 and 41.5% and a lifetime prevalence ranging from 14.2 to 71.0%, depending on its definition [1][2][3]. Neck pain is ranked as the fourth highest contributor to years lived with disability [4]. In Denmark, 6% of all visits to general practitioners and 23% of all visits to chiropractors or physiotherapist are due to neck pain [5].
More than 300 definitions of neck pain have been used in the epidemiological literature [6][7][8][9][10][11][12]. In 2009, the Joint Decade 2000-2010 Task Force on neck pain introduced a conceptual model of neck pain, defining neck pain as pain or discomfort between the superior nuchal line and the spine of the scapula [6]. Treatment approaches such as acupuncture, patient education, multidisciplinary rehabilitation, joint mobilisation, manipulation, and exercise have been shown to be effective treatments, but effect sizes have been small to moderate [7][8][9][10][11][12][13], and there is no clear evidence for any treatment being superior to another. One way to potentially improve the effect of treatment could be to target interventions to specific impairments that clearly discriminate between different types of patients with neck pain or between people with and without neck pain [14]. Existing classification systems build on pain distribution and neurological findings [15], and severity and impact of neck pain [16,17], whereas only one system of targeting treatment has been suggested [18]. However, none of these systems has been rigorously tested for its ability to clearly distinguish between people with and without neck pain.
Exercise treatment is widely used and has the ability to target specific impairments of the neck [19] or limit potential harmful postures such as carrying the head in a forward position [19]. Other parameters such as active range of motion [20][21][22][23][24][25], neck movement speed [26], conjunct motion [27], smoothness of motion [25,28], and kinaesthetic sense [25,[29][30][31][32][33][34][35] have been used to guide how exercises are delivered and performed in individual patients. Assessment of these factors requires that the measurements are reliable and valid in order to correctly guide interventions. Often, these measurements were obtained using electronic devises capable of continuous measurements or movement impairments such as impaired joint position sense. To our knowledge, only one review has addressed movement impairments (joint positioning sense) in people with neck pain [36]. Therefore, there is a need for an overview of the different movement impairments in the neck pain population, measured with electronic devices, in order to provide clinicians and researchers with state-of-the-art knowledge about electronic measurements of neck impairments; the reliability and diagnostic value of these measures, considering the technology and practical application of the movement test; and the type of neck patients.
The overall aim of this systematic review was to determine whether people with neck pain have different movement patterns when compared with people without neck pain. Firstly, we summarised the electronic devices used, the measurement methods, and the definitions of people with and without neck pain. Secondly, we compared electronic measurements of active range of motion, quality of neck motion, joint repositioning accuracy, and posture in people with and without neck pain.

Study design
This study was a systematic literature review based on criteria adapted from Cochrane diagnostic studies [37] and reported according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines [38] (Additional file 1).

Setting
This study was conducted at the Department of Sports Science and Clinical Biomechanics on University of Southern Denmark as part of the welfare tech project 'patient@home'.

Search strategy
We identified relevant studies from the databases PubMed and Embase. The search strategy was tailored with the assistance of an experienced research librarian. We limited the search to include only publications in English or Danish published between 1 January 2004 and 15 February 2019. The reference lists of all included papers were closely scrutinised for eligible studies. For the full search strategy, see Additional file 2.

Inclusion criteria
We included cross-sectional studies where the case population was judged to have non-specific neck pain, whiplash-associated disorder (WAD), cervical radiculopathy, or acute, sub-acute, or chronic neck pain of any duration. Furthermore, the study had to report at least one electronic measure of a movement impairment, joint position sense, or posture. The inclusion and exclusion criteria are specified in Table 1.

Selection of studies
The results from the literature searches were imported into EndNote©, and duplicates were removed. Three authors (BBH, HR, and PK) were involved in the screening process. Three authors (BBH and HR/PK) independently screened the titles and abstracts for relevance. If it was not possible to decide from title and abstract, a full-text screening was performed. In case of disagreement, a third author (PK/JH) was consulted.

Data extraction
BBH, PK, and JH did the data extraction independently. This included information about study population, testing circumstances, and test device. The results from the impairment measures were extracted, including active range of motion, movement speed, acceleration, jerk, head repositioning accuracy, and posture.

Quality assessment
The quality of the included studies was assessed using a purposeful tailoring of the quality assessment of studies of diagnostic accuracy included in systematic reviews (QUADAS) and Guidelines for Reporting Reliability and Agreement Studies (GRRAS) guidelines [39,40] where we remodelled the element relating to a reference standard and questions about case-control design. We piloted the modified tool using articles that we had excluded from the review.
Quality assessment was done in two sets where one half of the included studies were evaluated by BBH and PK and the second half by BBH and JH. Disagreements between the authors were discussed and consensus sought, and continued disagreements were then resolved by JH for the first half and PK for the second half.

Data reporting and analysis
Agreement in selecting studies and rating risk of bias were determined using Cohen's kappa [41]. The results, including risk of bias, were grouped by type of measurement and reported for subgroups of different types of neck pain. For comparable measures and homogeneous studies, the results were presented as forest plots. For measures where studies were too heterogeneous for that, data were narratively summarised.

Description of included studies
The literature search was conducted on 5 November 2014, and updated on 19 September 2017 and again on 15 February 2019. We identified 3348 unique studies after excluding 652 duplicates. A total of 90 papers were retrieved in full text of which 53 were excluded (see Additional file 3 for a list of excluded studies). An overview of the selection process and reasons for exclusion are shown in Fig. 1. We identified a total of 37 papers reporting on 36 studies [21,27,29,30,34,. These studies had case populations ranging between 7 and 120 participants with an average of 33 participants, and control populations ranging between 11 and 150 with an average of 35 participants. In 27 studies [22, 28, 30, 31, 35, 43, 45-51, 54-60, 62-64, 66, 69-71, 74], patients with neck pain were described as chronic or with pain duration of more than 3 months. In six studies [43,51,52,59,60], there was no description of pain duration. Definition of chronicity varied from no specification [29], duration of pain longer than 6 weeks [44], to neck pain lasting more than 2 years [30].

Quality of included studies
The initial inter-rater reliability of the risk of bias assessment between BBH and PK had an agreement of 76.4% and a kappa score of 0.59. Between BBH and JH, the agreement was 60.6% with a kappa score of 0.38. The Most of the studies had insufficient description of sample size, study population, characteristics of raters, and blinding of raters for clinical information and previous findings. In contrast, the descriptions of measurement devices, tests, recording methods, and statistical analyses were generally adequately reported. Detailed results of the risk of bias assessments are presented in Table 2.

Range of motion measures
Active range of motion was reported in 24 studies [21, 27, 30, 42, 45, 46, 48-55, 57, 59-61, 64, 67, 69] (Table 3). There were three different ways of reporting range of motion: half cycle range of motion, which is the range from neutral starting position to end position in a given direction; full cycle range of motion, which is the range from endpoint in one direction to the opposite endpoint; and a division between upper and lower cervical range of motion in       Across all 24 studies, people with neck pain had a smaller range of motion when compared with healthy controls.

Posture
In five studies [43,47,56,58], measures of posture were assessed. The characteristics of the studies are described in Table 6. In three of these studies [43,58], a working task of typing/computer work was assessed. In another one of these studies [47], the postural task of habitual sitting posture and perceived 'good' posture were evaluated, and in the remaining study [57], habitual standing posture was measured. 'Sagittal plane angle of head tilt' was the only parameter that was reported in all five studies. In two studies [44,74], no difference between the measured angles in the different groups was found, and in the three other studies, between one [57] and two angles [47,58] differed between the groups. However, none of the differences were consistent across the studies.

Discussion
Regardless of definition, people with varying types of neck pain have reduced active range of motion, reduced movement speed, and impaired head repositioning accuracy when compared with people without neck pain. However, due to lack of consistency in measurement parameters and variation in the postural tasks examined, it was not possible to quantify differences between people with and without neck pain for several of the included measures. We found substantial heterogeneity in the included studies regarding types of patients, types of measurements, and types of technology, and many studies had poor reporting, which resulted in high risk of bias.
Consequently, results of this review must be interpreted with caution. Firstly, study populations were poorly described, i.e. in 15 studies, the age and sex distribution were uneven across groups; secondly, the description of the neck pain groups was heterogeneous with eight different definitions of non-specific neck pain, and only five studies adequately reported the power calculation for their sample size. Furthermore, we found a general lack of description of the examiners' background and training, which may influence patient handling and application of measurement devices as most of the measurement devices are dependent on the examiners' ability to palpate landmarks on the subject, which is a challenge even for experienced clinicians [70,71]. Also, blinding of assessors is mostly not reported in the articles, which may be a concern because body language and communication generally may be affected if the assessors have knowledge of clinical information and previous test results. Lastly, the variation in test methods and measurement parameters was large, making it unfeasible to do meta-analyses. This heterogeneity in test condition is most likely contributing to the large degree of variation in the measured values for people without neck pain. For example, cervical flexion  [29] and 65.3° [50] and extension between 28 [29] and 79.4° [51].
The complexity of delivering measurements of active range of motion has been the subject of several systematic reviews [36,72,73]. Williams et al. [72] concluded that the simple non-electronic measurement devices were more reliable when measuring cervical range of motion when compared with more sophisticated electronic devices, whereas Micheils et al. [73] found that electronic devices were more reliable and valid in assessing motion patterns (the fly) and that neck pain patients had a greater degree of error when compared with people without neck pain. We found conflicting results with three studies having a greater degree of error and one with a lesser degree of error for the neck pain patients. Lastly, de Vries et al. [36] reviewed the literature dealing with joint positioning sense in people with neck pain and people without neck pain and concluded that joint positioning error was greater for people with neck pain, which corresponds with our findings, although they also included non-electronic measurement devices.
To our knowledge, this is the first systematic review combining studies dealing with measurements of movement in people with different types of neck pain and different types of movement and postural impairments. We adhered to the criteria adapted from the Cochrane diagnostic studies handbook and reported our work according to the PRISMA guidelines. We searched literature in PubMed and Embase databases