Postural sway does not differentiate individuals with chronic low back pain, single and multisite chronic musculoskeletal pain, or pain-free controls a cross-sectional study of 229 subjects

BACKGROUND CONTEXT: Physical activity in its various forms are the most recommended prevention and treatment strategy for chronic low back pain (CLBP). Standing postural stability is a prerequisite for many types of physical activities. Systematic reviews have investigated the evidence for an association between CLBP and postural stability but results remain inconclusive


Introduction
Chronic low back pain (CLBP) is a worldwide leading cause of disability and the most prevalent chronic musculoskeletal pain syndrome [1]. Chronic musculoskeletal pain syndromes are complex multifactorial biopsychosocial disorders, where it is seldom possible to draw any definite conclusion of causality of single factors leading to persistent and disabling pain [1,2]. According to the clinical presentation, chronic pain can be categorized into single or multisite pain [1,3]. Patients with multisite pain have more comorbid biopsychosocial health issues and poorer long-term prognosis [3].
Standing postural stability is one of the most fundamental human functions and is a prerequisite for locomotion. Physical activity across the domains of daily activities and physical exercises are the most recommended prevention and treatment strategies for CLBP [2,4,5]. Because postural stability is an integral part of physical activity, its association with CLBP and different chronic musculoskeletal pain syndromes have gained a considerable amount of research during the last decades [6,7]. Previous studies have proposed that adaptation to musculoskeletal pain has shortterm benefits in relation to the protection of injury or threat of injury but, in the long-term, may have detrimental due to low-order muscle-specific effects through to higher-order neural and behavioral maladaptations [6,8,9]. The most used postural stability measurement is standing bipedal postural sway on the force plate [10−13]. The force plate measures displacement of the center of pressure (CoP) and records a range of sway parameters. The most studied CoP parameters among CLBP patients are sway velocity and area of sway [14]. Systematic reviews studying postural sway CoP displacement between pain-free controls and subjects with CLBP have shown inconsistent results, but with a trend towards poorer postural stability in subjects with CLBP. Such poorer postural stability becomes especially evident in the more challenging postural conditions (ie, eyes closed and on an unstable surface) [10−13]. Notably, however, the higher the methodological quality of the individual studies, the weaker the observed association in experimental findings [11−13].
The primary objective of this large cross-sectional study was to determine if postural stability differs between pain-free controls and subjects with CLBP with or without leg pain and single and multisite chronic musculoskeletal pain subjects. The secondary objectives were to evaluate the association between postural stability and low back pain intensity and duration, age and gender, physical characteristics (height, weight, and body mass index (BMI), and multiple well-established pain-related factors with patient-reported outcome measures (PROMs) including central sensitization, kinesiophobia, depression, quality of life, disability, and effect of pain on sleep.

Ethics approval and consent to participate
Ethical approval for the study was obtained from the Research Ethics Committee of the Northern Savo Hospital District with identification number 1106/13.02.00/2018. Written informed consent was obtained from all subjects before the study. The study was conducted in accordance with the declaration of Helsinki. STROBE statement for guidelines for reporting cross-sectional studies [15] was adhered to in this study.

Setting and study subjects
Data were collected from a single Chiropractic private practice setting in Helsinki, Finland, between May 2019 and March 2020. A total of 237 subjects agreed to participate and booked clinical appointments through an online booking system. Subjects were recruited using online advertisements and via social media for a range of national Finnish musculoskeletal pain and spine-related organizations and healthcare colleagues. All 18 to 65 years old subjects from the general population meeting the study inclusion and exclusion criteria were invited to participate, whether they suffered pain or not. Inclusion criteria were; (1) age between 18 and 65 years (2) proficiency in written and spoken Finnish. Exclusion criteria were; (1) history of malignant tumor, (2) history of diagnosed trauma negatively affecting the central nervous system, including whiplash or mild traumatic brain injury, (3) history of diagnosed disease negatively affecting the central nervous system, including multiple sclerosis, Parkinson's disease, Alzheimer's disease or other dementia, (4) positive Slump test, (5) acute dizziness symptoms on the interview before postural stability tests, (6) unable to perform movement control and/or postural sway clinical tests in a safe manner indicating possible undiagnosed neurological conditions affecting the central nervous system.

Data collection
Postural stability data were collected using a force plate during the routine clinic visit on Thursdays between 8 AM and 12 AM , at which point the clinical interview and movement control assessments were also performed. A structured interview was done during the clinic visit before the force plate measurement regarding subjects' dizziness status. During the clinic visit, a Slump test was carried out to exclude subjects with symptomatic lumbar disc herniation (30).
None of the study subjects reported acute dizziness during the interview before postural stability tests or had a positive Slump test for acute sciatica thus excluding symptomatic intervertebral disc herniation. Five subjects were excluded due to clinical features suggestive of underlying neurological disorders identified during the prescreening assessment with low back movement control tests described by Luomajoki et al. [16], and/or markedly abnormal postural stability on the force plate. Such tests were interrupted when subjects experienced falls or were deemed to be at high risk of falls and subjects were excluded from the study. Three further subjects were excluded as they did not complete the study questionnaires at the time of their clinical appointments. After the exclusions, the total number of participants in the study was 229 (see Fig. 1 for a study flow chart).
Following the clinic visit, subjects were asked to complete an online web-based data entry detailing demographic information (age, gender, height, weight, and pain history) and PROMs at home. BMI was calculated using subjectreported height and weight data.

Variables
Clinical tests of postural control on the force plate Postural control was measured with a force plate (fourchannel portable computerized force plate (BT4; HUR Labs Oy, Tampere, Finland). Because the subjects completed the questionnaires after the clinic visit, the assessor and the first author (JM) were blinded from the participants' pain history and questionnaire scores. Postural control measurements included velocity and area of sway, which are the most commonly used outcomes for postural control in CLBP cohorts in previous studies [10,14].
The force plate was calibrated before each individual's measurement. Subjects were given the standardized instructions: to stand barefoot, feet as close together as comfortably as possible, to look straight ahead, and keep balancing in a relaxed manner keeping their arms at their sides. If subjects found foot stance as close as possible unnatural, they were instructed to bring their feet slightly further apart to create a more stable and natural feeling standing stance. Minor changes in the stance of feet should not have a considerable effect on the results [17] and results are more reliable when subjects can do slight adaptations to have their natural stance [18]. There was no clear fixation point for gaze, and the opposite wall was more than 3 meters away.
The tests were carried out in the same room with identical conditions for each subject, including distance to the opposite wall and light in the room.
Four different postural control tests were carried out. Each test lasted 60 seconds, which was identical to the time measurement in previous studies on CLBP patients with a similar testing protocol [10]. A 60-second test time is also used on a very similar and well-validated postural control test protocol in a vestibular disorder population [19]. A 5second prephase period occurred before the actual COP measurement of 60 seconds. The sampling frequency was set to 50 Hz, which was recommended by the manufacturer to balance consistent data acquisition and manageable data size. Four CoP displacement quiet stance measurements in respective order were: Eyes open on a stable surface (EOS).

Eyes closed on a stable surface (ECS).
Eyes open on an unstable foam surface (EOU). Eyes closed on an unstable foam surface (ECU).
Two postural sway parameters were studied on each test: (1) Mean velocity of the sway [mm/s] and (2) Area of the sway [mm 2 ]. A rectangular high density (50 kg/m 3 ) closed-cell Airex Balance Pad (delivered by a manufacturer with the force plate) was used for all tests requiring a calibrated foam surface in order to provide an unstable surface. All measurements were carried out once, in similar (nonrandomized) order, and there was no designated resting period between tests other than breaks, when subjects stepped off from the force plate while taking Airex Balance Pad on and off between EOS-ECS and EOU-ECU tests. The force plate protocol was identical to a routine clinical postural sway assessment to ensure direct clinical translation.

Subject-reported pain history questions
All subjects completed a structured web-based pain history assessment with binary questions (yes/no), including presence of chronic low back pain, referral to leg (if yes to CLBP), the experience of other ongoing chronic musculoskeletal pain, and presence of chronic headache. Subjects with CLBP were also asked to rate the severity of pain on a numerical pain rating scale (NPRS) from 0 to 10 and to indicate the duration of the pain as a function of the number of months with pain. CLBP was defined as pain present for more than 3 months and for more than 3 days per week [20,21].

Patient-reported outcome measurements
Numerical pain rating scale (NPRS) measures subjective pain on 11-point numerical scale. The scale is composed of 0 (no pain at all) to 10 (worst imaginable pain) [22].
Central Sensitization Inventory (CSI) was developed to be a screening tool for central Sensitization [23]. CSI is a two-part questionnaire where Part A includes 25 questions about CS-related symptomology. The total score range is "0" to "100" on Likert scale options: 0=never, 1=rarely, 2=sometimes, 3=often, and 4=always. Part B asks with "No/Yes, and year diagnosed" possible previously diagnosed Central Sensitization syndromes and related disorders. CSI part B is only for information and is not scored [24]. CSI has been translated and validated in a Finnish population [25]. Fibromyalgia [26,27], temporomandibular joint disorder [28], migraine, or tension headaches [29,30]  from CSI part B is known to have an effect on postural stability. Therefore, these previous diagnoses on CSI part B were taken into account when comparing subjects with CLBP and pain-free controls.
Tampa scale of Kinesiophobia (TSK) evaluates kinesiophobia (fear of movement). TSK is a 17-item questionnaire used to assess subjective kinesiophobia on Likert scale options: 1=strongly disagree 2=disagree 3=agree 4=strongly agree. The range of scores is from 17 to 68. Higher scores indicate an increasing degree of kinesiophobia [31]. TSK has been translated into Finnish and validated in the Finnish population [32].
Depression Scale (DEPS) is a 10-item questionnaire used to assess depressive symptoms on Likert scale options: 0=not at all, 1=a little, 2=quite a lot and 3=extremely. Higher scores indicate a higher possibility of a Major Depressive Disorder diagnosis [33]. The DEPS has excellent structural validity for screening depressive symptoms among patients with low back pain [34].
EuroQol (EQ-5D-5L) assesses health related quality of life in the five dimensions [35]. The dimensions are mobility, self-care, usual activities, pain/discomfort, and anxiety / depression. Each dimension has five response levels: 0=no problems, 1=slight problems, 2=moderate problems, 3=severe problems, 4=unable to/extreme problems. A second part of the EQ-5D-5L is the EQ visual analogue scale (EQ VAS) [35]. Because there is currently no Finnish standard value set available, a value set from Denmark was used to calculate the index value as recommended by the EuroQol EQ-5D-5L User Guide [36].
Roland-Morris Disability Questionnaire (RMDQ) evaluates disability related to chronic low back pain. RMDQ is a 24-item questionnaire of disability in chronic low back pain populations [37]. The RMDQ is scored by adding up the number of items checked "yes" on different low back painrelated daily activity disabilities. Total scores range from 0 to 24, with higher scores indicating a higher level of disability related to low back pain [38].
Pain and Sleep Questionnaire Three-Item Index (PSQ-3) assesses the impact of pain on sleep. PSQ-3 is a three-item sleep questionnaire designed to measure the impact of chronic pain on sleep during the past week [39]. PSQ-3 measures effect of pain on sleep on a numerical 11-point rating scale from 0 to 10. Zero indicates "never" and 10 indicates "always." Thus, the final score is ranged from 0 to 30. PSQ-3 is translated and validated in Finnish [40].

Statistical methods
Statistical analysis was performed using SPSS version 25 (IBM SPSS Statistics for Windows, Version 25.0. Armonk, NY: IBM Corp.). Statistical significance was defined as p<.05. The normality of the data was checked by Shapiro-Wilks tests. Demographics data were shown as percentages or means with standard deviations or medians with ranges depending on the distribution of the variables.
The minimum required sample size was calculated with average means and estimated standard deviation from the review comparing pain-free controls and subjects with low back pain [10]. The two-tailed hypotheses were calculated on two independent study groups with 0.05 probability of type I error, 0.80 effect size, and 0.8 statistical power. The calculation revealed that at least 25 subjects had to be included in each group.
Normally distributed variables were compared by independent samples t test or One-Way ANOVA with post hoc comparison of Least Significant Difference (LSD). Skewed distributed data were tested by Mann-Whitney U test. Categorical variables were compared by chi-square or Fisher's exact tests. The scatter plots were used to visualize and further investigate some of the key findings of the data.
A multivariable linear regression was calculated to evaluate associations of postural stability with demographics and PROMs by postural stability with eyes closed on unstable surface as dependent variable and gender, age, weight, height, NPRS, pain duration, CSI, TSK, DEPS, EQ-5D-5L, RMDQ, and PSQ-3 as independent variables. BMI was not included in this analysis as a variable because of potential multicollinearity effects between highly correlated variables of BMI and height and weight. Only subjects with CLBP were included in the multiple regression calculation. Sway velocity with eyes closed on unstable surface was chosen as the dependent variable, as sway velocity is considered to be the most accurate parameter of postural sway and previous studies report greatest abnormalities on postural stability in subjects with CLBP on the more challenging test conditions (ie, eyes closed and on unstable surface) [10][11][12][13]42].

Results
We enrolled 229 subjects, which consisted of 162 females and 67 males with mean age of 44.5 §11.8 years, height 171.2 §9.3 cm, weight 75.4 §16.9 kg, and BMI 25.6 §4.8. Subjects were grouped into a pain-free control group and group with CLBP with or without leg pain and single and multisite chronic musculoskeletal pain groups, based upon self-reported pain symptoms. Of the 229 total sample, 42 (18.3%) comprised the control group (reported no pain). Specifically, the pain-free controls reported no CLBP, no leg pain, pain scale 0/10, pain history 0 months, no other chronic musculoskeletal pain, and no chronic headaches. The remaining 187 subjects (81.3%) reported chronic pain, of which 77 (41.2%) reported CLBP without leg pain and 84 (44.9%) with leg pain, 79 (34.5%) reported pain in a single body area (one of the following: CLBP group with or without leg referral, other chronic musculoskeletal pain or chronic headache), and 108 (47.2%) reported multisite chronic pain (two or more of the following: CLBP with or without leg referral, other chronic musculoskeletal pain and/or chronic headache).
Of the total study cohort, only five subjects reported previous fibromyalgia diagnosis (1% in the pain-free control vs. 3% in CLBP groups), 32 temporomandibular joint disorder diagnoses (17% in the pain-free control vs 16% in CLBP groups), and 24 migraine or tension headache diagnoses (4% in the pain-free control vs 15% in CLBP groups).
The pain-free controls and chronic pain subjects were not statistically significantly different in gender, height, weight, or BMI. Demographics between pain-free controls and single and multiple site pain subgroups differed with respect to age. On average, the chronic pain groups were about 4 and 6 years older, showing statistically significant differences between controls vs CLBP with leg pain, single body area pain, and multisite pain groups. There were more males in the single pain group than the multisite pain group.
Pain intensity, duration and all pain-related PROMs measuring disability, central sensitization, kinesiophobia, quality of life and quality of sleep showed clear differences between all five group comparisons (pain-free control vs CLBP without leg pain, pain-free control vs CLBP with leg pain, pain-free control vs single site pain, pain-free control vs multisite pain, single-site pain vs multisite pain). There was a group difference in three of five comparisons in depression. Table 1.
There were only 2 significant differences of 40 comparisons (2/40, 5%) in postural stability parameters. The two differences were between pain-free controls and CLBP with leg pain and between pain-free controls and chronic multisite pain groups on eyes open unstable surface test on an area of sway parameter. Postural stability did not differentiate groups of single and multisite pain on any parameter. Overall, the pain-free control group showed marginally poorer postural stability on the most challenging test than the four chronic pain groups. Table 2.
Postural stability parameters showed only weak correlations with age, height, weight, and BMI. Table 3.
Postural stability parameters correlated very weakly and even weaker than demographics and physical characteristics with CLBP and its intensity and duration and with PROMs of central sensitization, kinesiophobia, depression, quality of life, disability, and effect of pain on sleep. Table 4.
Central sensitization and kinesiophobia, well-established chronic musculoskeletal pain-related factors showed very weak associations between individual scores and performance on the velocity of postural sway tests. Two of four eyes closed tests showed a negative association with central sensitization. Fig. 2.
The results of the multivariable linear regression indicated that the model explained 12.1% of the variance and that the model was a significant predictor of postural stability performance F(3.61), p=.001. Age was the only factor which contributed significantly to the model  Data presented as mean (95% Confidence interval lower and upper bound) or N (%). Comparison between controls without pain and chronic low back pain (CLBP) group without leg pain.
* Between controls and CLBP group with leg pain. y Between controls and chronic single body area pain. z Between controls and multisite pain.
x And between single body area pain and multisite pain. ǁ With statistical significance p<.05.
B=0.297. Table 5. Multivariable linear regression using the velocity of way parameter and other three postural stability tests of EOS, ECS, or EOU as a dependent variable with same independent variables provided essentially very similar results.

Discussion
We show that there are only very small and clinically unimportant differences in postural stability between painfree controls and subjects with CLBP with or without leg pain, nor in subjects with single or multisite chronic musculoskeletal pain. Hence, the results suggest that neither isolated nor widespread chronic musculoskeletal pain are associated with greater postural instability. Only 2 parameters of 40 total parameters showed a statistically significant intergroup difference, thus confirming that our results are robust and consistent.
Very small and clinically unimportant postural stability differences cannot be explained by demographics or physical characteristics, because the only group difference was age. Indeed, aging has been shown to be a strong indicator of poorer postural stability [43]. All four chronic pain groups included average of older subjects and hence should have led to more evident group differences between control groups. Moreover, age was the only significantly contributing factor for postural stability on multivariable linear regression analysis in our cohort.
The primary results are perhaps surprisingly unambiguous given previous findings showing differences in postural stability between subjects with and without CLBP. Systematic reviews in fact support the notion that methodological quality Table 2  Postural stability between pain-free controls and chronic pain groups and between single and multisite pain groups   Variables  Pain-free controls  N=42   CLBP without leg  pain N=77   CLBP with leg  pain N=84   Chronic single body  area  *Comparison between controls without pain and chronic low back pain (CLBP) group without leg pain. y Between controls and CLBP group with leg pain. z Between controls and chronic single body area pain.
x Between controls and multisite pain. ǁ And between single body area pain and multisite pain with statistical significance p<.05.  affects study findings, with studies of greater quality being more likely to find no difference between subjects with CLBP and pain-free controls [11−13]. Indeed, we sought to address many of the methodological limitations of the previous studies including small sample sizes [11,13], lack of detailed reporting of measurement protocol [12,13], lack of adequate inclusion and exclusion criteria and physical testing [10,12], and lack of reporting of subjects' baseline demographics and physical characteristics [13,14]. Also, overlapping other painrelated conditions affecting postural stability did not explain findings, because there were no unforeseen underlying diagnoses in the pain-free control group. Several previous studies have found that differences between subjects with CLBP and controls are more obvious on eyes closed conditions [10−13] due to impaired proprioception in patients with CLBP and therefore being less able to rely on this sense for balance when vision is denied [10,11]. We did not observe any differences between eyes open and eyes closed test conditions between groups, although we did see a normal trend of increased instability on eyes closed and on unstable surface tests within groups. Moreover, the control group showed marginally poorer postural stability than the pain groups on the most challenging test.
Neither pain intensity and duration nor multiple wellestablished chronic pain-related PROMs were related to poorer postural stability, and these findings were consistent across the dataset. Previous studies have shown inconclusive associations between postural stability and pain intensity and duration [12][13][14]44] and with central sensitization, disability, kinesiophobia, and quality of sleep [25,45−48].
In our results, demographic, and physical characteristics showed a higher correlation with postural stability than pain-related variables.
From a clinical perspective, we would argue that bipedal standing postural sway is not a reliable clinical test to understand the effects of pain upon balance function, particularly in the context of CLBP. Given the substantial burden of CLBP on the individual, and society [1,49] we would advise caution in performing unnecessary, time-consuming, and costly high-tech investigations measuring postural sway. Consistent with this, we are aware of a pilot randomized controlled trial evaluating the effect of an exercise training program on postural stability compared to sham treatment in CLBP subjects, where results showed only a small and clinically nonsignificant effect size [50].

Strengths
The major strengths of this study were the large cohort with well-defined study groups, strict inclusion and exclusion criteria including physical examination and interview, detailed clinically applicable test protocol, and use of multiple well-established measures. Moreover, this study was the first study to note clinically important pain distribution differences, which was not studied on previous research comparing controls with subjects with CLBP.

Limitations
We did not include a detailed history of exercise and physical activity in our baseline assessment and therefore cannot comment on whether baseline activity may influence postural stability. Moreover, whilst our physical examination for low back pain included the Slump test, inclusion of advanced imaging of the lower back and clinical tests such as active and passive leg raise and lower extremity sensory and motor clinical testing could have enabled classification of subjects with CLBP into specific etiological categories, which cannot be done with the Slump test alone [51].

Conclusion
We found only very small and clinically unimportant differences in postural stability between pain-free controls and subjects with CLBP with or without leg pain, and subjects with single or multisite chronic musculoskeletal pain. Moreover, postural stability did not reveal an association with leg more widespread distribution of pain or its intensity or duration or indeed correlate with central sensitization, kinesiophobia, depression, quality of life, disability, and effect of pain on sleep. Overall, chronic musculoskeletal pain, especially chronic low back pain, and well-established related factors do not appear to influence static postural stability.

Declaration of Competing Interest
There are no financial conflicts of interest to disclose.