new insight into motor adaptation to pain revealed by a combination of modelling and empirical...

ORIGINAL ARTICLE

New insight into motor adaptation to pain revealed by acombination of modelling and empirical approachesP.W. Hodges1, M.W. Coppieters1, D. MacDonald1, J. Cholewicki1,2

1 The University of Queensland, Centre of Clinical Research Excellence in Spinal Pain, Injury and Health, School of Health and Rehabilitation Sciences,

Brisbane, Qld, Australia

2 Center for Orthopedic Research, Michigan State University, Lansing, MI, USA

CorrespondencePaul W Hodges

E-mail: [email protected]

Funding sourceNational Health and Medical Research Council

(Australia).

Conflicts of interestNone declared.

Accepted for publication21 December 2012

doi:10.1002/j.1532-2149.2013.00286.x

Abstract

Background: Movement changes in pain. Unlike the somewhatstereotypical response of limb muscles to pain, trunk muscle responses arehighly variable when challenged by pain in that region. This has led manyto question the existence of a common underlying theory to explain theadaptation. Here, we tested the hypotheses that (1) adaptation in muscleactivation in acute pain leads to enhanced spine stability, despite variationin the pattern of muscle activation changes; and (2) individuals would usea similar ‘signature’ pattern for tasks with different mechanical demands.Methods: In 17 healthy individuals, electromyography recordings weremade from a broad array of anterior and posterior trunk muscles whileparticipants moved slowly between trunk flexion and extension with andwithout experimentally induced back pain. Hypotheses were tested byestimating spine stability (Stability Index) with an electromyography-driven spine model and analysis of individual and overall (net) adapta-tions in muscle activation.Results: The Stability Index (P < 0.017) and net muscle activity(P < 0.021) increased during pain, although no two individuals used thesame pattern of adaptation in muscle activity. For most, the adaptation wassimilar between movement directions despite opposite movementdemands.Conclusions: These data provide the first empirical confirmation that, inmost individuals, acute back pain leads to increased spinal stability andthat the pattern of muscle activity is not stereotypical, but instead involvesan individual-specific response to pain. This adaptation is likely to provideshort-term benefit to enhance spinal protection, but could have long-termconsequences for spinal health.

1. Introduction

Trunk muscle control changes in back pain. Thisincludes opposite changes of decreased (Hodges andRichardson, 1996; Leinonen et al., 2001; MacDonaldet al., 2009) and augmented (Radebold et al., 2000;Hodges et al., 2003b) muscle activity in acute (Hodgeset al., 2003b) and chronic pain (Radebold et al., 2000)and during symptom remission (Hodges and Richard-son, 1996; MacDonald et al., 2009) without consis-

tency in motor adaptation between muscles, patientgroups or experimental methods (van Dieen et al.,2003). This underpins confusion and debate, particu-larly when motor changes are used to design exerciseinterventions for back pain (O’Sullivan, 2000; McGill,2002; Richardson et al., 2004). Current treatmentstarget opposite goals to enhance (McGill, 2002) ordecrease (Richardson et al., 2004) activity of specificmuscles. Although variable responses in back painimply absence of a consistent outcome of the adapta-

1Eur J Pain •• (2013) ••–•• © 2013 European Federation of International Association for the Study of Pain Chapters

tion (van Dieen et al., 2003), the alternative interpre-tation is that variability reflects the trunk musclesystem’s redundancy (i.e. multiple muscle activationstrategies to achieve similar goals) (Latash and Anson,2006). Unfortunately, with few exceptions (Radeboldet al., 2000), most studies investigate few muscles,which limits consideration of overall responses topain.

Widely accepted theories of motor adaptation topain predict stereotypical responses. The ‘pain adapta-tion’ theory predicts decreased activity of muscles pro-ducing painful movements and facilitation ofantagonists (Lund et al., 1991). The ‘vicious cycle’theory predicts increased activity, with further paininduced by accumulation of metabolites from muscleischaemia (Roland, 1986). These stereotypical adapta-tions are inconsistent with variable trunk musclechanges in back pain (van Dieen et al., 2003). Recenttheoretical work proposes the adaptation of motorcontrol in acute pain involves redistribution of activitywithin and between muscles with the goal of protect-ing the body segment from real or perceived threat ofpain/injury in a manner specific to the individual/context (Hodges and Tucker, 2011). Yet despite varia-tion in trunk muscle responses, the overall outcomewill be enhanced spine stability (Hodges and Tucker,2011), defined as the potential to resist perturbationscausing intervertebral displacements (Reeves et al.,2007). This adaptation could be task-specific or indi-viduals may use a similar ‘signature’ adaptation acrossa range of tasks. Evidence of increased spine stabilityusing mathematical simulation of a small number ofpatterns of increased muscle activity thought to becommon in back pain provides some evidence for thishypothesis (van Dieën et al., 2003). However, empiri-cal evidence from muscle recordings is required to

take into account the complexity of potential changesin muscle activity, which may include increased anddecreased muscle activity.

We tested the hypothesis that motor adaptation inacute pain enhances spine stability despite variation inthe adaptation’s pattern (i.e. which muscles areaffected, and whether activity increases or decreases).We also tested the hypothesis that individuals woulduse a similar ‘signature’ pattern in different tasks.Hypotheses were tested by estimating spine stabilityusing an electromyography (EMG)-driven biome-chanical model and analysis of empirical data fromEMG recordings from a broad array of trunk musclesin response to experimentally induced back pain.

2. Methods

2.1 Participants

Seventeen males [mean (standard deviation) age – 25(6)years, height – 177(8) cm, weight – 75(12) kg] participatedin the study. Participants were excluded if they had anyhistory of low back pain that had limited function or causedthem to seek medical or allied health treatment. The Insti-tutional Medical Research Ethics Committee approved thestudy and all procedures were conducted in accordance withthe Declaration of Helsinki. Participants provided writteninformed consent.

2.2 Electromyography

EMG activity was recorded bilaterally from three trunkflexor muscles [rectus abdominis (RA), obliquus internus(OI) and externus abdominis (OE)] and three trunk extensormuscles [thoracic (TES) and lumbar erector spinae (LES) andlatissimus dorsi (LD)] using pairs of self-adhesive electrodes(3M, St. Paul, MN, USA) at sites described in detail elsewhere(Cholewicki et al., 1997) and shown in Fig. 1A and B. Theseare the only trunk muscles accessible for surface EMGrecording and provide good representation of overall trunkmuscle activity and the resultant spinal loads (Cholewickiet al., 1997). Skin was prepared with mild abrasion andwiped with alcohol. EMG data were pre-amplified 1000times close to the body (Neurolog, Digitimer, WelwynGarden City, Hertfordshire, UK), amplified further for twotimes, filtered between 20 and 1000 Hz (Neurolog) andsampled at 2000 Hz using a Power1401 (Cambridge Elec-tronic Designs, Cambridge, UK).

2.3 Procedure

Participants sat in frame on a slanted seat that was designedto position the lumbar spine in a mid-range lordosis(Fig. 1C). The pelvis was fixed with padded supports placedin front and behind the participant. A harness was placed

What’s already known about this topic?• Movement adapts in pain, but variability in the

response to pain has led some to suggest thatthere is no common mechanism or outcome ofthe adaptation.

What does this study add?• This study shows that spine stability is systemati-

cally increased when low back pain is inducedexperimentally.

• But, this is achieved by patterns of muscle acti-vation that vary between individuals.

• The adaptation may have short-term benefit, butwith potential long-term consequences.

New insight into motor adaptation to pain P.W. Hodges et al.

2 Eur J Pain •• (2013) ••–•• © 2013 European Federation of International Association for the Study of Pain Chapters

over the shoulders with a tilt sensor attached at the approxi-mate level of T9 to measure the angular displacement of thetrunk. This data was used to align EMG activity with respectto trunk angle rather than a specific time. The task involvedslow trunk movement between approximately -20 degrees(extension) and +20 degrees (flexion) for two repetitions ineach direction. The range of motion was monitored on lineand feedback provided to the participant if required. Therationale for this task is that although the activity of thetrunk flexor and extensor muscles is required to maintainthe extended and flexed posture against gravity, respectively,no resultant moment acts on the trunk in the mid-uprightposition and any recorded trunk muscle activity is thatrequired to maintain stability (control upright trunk positionif perturbed). This position is characterized by the lowest netmuscle activity (Cholewicki et al., 1997). The quasi-staticanalysis was enabled by movement at a slow speed that wascontrolled with feedback from a metronome. An auditorysignal was provided each 1 s and participants were encour-aged to take 7 s to complete the movement in each direction.Trials were performed before and during pain induced byinjection of hypertonic saline (5% concentration; 1.5 mLbolus injected over ~20 s) into the right longissimus muscle~5 cm lateral to the spinous process of L4 (Fig. 1D). Pain wasreported on an 11-point numerical rating scale (NRS)anchored with ‘no pain’ and ‘worst pain imaginable’. In 15

participants, data were also recorded for two trials after thepain had resolved when participants reported 0 on the NRS.

Maximal voluntary contractions (MVC) against manualresistance were performed for each trunk muscle for normal-ization of the EMG recordings for calculation of the spineStability Index (SI; see below). The tasks were: TES and LES– trunk extension in prone with resistance to the thighs andupper trunk; LD – shoulder adduction and extension againstmanual resistance at the elbow with the participant in sittingand the arms abducted and externally rotated 90 degrees; RA– trunk flexion in supine with resistance to the thighs andupper trunk; OE and OI – trunk rotation in supine with theknees bent and resistance applied to the bent knees and thearms (in 90 degrees shoulder flexion and full elbow exten-sion). Contractions were maintained for ~3 s and the largestamplitude over three repetitions recorded for analysis.

2.4 Data analysis

Data analysis was conducted using Matlab (The MathWorks,Inc., Natick, MA, USA). The QRS complexes of the electro-cardiogram (heart beat) were removed from the raw EMGusing a modified turning point filter and adaptive sampling(Aminian et al., 1988). Next, EMG data were rectified andlow pass filtered at 1 Hz (dual pass, fourth-order Butter-worth). The net muscle activity was quantified by calculatingthe root mean square (RMS) of the 12 EMG signals at eachtrunk angle. The minimum net muscle activity (minimumRMS EMG) was identified and the trunk angle at which thisoccurred was recorded (referred to as the neutral position).Data for this analysis were not normalized to MVC, as theywere used to estimate the net overall muscle activity and toidentify the trunk position associated with minimum netmuscle activity. However, this analysis did not take intoaccount the relative contribution of each muscle to spinestability because it ignored differences in moment arm andmuscle cross-sectional area. These factors were taken intoaccount in calculation of the spine SI.

For calculation of the spine SI, EMG data in the neutralposition were used in a spinal stability model that has beendescribed in detail elsewhere (Cholewicki and McGill, 1996).Briefly, MVC-normalized EMG amplitudes recorded from 12muscles were used to estimate muscle force and stiffness for90 muscle fascicles represented in the model. The SI quan-tifies the average curvature of the system’s potential energyin the vicinity of static equilibrium. This potential energy isthe difference between work performed by external forcesacting on the spine and elastic energy stored in muscles,whose stiffness is proportional to the muscle force. The indexis a function of the rotational stiffness at each of the 18degrees of freedom in the model (six lumbar intervertebraljoints and three rotations at each) and provides a compara-tive measure of structural robustness of the spine to externaldisturbances that would cause intervertebral displacements(Howarth et al., 2004). The quasi-static character of the tasks(resultant trunk velocity was approximately 6°/s) justifiedthe use of a static, structural buckling analysis in this study

Figure 1 Experimental set up. Surface electromyography electrode

placement for recording of (A) trunk flexor muscles [rectus abdominis

(RA), obliquus externus (OE) and internus (OI) abdominis] and (B) exten-

sor muscles [latissimus dorsi (LD), thoracic (TES) and lumbar (LES)

erector spine]. (C) Participant positioned in the frame with pelvic

restraint and tilt sensor to record trunk angle relative to gravity. (D)

Placement of the needle for injection of hypertonic saline in the experi-

mental pain trials.

P.W. Hodges et al. New insight into motor adaptation to pain


and this approach was validated previously (Cholewickiet al., 1997). Data were averaged over two repetitions.

2.5 Statistical analysis

The spine SI and minimum net muscle activity were com-pared between Pain conditions (no pain vs. pain) andbetween Directions of movement (front-back vs. back-front)with repeated measures analyses of variance (ANOVA). Posthoc testing was undertaken with Duncan’s multiple rangetest. Differences in EMG for individual muscles (for datafrom all participants considered together) were investigatedby comparison between trials with and without Pain,between Directions and between Muscles with an ANOVA. Inaddition, data were displayed pictorially to evaluate thechanges in activity of each muscle for each individual par-ticipant. Thus, changes in activity for each muscle were pre-sented in black if activity increased, grey if the activitydecreased, and white if the activity did not change. For thisanalysis, data were defined as increased or decreased if theychanged by > 15% (Chapman et al., 2008) from valuesrecorded in the trials without pain. The proportion of indi-vidual participants who had an increase or decrease in activ-ity for each individual muscle was recorded, as were thenumber of participants who had the same or opposite changein muscle activity between directions of movement. Datawere presented descriptively for these analyses. Data arepresented as mean (standard deviation) throughout the text.Significance was set at P < 0.05.

3. Results

3.1 Pain

Participants reported peak pain of 6.1(2.7) out of 10on the NRS after injection of hypertonic saline intolongissimus. All movement trials were completedbefore pain fell below 4 out of 10 on the NRS.

3.2 Minimum net muscle activity

The mean (range) amplitude of EMG across muscles atbaseline was 1.9 (0.2–7.8) % MVC. Fig. 2 shows datafor a representative participant. The heavy dashed linethat indicates the net muscle activity (RMS of all 12trunk muscles’ EMG) clearly demonstrates an increasein the minimum net muscle activity during pain. Withslow movement from front-back, the minimum netmuscle activity increased by 30(42)% during pain, andby 14(34)% when moving from back-front (Interac-tion: Pain*Direction – P = 0.021, Post hoc: front-backP < 0.001 and back-front P = 0.027, Fig. 3A). Theminimum net muscle activity increased during pain in76% (13 out of 17) and 65% (11 out of 17) of partici-pants for the front-back and back-front movements,

respectively. In three participants, the minimum netmuscle activity decreased in both movement direc-tions. Analysis of individual muscles showed that evenin those participants with decreased minimum netmuscle activity, the activity of some muscles wasincreased (see below).

An additional analysis was conducted for the 15participants who had data available after the resolu-tion of pain. In those participants, the minimum netmuscle activity after pain was not different to thatbefore pain (Post hoc: front-back – P = 0.710, back-front – P = 0.740, Fig. 3).

3.3 Spine SI

The spine SI was calculated at the trunk angle atwhich the minimum net muscle activity was identified(neutral position). Concomitant with the minimumnet muscle activity, the spine SI was increased duringpain and this was similar for both directions of trunkmotion (Main effect: Pain – P = 0.017, Interaction:Pain*Direction – P = 0.15; Fig. 3B). The SI wasincreased in the back-front direction in 82.4% (14 outof 17) of participants, and 70.6% (12 out of 17) ofparticipants in the front-back direction. There was nodifference in the spine SI between the two directionsof motion (Main effect: Direction – P = 0.240). For the15 participants who had data available after the reso-lution of pain, the SI after pain was not different tothat before pain (Post hoc: P = 0.83, Fig. 3).

Figure 2 Representative electromyography (EMG) activity as a function

of trunk angle during slow movement from front to back. Trunk extensor

(grey line) and flexor (black line) muscle EMG is shown. The root mean

square (RMS) EMG activity of all 12 trunk muscles (net muscle activity) is

shown as a thick dashed line and its minimum is indicated at angle zero

with the vertical dashed line. Note the increase in the minimum net

muscle activity during pain.



3.4 EMG amplitude for individual muscles atthe angle of minimum net muscle activity

When data were analysed for the entire participantgroup together, TES EMG activity on the right side ofthe body was increased in the neutral position (Inter-action: Pain*Muscle – P = 0.009, Post hoc: P < 0.001;Fig. 4) and, although not significant, there was a ten-dency for increased TES activity on the left side (Posthoc – P = 0.064). Left OE EMG was decreased (Post hoc:P = 0.042). These changes were not dependent on thedirection of trunk movement (Interaction: Direction*Pain – P = 0.100, Interaction: Direction*Pain*Muscle –P = 0.280). There was no difference for any othermuscle (Post hoc: P-value range P = 0.120 to P = 0.900).

Analysis of the EMG data for individual musclesfrom participants as a group provides limited under-standing of the changes in muscle activity with pain inindividual participants. Fig. 5 provides a visualsummary of the complex and variable pattern ofchange in trunk muscle activity during pain. Panel A is

organized with pairs representing the data for indi-vidual participants during the two movement direc-tions (top – front-back, bottom – back-front). Panel Bshows the proportion of participants in which thesame response was identified in the front-back andback-front directions, i.e. the consistency of thechange in muscle activation between task directions.The change in EMG amplitude for individual muscleswas opposite between trunk movement directions inonly 10.7% of muscles across all participants. Thus,the adaptation was almost identical for the two con-ditions, despite the opposite direction of motion.

There was considerable variation in the pattern ofchange in trunk muscle activity between participantsand no two participants showed an identical pattern ofchange. Across participants and directions, thenumber of muscles in which EMG increased by 15%or more ranged between 2 and 10 (Fig. 5A). In thefront-back direction, left OI EMG was increased in65% of participants; left RA, LD and TES in 59%; rightRA, OE, OI, LES and TES in 53%, and right LD, left OEand left LES in 36–47% (Fig. 5C). With the back-frontdirection, EMG increased by 15% or more in 59% ofparticipants for right OE, left OI and right TES andLES; 53% for left LD and right OI; 47% for left andright RA; 17–35% for left TES and LES, left OE andright LD (Fig. 5D). Activity was most commonlydecreased by 15% or more for muscles that rotate thetrunk to the right (OE on the left side of the body:front-back – 53%; back-front – 58%; right OI: bothdirections – 35%) and the left LES (29–35%). Allother muscles were decreased by more than 15% inless than 29% of participants.

Figure 3 Mean (standard deviation) of (A) minimum net muscle activity

(minimum root mean square electromyography [EMG] of all 12 trunk

muscles) and (B) Stability index (SI) before and during experimental pain.

Minimum net EMG activity and SI are also shown for the 15 participants

with measures made after the resolution of pain. Note the increase in both

parameters during pain. *P < 0.05.

Figure 4 Mean electromyography (EMG) activity for each muscle at the

trunk angle identified to have the minimum net muscle activity (minimum

root mean square EMG of all 12 trunk muscles) for front to back (left) and

back to front trunk movements. Standard deviation is shown. RA, rectus

abdominis; OE, obliquus externus abdominis; OI, obliquus internus abdo-

minis; LD, latissimus dorsi; TES, thoracic erector spine; LES, lumbar

erector spine; r, right; l, left. *P < 0.05.



4. Discussion

Although experimental back pain induced variablepatterns of increased and decreased trunk muscleactivity, when all muscles were considered together,the net trunk muscle activity and estimated spine sta-bility increased. This finding concurs with the hypoth-esis that the nervous system responds to pain byincreasing muscle activity to protect the spine, validat-ing a key prediction of a contemporary theory ofmotor adaptation to pain (Hodges and Tucker, 2011).

4.1 Methodological considerations

The present results require consideration of severallimitations. First, analysis based on net trunk muscleEMG activity is limited because it does not recognizethe muscles’ different cross-sectional areas, momentarms and effects on spine stability. Therefore, we alsocomputed the SI, which accounts for these variables.Congruence between findings from both methods aidsinterpretation of EMG data from individual muscles.Further, restriction of hip motion limits potential forhip muscle adaptation. Although, this improves accu-racy of our estimates of spine stability, investigationincluding the hip would provide additional insight ofadaptation strategies available in a real-world context.

Second, the biomechanical model used to estimatethe SI is static and can only provide analysis of thespine at a single point in time. However, the spine is a

dynamic system with feedback control. Therefore, theslow movement task was quasi-static to limit dynamicaspects of trunk muscle responses.

Third, adaptation to pain was present during a taskrequiring low muscle forces. Although relevant toeveryday activity such as standing or sitting without aback support, further work should determine whetheradaptation is similar in higher effort tasks.

4.2 Spine stability increased despite variableresponses of trunk muscles to pain

Theoretical predictions of motor adaptation to pain arevariable. Decreased activity of muscles producing apainful movement (agonist) and increased antagonistactivity predicted by the ‘pain adaptation’ theory toreduce movement amplitude/velocity was not sup-ported by our data; there was neither consistentlydecreased trunk flexor and increased extensor activityduring back-front movement, nor the converse withopposite movement. Although this prediction is sup-ported in simple systems with few muscles [e.g. elbow(Ervilha et al., 2004), jaw (Svensson et al., 1995; Sohnet al., 2000) or ankle (Graven-Nielsen et al., 1997)],the complexity of the redundant trunk muscle systemappears not associated with stereotypical change.

Decreased activity in some muscles is not explainedby ‘vicious cycle’ theory predictions of increased activ-ity. Other theories propose some back pain is associ-ated with spine instability (Panjabi, 1992). Although

Figure 5 Individual data for change in elec-

tromyography (EMG) activity of each muscle

during pain. (A) Data are shown for each par-

ticipant (S1–17) in the front to back (F-B;

upper) and back to front (B-F; lower) directions

of trunk movement. Black indicates increased

(15% or more) EMG activity, grey indicates

decreased (15% or more) activity and white

indicates a change of less than 15% from the

pre-pain condition. (B) The proportion of par-

ticipants who had the same change in EMG

activity for both tasks. (C) The proportion of

participants with increased, no change and

decreased EMG for each muscle during front to

back movement. (D) The proportion of partici-

pants with increased, no change and

decreased EMG for each muscle during back to

front movement. RA, rectus abdominis; OE,

obliquus externus abdominis; OI, obliquus

internus abdominis; LD, latissimus dorsi; TES,

thoracic erector spine; LES, lumbar erector

spine; r, right; l, left.



possible when passive spine support is disrupted (e.g.trauma), this was not present in this experimentwhere pain induced increased spinal stability withoutchange in passive stiffness.

The observed individual-specific, non-stereotypicalchanges in muscle activity parallel the predictions of acontemporary theory of adaptation to pain, whichstates redistribution of activity between musclesincreases net muscle activity to protect the painful part(Hodges and Tucker, 2011). Several earlier experimen-tal observations provided foundation for this predic-tion. First, a variable pattern of increased/decreasedmuscle activity was reported for a limited number oftrunk muscles with experimental back pain (Hodgeset al., 2003b), but the net effect on spine stability wasnot assessed. Others proposed a net increase in trunkmuscle activity based on literature review (van Dieenet al., 2003) and biomechanical model simulation ofthree stereotypical patterns of increased muscle activityincreased stability (van Dieën et al., 2003). Althoughpromising, that study did not account for diversity ofadaptation identified here and did not includeco-existent decreased activity of some muscles whichcould counteract the effect of increased activation.Finally, recent work showed increased trunk stiffness inresponse to perturbation during remission from recur-ring back pain (Hodges et al., 2009). Although provid-ing some validation of the interpretation of the presentdata, that method could not distinguish between activeand passive contributions to spine stiffness.

Although some argue adaptation to augment spinestiffness would be necessary to compensate forreduced support from passive structures (i.e. injury)(Panjabi, 1992; van Dieën et al., 2003), this cannotexplain our data as pain was induced without injury.However, adaptation of large superficial muscles maycompensate for a decreased contribution of the deepermuscles to spine stiffness. Deeper trunk muscles, suchas transversus abdominis, contribute to spine stability(Hodges et al., 2003a, 2004; Barker et al., 2005) butare consistently compromised when clinical (Hodgesand Richardson, 1996) or experimental back pain(Hodges et al., 2003b) is present. Activity of deepmuscles was not monitored here.

Experience of pain, independent of tissue injury,was sufficient to trigger a response that enhancedspine stability. This parallels earlier observations ofincreased/decreased muscle activity in experimentalmuscle pain (Hodges et al., 2003b). A similar responsecan be evoked by threat of pain, in the absence ofnociceptor stimulation (Moseley et al., 2004). Thenervous system appears to take protective action inthe presence of both a real or predicted threat to body

tissues. This has implications for patients with persis-tent back pain as adapted responses may be main-tained, despite tissue healing. Although this studycannot resolve whether the adaptation was caused bynociceptive stimulation or the threat of pain, this doesnot detract from the key observations.

Is the adaptation to increase spine stability/protection helpful? One interpretation is the adapta-tion prevents further pain and/or injury. However, thisbenefit may be limited to the short term, with poten-tial for negative long-term consequences caused by:sustained increase in spine load from the net increasein muscle activity (Kumar, 1990; Marras et al., 2004),impaired spine movement and its contribution toshock absorption/dampening (Mok et al., 2007) orreduced movement variability which compromisesload sharing between spinal structures (e.g. muscles,joint surfaces, ligaments) which is linked to pain inother regions (Hamill et al., 1999). Here, the adaptedmotor response resolved once pain recovered, butmany individuals with recurring pain maintain abnor-mal muscle activation during symptom remission(Hodges and Richardson, 1996; MacDonald et al.,2009). Whether, maintenance of the adaptationunderlies persistence/recurrence of pain requiresinvestigation in longitudinal studies.

The SI did not increase in three participants in thefront-back direction and five participants in the oppo-site direction. Instead, the SI decreased by 12(8) and8(5)%, respectively. Why this subgroup had the oppo-site response is unclear, but most of these participants(two out of three and three out of five, respectively)had an SI in the highest five of all participants prior topain induction. Their SI may have already exceededthat required to complete the task and a modifiedpattern of muscle activity to achieve stability duringpain may have been more important than an absoluteincrease in SI. It is also possible that spine stabilityincreased during pain, but using muscles other thanthose recorded (e.g. quadratus lumborum, psoasmajor).

4.3 The pattern of change in muscle activityvaried between individuals

Inconsistent with earlier theories (Roland, 1986; Lundet al., 1991), the nervous system did not changemuscle activity stereotypically during pain, but theoutcome of adaptation was predictable. In the redun-dant trunk muscle system, many solutions are avail-able to increase stability. There are several reasonswhy an individual may select a specific pattern. First,the strategy may be a learnt behaviour based on



habitual postures or patterns of movement [e.g. sittingwith extension from lumbar to mid-thoracic levelsfavours TES activity (Claus et al., 2009; Astfalck et al.,2010) and people who naturally sit in this mannermay adopt a strategy involving this muscle]. Second,anthropometry (relative length of segments) mayinfluence moment arms and response selection. Third,adaptation may relate to the site of pain. The left OImuscle, which rotates the trunk to the left (away fromthe posterior right injection site), most commonlyincreased activity. The pattern of adaptation to painmay also relate to movement subgroups identified inback pain in the clinical literature (Janda, 1996;Sahrman, 2002; Dankaerts et al., 2006).

Did individuals use a consistent pattern of adapta-tion between tasks? Largely, the pattern was similarbetween tasks; only 10% of the adaptations wereopposite in direction (e.g. decreased with one move-ment but increased with the other) and the musclesthat had most consistently augmented activity (e.g.trunk rotator muscles such as the oblique abdominalmuscles) were rarely affected in an opposite mannerbetween tasks for an individual. Thus, many partici-pants appeared to adopt a relatively consistentresponse during pain. This concurs with the clinicalassumption that back pain patients can be classifiedinto some movement subgroups regardless of task.

5. Conclusion

This study provides the first empirical confirmationthat, in most individuals, acute back pain is associatedwith adaptation of the trunk muscles in a mannerthat increases spinal stability and is consistent withthe goal of enhancing protection of the spine, but thepattern of muscle activity is not stereotypical acrossindividuals. Instead, the adaptation involves anindividual-specific response that is not predicted bymost existing theories of motor adaptation to pain.The adaptation of muscle activity is maintained acrosstasks within an individual despite opposite movementdemands. This adaptation is likely to provide short-term benefit, but could have long-term consequencesfor spinal health.

Author contributions

All authors were involved in conceptualization and design ofthe study, acquisition of data, analysis and interpretation ofdata. P.W.H. and J.C. drafted the article. All authors discussedthe results and commented on the manuscript and approvedthe final version. P.W.H. acquired funding.

Acknowledgements

This study was supported by a Senior Principal ResearchFellowship to PH [ID1002190] and a project grant[ID401598] from the National Health and Medical ResearchCouncil of Australia. We thank Wolbert van den Hoorn forcontribution to analysis of EMG data and Kylie Tucker forassistance with data collection.

References

Aminian, K., Ruffieux, C., Robert, C. (1988). Filtering by adaptive sam-pling (FAS). Med Biol Eng Comput 26, 658–662.

Astfalck, R.G., O’Sullivan, P.B., Straker, L.M., Smith, A.J., Burnett, A.,Caneiro, J.P., Dankaerts, W. (2010). Sitting postures and trunk muscleactivity in adolescents with and without nonspecific chronic low backpain: An analysis based on subclassification. Spine 35, 1387–1395.

Barker, P., Guggenheimer, K., Grkovic, I., Briggs, C., Jones, D., Thomas,C., Hodges, P. (2005). Effects of tensioning the lumbar fasciae on seg-mental stiffness during flexion and extension. Spine 31, 397–405.

Chapman, A., Vicenzino, B., Blanch, P., Hodges, P. (2008). Patterns of legmuscle recruitment vary between novice and highly trained cyclists. JElectromyogr Kinesiol 18, 359–371.

Cholewicki, J., McGill, S.M. (1996). Mechanical stability of the in vivolumbar spine: Implications for injury and chronic low back pain. ClinBiomech 11, 1–15.

Cholewicki, J., Panjabi, M.M., Khachatryan, A. (1997). Stabilizing func-tion of trunk flexor-extensor muscles around a neutral spine posture.Spine 22, 2207–2212.

Claus, A.P., Hides, J.A., Moseley, G.L., Hodges, P.W. (2009). Different waysto balance the spine: Subtle changes in sagittal spinal curves affectregional muscle activity. Spine 34, E208–E214.

Dankaerts, W., O’Sullivan, P.B., Straker, L.M., Burnett, A.F., Skouen, J.S.(2006). The inter-examiner reliability of a classification method fornon-specific chronic low back pain patients with motor control impair-ment. Man Ther 11, 28–39.

Ervilha, U.F., Arendt-Nielsen, L., Duarte, M., Graven-Nielsen, T. (2004).The effect of muscle pain on elbow flexion and coactivation tasks. ExpBrain Res 156, 174–182.

Graven-Nielsen, T., Svensson, P., Arendt-Nielsen, L. (1997). Effects ofexperimental muscle pain on muscle activity and coordination duringstatic and dynamic motor function. Electroencephalogr Clin Neurophysiol105, 156–164.

Hamill, J., van Emmerik, R.E., Heiderscheit, B.C., Li, L. (1999). A dynami-cal systems approach to lower extremity running injuries. Clin Biomech14, 297–308.

Hodges, P., Kaigle Holm, A., Holm, S., Ekstrom, L., Cresswell, A., Hansson,T., Thorstensson, A. (2003a). Intervertebral stiffness of the spine isincreased by evoked contraction of transversus abdominis and the dia-phragm: In vivo porcine studies. Spine 28, 2594–2601.

Hodges, P., Shirley, D., Eriksson, A.E.M., Gandevia, S.C. (2004). Intra-abdominal pressure can directly increase stiffness of the lumbar spine. JBiomech 38, 1873–1880.

Hodges, P., Tucker, K. (2011). Moving differently in pain: A new theory toexplain the adaptation to pain. Pain 152, S90–S98.

Hodges, P., van den Hoorn, W., Dawson, A., Cholewicki, J. (2009).Changes in the mechanical properties of the trunk in low back pain maybe associated with recurrence. J Biomech 42, 61–66.

Hodges, P.W., Moseley, G.L., Gabrielsson, A., Gandevia, S.C. (2003b).Experimental muscle pain changes feedforward postural responses ofthe trunk muscles. Exp Brain Res 151, 262–271.

Hodges, P.W., Richardson, C.A. (1996). Inefficient muscular stabilisationof the lumbar spine associated with low back pain: A motor controlevaluation of transversus abdominis. Spine 21, 2640–2650.

Howarth, S.J., Allison, A.E., Grenier, S.G., Cholewicki, J., McGill, S.M.(2004). On the implications of interpreting the stability index: A spineexample. J Biomech 37, 1147–1154.



Janda, V. (1996). Evaluation of muscular imbalances. In Rehabilitation ofthe Spine: A Practitioners Manual, C. Liebenson, ed. (Baltimore: Williamsand Wilkins) pp. 97–112.

Kumar, S. (1990). Cumulative load as a risk factor for back pain. Spine 15,1311–1316.

Latash, M.L., Anson, J.G. (2006). Synergies in health and disease: Relationsto adaptive changes in motor coordination. Phys Ther 86(8), 1151–1160.

Leinonen, V., Kankaanpaa, M., Luukkonen, M., Hanninen, O., Air-aksinen, O., Taimela, S. (2001). Disc herniation-related back painimpairs feed-forward control of paraspinal muscles. Spine 26, E367–E372.

Lund, J.P., Donga, R., Widmer, C.G., Stohler, C.S. (1991). The pain-adaptation model: A discussion of the relationship between chronicmusculoskeletal pain and motor activity. Can J Physiol Pharmacol 69,683–694.

MacDonald, D., Moseley, G.L., Hodges, P.W. (2009). Why do some patientskeep hurting their back? Evidence of ongoing back muscle dysfunctionduring remission from recurrent back pain. Pain 142, 183–188.

Marras, W.S., Ferguson, S.A., Burr, D., Davis, K.G., Gupta, P. (2004).Spine loading in patients with low back pain during asymmetric liftingexertions. Spine J 4, 64–75.

McGill, S. (2002). Low Back Disorders: Evidence Based Prevention and Reha-bilitation (Champaign, IL: Human Kinetics Publishers, Inc).

Mok, N.W., Brauer, S.G., Hodges, P.W. (2007). Failure to use movement inpostural strategies leads to increased spinal displacement in low backpain. Spine 32, E537–E543.

Moseley, G.L., Nicholas, M.K., Hodges, P.W. (2004). Does anticipation ofback pain predispose to back trouble? Brain 127, 2339–2347.

O’Sullivan, P.B. (2000). Lumbar segmental ‘instability’: Clinical presenta-tion and specific stabilizing exercise management. Man Ther 5, 2–12.

Panjabi, M.M. (1992). The stabilizing system of the spine. Part I. Function,dysfunction, adaptation, and enhancement. J Spinal Disord 5, 383–389.

Radebold, A., Cholewicki, J., Panjabi, M.M., Patel, T.C. (2000). Muscleresponse pattern to sudden trunk loading in healthy individuals and inpatients with chronic low back pain. Spine 25, 947–954.

Reeves, N.P., Narendra, K.S., Cholewicki, J. (2007). Spine stability: The sixblind men and the elephant. Clin Biomech 22, 266–274.

Richardson, C.A., Hodges, P.W., Hides, J.A. (2004). Therapeutic Exercise forLumbopelvic Stabilisation: A Motor Control Approach for the Treatment andPrevention of Low Back Pain (Edinburgh: Churchill Livingstone).

Roland, M. (1986). A critical review of the evidence for a pain-spasm-paincycle in spinal disorders. Clin Biomech 1, 102–109.

Sahrman, S. (2002). Diagnosis and Treatment of Movement Impairment Syn-dromes (St Louis: Mosby, Inc).

Sohn, M.K., Graven-Nielsen, T., Arendt-Nielsen, L., Svensson, P. (2000).Inhibition of motor unit firing during experimental muscle pain inhumans. Muscle Nerve 23, 1219–1226.

Svensson, P., Arendt-Nielsen, L., Houe, L. (1995). Sensory-motor inter-actions of human experimental unilateral jaw muscle pain: A quanti-tative analysis. Pain 64, 241–249.

van Dieen, J.H., Selen, L.P., Cholewicki, J. (2003). Trunk muscle activa-tion in low-back pain patients, an analysis of the literature. J Elec-tromyogr Kinesiol 13, 333–351.

van Dieën, J.H., Cholewicki, J., Radebold, A. (2003). Trunk musclerecruitment patterns in patients with low back pain enhance the sta-bility of the lumbar spine. Spine 28, 834–841.



new insight into motor adaptation to pain revealed by a combination of modelling and empirical...

Documents