A comparison of trunk control in people with no history, standing-induced, and recurrent low back pain during trunk extension

ABSTRACT

Objectives: This study compares people with recurrent low back pain (rLBP) and people with pre-clinical low back pain (standing-induced low back pain developers; PDs) to each other and back-healthy controls (non-pain developers; NPDs). Movement variability and muscular co-activity related to coordination are important for both rLBP and PDs, and these two groups also have altered static spine extension.

Methods: Eleven participants with recurrent low back pain, and twenty-one asymptomatic participants, categorized as PDs (11) and NPDs (10) through an established standing protocol, volunteered for this study. Three phases of standing extension motion (lean, hold, and return to neutral) were analyzed. Root mean square angular jerk was calculated from trunk and pelvis kinematics, co-activation of the trunk and hip musculature were assessed in four-muscle sets.

Results: Root-mean-square jerk was greater when returning to neutral than when leaning back during standing extension in all three groups. People with rLBP had reduced co-activity in their trunk extensors, people classified as PD had more co-activity in their hip extensors compared with the other groups, and anterior trunk co-activity was phase-dependent, and similar between groups.

Discussion: Movement control alterations with low back pain may start as an over-protective co-activation strategy in those with standing-induced LBP and progress to an under-protective strategy in those with recurrent low back pain. Level of Evidence: 3.

1. Introduction

Poor movement control is a common feature that separates people with nonspecific low back pain (LBP) from their asymptomatic counterparts [1–3]. A joint stability approach suggests some intrinsic inability to control muscle forces can render muscles unable to mitigate injurious loads or produce injurious forces [4,5]. This can lead to overloading of passive structures such as ligaments and joint capsules, and eventually, result in pain or injury of those tissues [6]. Recurrent or chronic LBP can, in turn, alter the brain’s ability to process afferent or generate appropriate efferent information [7], potentially affecting their movement control. In addition to neurophysiological changes, a person’s current LBP state – whether in a painful episode or in remission – can also affect their movement or muscular control [8]. Assessing multiple populations across a spectrum of LBP conditions (asymptomatic, asymptomatic-at-risk, clinical LBP-in-remission, clinical LBP) can help elucidate the progression of movement control strategies as a person develops LBP. For this study, we focus on three groups. The first group are people who report no LBP in their daily lives and no LBP following 2 hours of continuous standing (NPD – non-pain developers). The second group are people who report no LBP in their daily lives, but develop LBP during 2 hours of continuous standing (PD – pain developers) [9,10]. The third group are those who have sought medical treatment for recurrent episodes of LBP (rLBP) and are predicted to respond well to a muscular stabilization rehabilitation program [11].

Despite different LBP histories, the PD and rLBP groups share many characteristics. Both groups experience similar LBP qualitatively [12], but PDs exclusively experience LBP during prolonged standing [9] while the episodes of LBP in those with rLBP occur less predictably [13]. Both those classified as PDs and people with rLBP have shown altered muscular control compared to back-healthy controls. Differences common to the PD and rLBP groups include a reversed trunk muscular timing during extending out of a flexed posture [14,15], and in deviations in the upper lumbar spine (L1 to L3) motion in static trunk extension [16–18]. Both PD and rLBP groups have been shown to benefit from trunk muscular stabilization treatments aimed at correcting specific muscular coordination differences compared to people in back-healthy control groups [10,19]. Additionally, both groups show differences in synergistic muscular co-activity – where co-active muscles generate similar direction moments about a joint [20,21] or antagonistic co-activity (muscles have opposite moment-generating directions) [22,23] depending on the context. In line with the aforementioned neuromuscular differences in PD and rLBP populations relative to controls, these two populations may be the same people at different points in their lives since PDs are up to 3 times more likely than NPDs to seek clinical treatment for LBP later in life [24].

There is evidence to suggest that those in PD and rLBP groups have impaired movement coordination grounded in insufficient muscular control that is theoretically linked to their LBP. Both participant groups also demonstrate differences specific to static standing spine extension [17,18]; however, these two populations have yet to be compared directly to each other or a common control group. If the PD and rLBP groups show similar outcomes in terms of variance in trunk motion through angular jerk [25] or active stiffening through muscular co-activity [26], then it would provide a framework for how a person can move from predictable, non-disabling LBP (PD) to recurrent and potentially disabling episodes of LBP through inappropriate movement control. The purpose of this study was to determine if people classified as PDs or rLBPs have similar movement control of trunk extension. It was hypothesized that both the PD and rLBP groups would have greater muscular co-activity and angular jerk compared to a control group (NPDs) in this task.

2. Methods

An institutional review board at Regis University reviewed and approved this study. Informed consent was obtained prior to all procedures.

2.1. Participants

Participants with and without a history of LBP (pain between the twelfth rib and gluteal fold) were recruited. Participants without prior LBP history were categorized as PD or NPDs based on LBP reporting during 2 hours of continuous standing. To be considered PD, participant’s LBP reports were required to exceed a threshold of 10 mm increase from baseline on a 100 mm visual analog scale during standing. The remaining participants with no LBP history were categorized as NPDs [9,22]. This method of distinguishing NPD and PD groups has been found to be repeatable over a 4-week timescale [27], and useful in predicting future low back pain development risk [24]. In addition to having a history of recurrent LBP, inclusion into this group required that participants were currently in symptom remission, and also met clinical criteria consistent with the stabilization/motor control category of the Treatment-Based Classification system for group homogeneity purposes [11]. A patient history consisting of the duration of their LBP, total number of episodes of LBP, pain-related disability through the Oswestry Disability Questionnaire [28] and movement avoidance tendencies using the Fear-Avoidance Beliefs Questionnaire [29] were also obtained as part of the screening process. Participant characteristics are in Table 1. All participants were between the ages of 18 and 55 years with no prior back or hip surgeries. Participants were excluded if they reported having an adhesive or rubbing alcohol allergy, as were females who had been pregnant within the 12 months preceding data collection. A licensed physical therapist screened all participants to determine their eligibility prior to enrollment.

Table 1.

Participant demographic information. Data are presented as Mean ± 1 SD.

	NPD (n = 11)	PD (n = 10)	rLBP (n = 11)
Age (yrs)*	25.2 ± 5.2	25.6 ± 3.9	35.8 ± 10.9
Height (m)	1.71 ± 0.10	1.75 ± 0.09	1.72 ± 0.06
Mass (kg)	67.4 ± 13.3	72.0 ± 13.7	63.5 ± 7.0
BMI (kg/m2)	22.9 ± 3.0	23.5 ± 2.9	21.8 ± 1.6
% Female	54.5	60.0	63.6
Current Low Back Pain VAS (100 mm)	0.6 ± 0.9	0.0 ± 0.0	2.7 ± 3.1
Duration (yrs)	–	–	7.0 ± 7.8
# of Episodes	–	–	9.1 ± 8.4
ODQ (%)	–	–	13.3 ± 7.0
FABQ (PA + W)	–	–	19.0 ± 6.2

BMI, Body Mass Index; VAS, Visual Analog Scale; ODQ, Oswestry Disability Questionnaire; FABQ (PA + W), Fear-Avoidance Beliefs Questionnaire (the total of the Physical Activity and Work-related components).

*Not equivalent between pain groups (p < 0.05).

2.2. Procedures

Participants completed a three-phase trunk extension motion where they were instructed to bend backward (lean) at a natural pace and look at the ceiling above them for 3 to 5 s (hold) before returning to a neutral standing posture (return) (Figure 1). Throughout this motion, participants had their arms crossed over their chest with each hand placed on the contralateral shoulder. Participants were shown the task by an experimenter and given 2–3 attempts to replicate it, the first attempt with sufficient marker visibility that participants felt was an accurate reproduction of the demonstration was used for analysis. This approach was used to capture a natural motion while minimizing learning effects and being repeatable across participants.

Figure 1.

Standing trunk extension phases. The lean (A to B), hold (B to C), and return phases (C to D) were using the trunk/pelvis extension angular motion; the return phase included overshoot and repositioning.

2.3. Kinematics

Trunk and pelvis positions were tracked using an active marker camera motion capture system (Optotrak 3D Investigator, NDI, Waterloo ON) sampled at 50 Hz. Rigid clusters of three infrared-emitting diodes were secured to participant’s postero-lateral rib cage (T6 level) and pelvis (S2 level) using adjustable Velcro straps. Digital markers defined the locations and orientations of the trunk (bilateral acromion processes and iliac crests) and pelvis segment (bilateral anterior and posterior superior iliac spines) in rigid-link modeling software (Visual 3D v5, C-Motion Inc., Germantown MD). The orientation of the trunk (rib cage) segment relative to the pelvis segment was defined as the trunk/pelvis angle. The trunk/pelvis angle was computed using a flexion/lateral bend/axial twist rotation sequence and a neutral standing posture defined a zero-degree trunk/pelvis angle for all three axes.

Three phases of the standing extension task (lean, static and return) were identified by modifying an algorithm previously used to describe hip and pelvic motion [30]. The initiation of the lean and return phases was defined by the earliest time point where the flexion component of the trunk/pelvis angle had both an angular displacement exceeding 1° and an angular velocity exceeding 5% of the maximum angular velocity in the appropriate direction. The end of the lean and return phases was defined as time points when the angular velocity remained below 5% of the maximum observed value for at least 0.5 s (25 frames). The hold phase was bounded between the end of the lean and the start of the return phase.

Movement control during standing extension was quantified by computing the root-mean-square of the angular jerk (a third derivative of angular position with respect to time) for a time-varying angle $\theta$ (RMS jerk; Equation (1)) in the lean and return phases. A higher RMS jerk indicates greater variance in an angular position [25].

$$RMSJerk=\sqrt{\frac{1}{N}\sum _{n=1}^N{\left(\frac{d^3{\theta }_n}{d{t}^3}\right)}^2}$$ (1)

Trunk/pelvis extension angles were averaged over the hold phase of the standing extension task. RMS jerk for all three components (flexion, lateral bend, and axial twist) of the trunk/pelvis angles were computed during the lean and return phases of the standing extension task.

2.4. Electromyography

Muscle activities were recorded bilaterally from six muscles about the trunk and hip using surface electromyography (EMG): T9 Erector Spinae (TES – lateral to the T9 spinous process), L3 Erector Spinae (LES – lateral to the L3 spinous process), Gluteus Maximus (GMX – halfway between the posterior superior iliac spine and greater trochanter), long head of Biceps Femoris (BCF – 2/3 the distance between the greater trochanter and lateral femoral condyle), External Oblique (EXO – halfway between the inferior rib cage and iliac crest), and Internal Oblique (INO – halfway to the midline between the anterior superior iliac spines). Two disk-shaped Ag/AgCl surface electrodes (Blue Sensor, Medicotest Inc., Olstykke, Denmark) were placed 2 cm apart over muscle sites located through muscle testing and palpation. Sites were dry-shaved and cleaned with isopropyl-alcohol swabs prior to electrode placement.

Double-differential electrical signals (CMRR: 115 dB, impedance 10¹⁰ Ohms) were band-pass filtered from 10 to 1000 Hz and gained from 500 to 2000 to maximize input resolution (AMT-8, Bortec, Calgary AB). Gained signals were sampled at 2000 Hz using a 16-bit conversion card and synchronized to motion capture data (1^stPrinciples, NDI, Waterloo ON).

EMG post-processing was performed using Matlab (v2015a, The Mathworks Inc., Natick MA). Sampled EMG data were bias-removed and rectified prior to feeding through a fourth order dual-pass 2.5 Hz cutoff Butterworth filter to represent muscle activations. Muscle activations had resting activations, defined as the minimum of mean supine and mean prone activity, removed prior to normalization to reference voluntary contractions (RVCs) since it is recommended to avoid maximal voluntary contractions with clinical populations [31]. For the TES and LES RVC, participants extended their head, neck, and chest from prone to a 10-cm shoulder-surface distance [31]. For the GMX and BCF RVC, participants lay prone with their knees flexed to 90° and raised their distal thighs 3 cm off the surface [32]. For the EXO and INO RVC, participants raised both legs 15 cm off the surface from a supine position with their hips flexed to 45°, knees flexed to 90° and feet flat on the surface [31]. RVCs were held for 3 s, with the maximum processed voltages (bias-removed, rectified, filtered and rest-removed) during the held portion being used for normalization.

The commonality dimension of co-activity was used to determine the relative normalized muscle activation among sets of muscles (Equation (2)) [33]. $Com(m)$ represents the proportion of the total area under the curve that is shared by the $m$ muscles within the current set at each point in time. Since EMG data were normalized to an RVC, larger values of $Com(m)$ indicate relative activations that are more similar to the RVC task. Co-activity of the muscle sets in Table 2 was averaged over the three phases identified during standing extension (lean, hold, and return).

$$Com(m)=\mathrm{m}\mathrm{i}\mathrm{n}(m)/\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}(m)$$ (2)

Table 2.

Muscles in each set for $Com(m)$ calculations.

Set name	#	Muscles included
Trunk8	8	BL-TES, BL-LES, BL-EXO, and BL-INO
RTrunk4	4	R-TES, R-LES, R-EXO, and R-INO
LTrunk4	4	L-TES, L-LES, L-EXO, and L-INO
ATrunk4	4	BL-EXO and BL-INO
PTrunk4	4	BL-TES and BL-LES
Hip4	4	BL-GMX and BL-BCF

BL, bilateral; R, right; L, left; Refer to Section 2.4 for muscle abbreviations.

Trunk8 captures the eight muscles measured spanning the trunk; RTrunk4/LTrunk4 captures the right and left halves (four muscles each) of the Trunk8 group; ATrunk4/PTrunk4 capture the anterior and posterior halves of the Trunk8 group; Hip4 refers to the four muscles of the posterior hip measured.

2.5. Statistical analyses

Statistical modeling was performed in RStudio (v.3.3.2, RStudio Inc., Boston MA). Differences in range of motion during trunk extension were compared using a one-way ANOVA using a between factor of pain group (NPD/PD/rLBP). Comparisons of RMS jerk and co-activity between phases of the standing extension task used a within factor of phase (Lean/Return for RMS Jerk, Lean/Hold/Return for co-activity) and a between factor of pain group (NPD/PD/rLBP). RMS jerk had an additional within factor of direction (Flexion/Lateral Bend/Twist). Greenhouse-Geisser corrected probabilities were used for ANOVAs where sphericity assumptions were violated. Significance was set at p $\le$ 0.05, the effect size of all significantly different groups was quantified using Cohen’s d.

3. Results

All participant groups achieved similar ranges of motion during trunk extension (NPD: 21.8 ± 9.7°; PD: 27.9 ± 10.4°; rLBP: 23.1 ± 6.3°; F_2,29 = 1.20; p = 0.317) and were able to perform all study components without experiencing pain or discomfort.

There was a Phase*Direction interaction in RMS jerk (F_1.8,52.8 = 3.68; p = 0.036; Figure 2) where there was more angular variability in the return phase (17.5 ± 8.3°/s³) compared to the lean phase in the flexion axis (12.9 ± 5.8°/s³; p < 0.001, d = 0.646), but not the other two axes (p > 0.281). Participant groups had similar RMS jerk values to each other (F_2,29 = 0.14; p = 0.871; Figure 2).

Figure 2.

RMS jerk during the lean and return phases of standing extension for all three pain groups. Sub-plot headings indicate the trunk/pelvis angle component. An asterisk denotes significant differences (p < 0.05); error bars show one SD.

There were main effects of group on $Com(m)$ within the PTrunk4 (F_2,29 = 6.50; p = 0.005) and Hip4 (F_2,29 = 3.12; p = 0.049) muscle sets (Figure 3). The rLBP group (0.287 ± 0.182 RVC/RVC) had lower PTrunk4 $Com(m)$ than the NPD (0.436 ± 0.202 RVC/RVC; p = 0.042, d = 0.775) and PD groups (0.510 ± 0.138 RVC/RVC; p = 0.001, d = 1.375), while the PD group (0.235 ± 0.164 RVC/RVC) had greater Hip4 $Com(m)$ than the NPD group (0.113 ± 0.147 RVC/RVC; p = 0.050, d = 0.778).

Figure 3.

Commonalities ($Com(m)$) during standing extension. Subplot headings indicate the muscle set from Table 2. Significant phase main effects (p < 0.05) are indicated through letter labels where phases labeled with different letters were significantly different from each other (A is different from B, but both are similar to AB). Muscle sets showing significant main effects of group are labeled with an asterisk (*). Error bars show one SD; 1.0 RVC/RVC indicates that all muscles within a set have the same mean %RVC.

The main effects of phase on $Com(m)$ were observed in all muscle sets except the PTrunk4 set (F_2,29 > 3.23; p < 0.047). The greatest $Com(m)$ was during the hold phase in the ATrunk4 and Hip4 sets (p < 0.034, d > 0.327), while this phase had the lowest $Com(m)$ in the Trunk8, RTrunk4, and LTrunk4 muscle sets (p < 0.001, d > 0.898).

4. Discussion

The purpose of this study was to compare the kinematic variability and muscular co-activity during trunk extension in three groups of people along the spectrum of LBP symptoms. Our hypothesis that the PD and rLBP groups would show similar outcomes of movement coordination variables was rejected in light of findings suggesting two distinct adaptations to standing spine extension in our samples. Synergistic co-activity seen in the PTrunk4 and Hip4 muscle sets could differentiate the NPD, PD, and rLBP groups. People classified as PDs had most common overlap, followed by the NPD group, with the rLBP group having the lowest common muscle activity. These differences were not seen in kinematic variability (RMS jerk) or total range of motion but were observed in all phases of standing spine extension. Both kinematic and muscular measures of movement coordination showed phase-dependent patterns that were common to all three participant groups.

One interpretation of the differences in muscular co-activity is that people unable to provide sufficient co-activity are at a greater risk of low back dysfunction following challenging tasks [3,34]. The RVC used presently to normalize activity for all muscles in the PTrunk4 group (bilateral T9 and L3 erector spinae) was active prone extension, which is defined to have $Com(m)$ of 1. The two main differences between the prone RVC and the standing test task are that standing extension requires postural control for balance and gravity acts to aid standing spine extension rather than oppose it such as in the prone position. These added demands make the standing extension task more challenging than the prone extension RVC and this could explain the differential activation in this four-muscle set. Challenging tasks have previously been shown to elicit greater differences between those with and without a history of LBP [3,35,36], and there is some evidence to support this in the PD group [20]. Models of joint stability often predict that challenging tasks are more likely to generate low back dysfunction [5,37]. The rationale for this prediction is: larger external demands necessitate higher internal loads to prevent excessive tissue strains which can precipitate damage or acute pain [38,39]. In this framework, higher commonality is desired to adapt to the greater external demands of more challenging tasks. While the PD group responded with potentially too much common activity, the rLBP group appears more likely to generate insufficient common activity. The response of people classified as PDs would be labeled as an ‘over-protective’ response, while the response of the rLBP group may be ‘under-protective’ [4,5]. Assuming that the PD group and rLBP group are similar populations at different points in time, a transition from an over-protective to an under-protective response could occur, and the timing of this shift relative to symptom reporting has the potential to explain LBP progression. More direct longitudinal approaches would be the next step in providing evidence for or against this hypothesis.

An alternative to the joint stability interpretation of co-activity is that the reduced common activity in the rLBP group demonstrates greater regional variability in trunk muscular control. Recent work examining this rLBP population supports the theory that localized changes are pivotal in distinguishing movement characteristics from back-healthy controls [17,35]. Specifically, the rLBP group had greater upper lumbar motion (L1–L3) than other LBP groups with more chronic LBP symptoms during active prone spine extension [17]. This perspective would indicate that the PD group would show the least regional variability in trunk motion. Radiographic evidence shows that people classified as PDs may have proportionately greater L1/L2 to L3/L4 intervertebral extension than those classified as NPDs [16]. However, the larger PTrunk4 commonality in the PD group relative to the NPD group would suggest less, not more regional variability. Additionally, regional variability in the trunk movement should translate into higher RMS jerk values [40], which were not observed presently. While it remains unclear how intervertebral coordination translates to muscle activity measured by surface EMG, these two measures of movement coordination both indicate that the upper lumbar levels are important in understanding the mechanics of movement in LBP populations.

There are clinical implications for practitioners who interact with people in the PD or rLBP groups. While both groups experience episodic LBP and benefit from motor coordination training [10,41], they employ different neuromuscular strategies during standing extension which may transfer to other tasks. Inquiring about pain development during prolonged standing might be important information that could direct clinicians toward implementing the best evidence-based approach for their clients. A person in the rLBP group is likely more disabled by their LBP when episodes occur and are more likely to seek treatment for their symptoms. While inclusion into the PD group may not prompt immediate treatment, it is a known risk factor for LBP development [24] and the LBP developed during standing can be mitigated through training [10] or ergonomic interventions [42,43]. Additionally, an active hip abduction test can predict inclusion in the PD group in back-healthy individuals [44]. Therefore, it is beneficial for clinicians to identify if their clients develop LBP during prolonged standing to prevent their pain from becoming less predictable and more disabling.

There were potentially confounding effects based on sample demographics in this study. Although rLBPs were older than PDs, both groups were representative of previously reported samples [18,22,41,45] and were similar in terms of sex distribution, and body mass index. Studies examining age effects in muscle activities during balancing [46] and fatiguing [47] tasks often use 65 years of age to define their older populations, and both groups were younger than that threshold. The low disability (13.3 ± 7.0% ODQ) of those in the rLBP group may mask any differences compared to the other groups studied. However, a higher functioning sample is expected when requiring participants with LBP histories to be in between episodes to limit confounding pain effects on movement [8].

5. Conclusions

We provide evidence that persons at risk for developing LBP coordinate motion differently than persons with a history of LBP, with both groups showing differences compared to an asymptomatic, low-risk group. Differences were localized to the posterior chain of muscles spanning the trunk and hip, with main differences arising in synergistic co-activity. Based on these differences, those only experiencing predictable episodes of LBP tended to use an over-protective co-activity strategy, while those with more severe and frequent episodes of LBP used an under-protective strategy relative to a control population. Early intervention in preventing the under-protective strategy could be beneficial in reducing the progression into recurrent LBP.

Biographies

Daniel Viggiani, MSc, is currently completing his PhD at the University of Waterloo. His thesis work aims to relate prolonged and repetitive exposures to alterations in mechanical pain sensitivity in the low back. His other research interests have touched on muscle fatigue, movement coordination, and prolonged standing; all in regards to their relationship to low back pain

Erika Nelson-Wong, PhD, DPT, is a professor in the School of Physical Therapy and a co-director of the Clinical Biomechanics Lab at Regis University. Her primary research interest is in identification of factors associated with development of low back pain and response to physical therapy interventions, with a focus on movement strategies during functional tasks, particularly during occupational standing.

Bradley S. Davidson, PhD, is an associate professor in mechanical engineering and director of the Human Dynamics Laboratory at the University of Denver. His research emphasizes the measurement of human movement and motor control, with applications in physical medicine and rehabilitation.

Jack P. Callaghan, PhD, CCPE, is a professor at the University of Waterloo and the director of the Centre of Research Expertise for the Prevention of Musculoskeletal Disorders (CRE-MSD). His research program examines biological responses to cumulative loading exposure from both pain generating and tissue altering/injuring perspectives in the low back.

Disclosure statement

No potential conflict of interest was reported by the authors.