Orthop Clin N Am 34 (2003) 245 – 254
Core stability exercise in chronic low back pain
Paul W. Hodges, BPhty (Hons), PhD
Department of Physiotherapy, The University of Queensland, Brisbane, Qld 4072, Australia
Exercise is commonly used in the management
of chronic musculoskeletal conditions, including
chronic low back pain (CLBP). The focus of exer-
cise is varied and may include parameters ranging
from strength and endurance training, to specific
training of muscle coordination and control. The
assumption underpinning these approaches is that
improved neuromuscular function will restore or
augment the control and support of the spine and
pelvis. In a biomechanical model of CLBP, which
assumes that pain recurrence is caused by repeated
mechanical irritation of pain sensitive structures [1],
it is proposed that this improved control and stability
would reduce mechanical irritation and lead to pain
relief [1]. Although this model provides explanation
for the chronicity of LBP, perpetuation of pain is
more complex, and contemporary neuroscience
holds the view that chronic pain is mediated by a
range of changes including both peripheral (eg,
peripheral sensitization) and central neuroplastic
changes [2]. Although this does not exclude the role
of improved control of the lumbar spine and pelvis
in management of CLBP, particularly when there is
peripheral sensitization, it highlights the need to
look beyond outdated simplistic models. One factor
that this information highlights is that the refinement
of control and coordination may be more important
than simple strength and endurance training for the
trunk muscles. The objective of this article is to
discuss the rationale for core stability exercise in the
management of CLBP, to consider critical factors for
Financial support was provided by the National Health
and Medical Research Council of Australia and The Swedish
Research Council.
E-mail address: p.hodges@shrs.uq.edu.au
0030-5898/03/$ – see front matter D 2003, Elsevier Inc. All rights reserved.
doi:10.1016/S0030-5898(03)00003-8
its implementation, and to review evidence for
efficacy of the approach.
The rationale for core stability exercise
Core stability exercise can be defined loosely as
the restoration or augmentation of the ability of the
neuromuscular system to control and protect the spine
from injury or reinjury. The term is variously used to
describe a spectrum of exercise approaches that have
the common goal to improve lumbopelvic control,
with varied rationales. In general, strategies can be
divided into two main groups: those that aim to
restore the coordination and control of the trunk
muscles to improve control of the lumbar spine and
pelvis [3,4], and those that aim to restore the capacity
(strength and endurance) of the trunk muscles to meet
the demands of control [5,6]. Although both compo-
nents are necessary, it is useful to consider the
rationales for each.
The ‘‘muscle capacity’’ model of core stability
exercise is based on the well-established premise that
stability of the spine is dependent on the contribution
of muscle. The contemporary view of spinal stability
is based on the Euler model that considers the control
of buckling forces [7 – 9]. This is based on the
understanding that buckling failure of the lumbar
spine, devoid of muscle, occurs with compressive
loading of as little as 90 N. This model argues that
muscle activity is required to act like guy wires to
stiffen intervertebral joints that they span to maintain
the lumbar spine in a mechanically stable equilibrium
[7 – 9]. This definition is relatively static and suggests
the maintenance of a set position of the spine. Few
studies have considered this model in more dynamic
terms [10]. Consistent with the proposal that muscle
activity is required to control stability and restrict
trunk motion; studies have identified increased activ-
ity [11,12] and coactivity of antagonist trunk muscles
[13], presumably to augment the control of the spine
in individuals with CLBP. On the basis of this
philosophy, a range of exercise strategies has been
developed to overcome the buckling force and aug-
ment the stability of the spine. In general, these
strategies involve exercises that encourage mainte-
nance of a stable lumbopelvic position, while forces
are applied with limb movements and loads.
The ‘‘control’’ model of core stability exercise is
based on the premise that lumbopelvic function and
health are dependent on the accurate interplay of the
trunk muscles. Stability and control of the spine is
dependent not only on the muscles but also on the
central nervous system (CNS) which must determine
the requirements of stability to plan and implement
strategies that meet these demands [1]. When the
CNS can predict that the control of the spine will be
challenged by an internal or external force (eg, as a
result of the reactive forces from a moving limb), it
has the capacity to plan strategies of muscle activity
in advance of the movement to prepare the spine (ie,
feedforward control). In other cases, when the per-
turbation is unexpected, the CNS can initiate a
response of the trunk muscles [14,15]. Thus, accurate
control of the lumbar spine and pelvis is dependent
not only on the capacity of muscle, but also on the
sensory system that provides information about the
status of stability, recognition of perturbation, and
the development of the ‘‘internal model of body
dynamics’’ that provides the CNS with the capacity
to predict the outcome of mechanical events and
predict the interaction between body segments and
the environment. Accurate control is also dependent
on the capacity of the motor system to plan the
appropriate strategy to meet the demands of stability.
This system is not only responsible for the control
of buckling forces, but also for the control of the
overall orientation of the spine and pelvis, and
intervertebral translation and rotation during trunk
movement. Although the control of these two ele-
ments cannot be completely separated, particular
attention must be paid to control of translations and
rotations. For instance, during an arc of movement, it
is important to control the coordination between
translation and rotation at the intervertebral levels
[16]. It has been shown that, if the spine is modeled
with one segment with no muscle attachment, the
spine is as stable as having no muscle at all [8]. Thus,
segmental control is an essential component for
spinal stability. There is an increasing body of litera-
ture that has outlined the normal strategies for control
of the spine, as well as changes in this control when
people have LBP. As a result, many contemporary
approaches to core stability exercise focus on reha-
bilitation of the control of the trunk muscles. In many
cases, this focus has been on retraining the function
of the deep intrinsic muscles of the lumbar spine and
pelvis (eg, transversus abdominis and lumbar multi-
fidus), and then on integration of the activity of the
deep and superficial trunk muscles in functional
tasks. The hypothesis here is that the coordinated
activity of the deep spinal muscles plays an important
role in the fine-tuning of intersegmental motion of the
spine and pelvis (eg, sacroiliac joints). Though these
muscles may not provide the largest contribution
to stability, their function is essential for optimal
spinal health.
It is impossible to separate these two rationales
completely, but there are differences in focus that
have lead to variations in the exercises incorporated
into a core stability regime. The following section
addresses the neuromuscular control of stability to
provide greater insight into the clinical decisions for
implementation of core stability exercise in the man-
agement of CLBP.
Normal and abnormal control of the lumbar spine
and pelvis
Many muscles span the lumbopelvic region and
contribute to its control and stability. In fact, all
muscles are required for this function. There is
considerable biomechanical and motor control evi-
dence, however, to argue that different muscles and
different control strategies may be involved in the
control of different elements of stability, ie, interver-
tebral control, and the control or buckling forces and
orientation of the lumbar spine and pelvis. The motor
system has considerable redundancy, with many
muscles capable of performing similar functions.
 P.W. Hodges / Orthop Clin N Am 34 (2003) 245–254
or a few segments and that have a limited moment
arm to move the joint, but an ideal anatomy to control
intervertebral motion. Bergmark included muscles
such as the lumbar multifidus in this group; however,
other muscles that satisfy these criteria are trans-
versus abdominis (TrA), intertransversarii and inter-
spinales, posterior fibers of psoas [19], and medial
fibers of quadratus lumborum [20]. Obliquus internus
has fibers that parallel the TrA and may contribute to
the force closure of the sacroiliac joint (SIJ) [21]. By
contrast, the global muscles cross several joints with
attachments to the pelvis and the thorax. These
muscles have a larger moment arm (ie, greater
torque-generating capacity) and are suited to the
control of orientation and external forces. Examples
of the global muscles include: rectus abdominis,
obliquus externus and internus abdominis, thoracic
portions of the longissimus and iliocostalis, lateral
fibers of quadratus lumborum, anterior fibers of
psoas, and latissimus dorsi. There is overlap between
these systems, and some muscles share features of
both, such as the lumbar longissimus and iliocostalis,
and the superficial fibers of multifidus having one
attachment to the lumbar vertebrae and sharing some
features of the local system.
Two local or deep intrinsic lumbopelvic muscles
have received attention specifically in the literature:
the TrA and the lumbar multifidus. TrA is a broad
sheet-like muscle with extensive attachments to the
lumbar vertebrae via the thoracolumbar fascia, and to
the pelvis and rib cage. The muscle fibers have a
relatively horizontal orientation and, therefore, min-
imal ability to move the spine, although they may
contribute to rotation [22,23]. Contribution to spinal
control is likely to involve modulation of intra-
abdominal pressure (IAP) and tensioning the thora-
columbar fascia. TrA has been shown to be associated
closely with control of IAP [22,24], and recent data
confirm that spinal stiffness is increased by IAP
[25,26]. Fascial tension may restrict intervertebral
motion directly or provide gentle segmental compres-
sion via the posterior layer of the thoracolumbar
fascia [27]. Recent porcine studies confirm that the
combined effect of IAP and fascial tension is required
for TrA to increase intervertebral stiffness, and the
mechanical effect of its contraction is reduced if the
fascial attachments are cut [25]. For sacroiliac sup-
port, TrA acts via the ilia to compress the SIJ
anteriorly [21], and this finding has been confirmed
in vivo [28].
The lumbar multifidus has five fascicles that arise
from the spinous process and lamina of each lumbar
vertebra and descend in a caudo-lateral direction [29].
The most superficial fibers cross up to five segments
247
and attach caudally to the ilia and sacrum, whereas the
deep fibers attach from the inferior border of a lamina
and cross a minimum of two segments to attach on the
mamillary process and facet joint capsule [30]. The
superficial fibers are distant from the centers of rota-
tion of the lumbar vertebrae, have an extension mo-
ment arm, and can control the lumbar lordosis [29]. By
contrast, the deep fibers have a limited moment arm
and a minor ability to extend the spine [31].
Multifidus can control intervertebral motion by
generation of intervertebral compression [32]. The
proximity of deep multifidus to the center of rotation
results in compression with minimal extension
moment. In addition, multifidus may contribute to
the control of intervertebral motion by control of
anterior rotation and translation of the vertebrae
[29], or by tensioning the thoracolumbar fascia as it
expands on contraction [33]. Several studies have
provided in vitro and in vivo evidence of the ability
of multifidus to control intervertebral motion [32,34].
The contribution of the superficial global muscles
to lumbopelvic movement and stability generally is
predictable based on the moment arm and the di-
rection of force provided by the muscles [17,35]. That
is, flexors generate flexion torque and oppose exten-
sion. In addition, it generally has been considered that
antagonist trunk muscles are coactivated to stiffen the
spine and prevent buckling [36]. For example, greater
activity of the superficial abdominal muscles has
been demonstrated during isometric trunk tasks than
predicted by a biomechanic model [35], and antago-
nistic activity of obliquus externus (OE) and internus
abdominus (OI) has been recorded during lifting and
isometric trunk efforts [35].
Optimal function of both systems is required to
maintain spinal function. The local system has only a
limited ability to influence the control of orientation,
and the global system has only a limited ability to
control intervertebral motion. It is important to con-
sider how these muscles are coordinated in function.
Studies of feedforward and feedback control of the
trunk suggest that the CNS uses separate strategies for
the control of intervertebral motion and orientation of
the spine. Initial studies of the deep muscles inves-
tigated locomotion and trunk movements [22,37]. In
these tasks, the superficial abdominal and extensor
muscles were active in a manner that was specific to
direction. TrA was tonically active, however, and not
influenced by the direction of movement [22,37].
Precise evidence of this differential control has come
from investigation of predictable challenges to spinal
stability caused by rapid limb movements. Early
studies of arm movements showed activity of leg
muscles and erector spinae before flexion of the arm
248 P.W. Hodges / Orthop Clin N Am 34 (2003) 245–254
[38]. More recently, studies have confirmed that TrA
and the deep fibers multifidus are activated as a
component of this anticipatory response [39 – 41].
Though the response of the superficial muscles is
linked to the direction of force [40,42], the activity of
TrA and deep multifidus is independent of the
direction of force [40 – 42]. That is, the superficial
muscles control the orientation of the spine, whereas
the deep muscles provide control of intersegmental
motion that is not specific to the direction of force.
An important additional finding of these studies is
that the CNS does not make the spine rigid by
coactivation of the superficial muscles but, instead,
uses controlled movement to help counteract the ap-
plied forces [42]. The differential control of the deep
and superficial muscles has been confirmed in exper-
iments that have manipulated task expectation [43].
Thus, feedforward studies indicate that the CNS
uses specialized strategies for each element of spinal
control (ie, nondirection specific, early, tonic coac-
tivation of the deep muscles and direction specific,
phasic activation of the superficial muscles). This is
supported by studies of feedback-mediated control.
Many studies have investigated activity of the super-
ficial muscles in response to external perturbations
such as a load dropped into a box held in the hands in
front of the body [44], translation of the support
surface [45], or trunk loading [24]. These studies
report direction-specific activity of the superficial
muscles to maintain spinal orientation. By contrast,
studies of TrA report activity irrespective of the
direction of force [24]. Recent data indicate that the
deep and superficial fibers of multifidus are differ-
entially activity in response to trunk loading [15].
This was only apparent, however, when the loading
was predictable.
Many studies report changes in motor control in
people with acute and chronic LBP. Though there is
considerable variability in results, differential changes
in activity of deep and superficial muscles have been
relatively consistent. In terms of deep muscle activity,
there is evidence of delayed activity of TrA in asso-
ciation with rapid limb movements in CLBP [46]. The
changes in TrA have been replicated when pain is
induced by intramuscular injection of hypertonic sa-
line into the longissimus muscle [47]. Notably in
studies of CLBP the changes were identified in people
who had a history of LBP but were in remission from
their symptoms. Although these studies have reported
a delay in activation, it is likely that the change is not
confined to this parameter but, instead, may be reflec-
tive of a general change in control. For example, tonic
activity of TrA, which is normally observed during
repetitive trunk [22] and limb movements [48], is
reduced during experimental pain [47], and relative
EMG activity of the rectus abdominis and the infero-
lateral abdominal wall is altered in CLBP [40].
There is preliminary evidence that the deep para-
spinal muscles show similar changes in activity. Dur-
ing functional tasks, there is reduced amplitude of
activity of the multifidus in CLBP [50], and altered
responses have been observed during loading of the
trunk. For example, when a load is unexpectedly
dropped into the hands, there is normally a short-
latency response of the paraspinal muscles [15,51];
in healthy control subjects, when the loading can be
anticipated, the response of the deep fibers of multi-
fidus occurs earlier [15]. When people with sciatica
catch a load that is predictable, however, the earlier
response of the paraspinal muscles does not occur [51].
Others, using an unexpected loading paradigm, report
both delayed response [44] and no change [12] in
paraspinal muscle activity. These changes in activity
are consistent with changes in the morphology and
fatigability of the muscle, which could be explained by
altered use of the muscle. For example, studies report
changes in muscle fiber composition [52], increased
fatigability [53], and a reduced cross-sectional area
that occurs as little as 24 hours after the onset of acute,
unilateral LBP [54]. In CLBP, there is evidence of fatty
infiltration into the paraspinal muscles [55].
There is a large body of literature that investigates
changes in the superficial trunk muscles in LBP.
Although the changes have been studied extensively,
there is considerable disagreement as to whether the
superficial muscles have reduced endurance and
strength capacity [56,57]. It has been suggested that
these changes may be more related to inactivity than
to pain [57]. Likewise, studies of superficial para-
spinal muscle activity have had variable results, with
some reporting increased activity [58], others report-
ing decreased [50] or asymmetrical activity [59], and
still others reporting no change in activity [60]. One
finding that has been observed consistently in people
with LBP is sustained activity of the erector spinae
muscles at the end of the range of spinal flexion—
a point at which the erector spinae muscles are
normally inactive (the so-called flexion-relaxation
response) [61]. This finding has been replicated by
experimental pain [12] and has been shown to limit
intervertebral motion [62]. The normal periods of
silence in erector spinae activity between heel con-
tacts are reduced, furthermore, in LBP patients and in
otherwise asymptomatic participants given experi-
mentally induced LBP [11], which may serve to
splint the area during this period. Others have iden-
tified delayed reduction of activity and increased
coactivation of superficial muscles in response to
 P.W. Hodges / Orthop Clin N Am 34 (2003) 245–254
removal of a load from the trunk in a subset of CLBP
[13]. Experimentally induced pain caused variable
responses of the superficial trunk muscles in asso-
ciation with rapid limb movements [47]. Importantly,
however, although there was considerable intersub-
ject variability in the pattern of superficial trunk
muscle activity, at least one superficial muscle was
augmented during pain in every subject. Notably,
despite the variation in superficial muscle activity,
hypoactivity of the intrinsic spinal muscle, TrA, was
a consistent finding across the group in that study.
In addition to changes in muscle recruitment,
impairment of other elements of the sensorimotor
system has been identified in CLBP that may affect
the normal control of the lumbar spine and pelvis. For
example, studies have reported reduced acuity [63]
and impaired ability to perform repositioning tasks
[64]. Because both feedforward- and feedback-medi-
ated components of motor control are dependent on
sensory input, any change in sensory input is likely to
be important.
In summary, the evidence suggests that, with LBP,
the activity of the deep and superficial muscles are
affected differentially. Although the results are vari-
able, there is evidence for augmented activity of at
least one of the superficial muscles in CLBP, whereas
the control of the deep intrinsic spinal muscles is
invariably impaired irrespective of the pathology.
Although the increase in superficial trunk muscle
activity can prevent lumbar spine buckling and
increase spinal stiffness [9,17], it is also associated
with increased compressive loading of the spine [36],
which has long been considered a risk factor for
spinal degeneration and pain. It is known, moreover,
that the CNS uses movement rather than stiffening of
the spine to overcome challenges to stability [42] and
reduce energy expenditure. A strategy of trunk stiff-
ening, although requiring less complex neural con-
trol, may compromise optimal spinal loading. Finally,
the response to pain of stiffening the spine appears to
be at the ‘‘cost’’ of a loss of the fine tuning of
intersegmental motion of the spine provided by the
deep muscles.
The mechanism of these changes is not com-
pletely understood, nor whether the changes precede
the pain or are a result of the painful episode. Though
it is not possible to exclude the possibility that motor
control changes may predispose an individual to pain,
recent evidence argues that pain may replicate the
changes seen in clinical populations [36]. The mech-
anism for the effect of pain may involve factors such
as changes in excitability along the motor pathway or
more complex changes caused by fear-avoidance
behaviors. This topic is reviewed elsewhere [65].
249
Clinical factors in the implementation of core
stability exercise
From the preceding discussion, it is clear that the
specific nature of the core stability exercise program
can vary depending on whether the focus is on control
or muscle capacity. In reality all components require
consideration, and a specific intervention must be
tailored to the individual patient’s presentation. This
section addresses factors to consider in the imple-
mentation of the spectrum of exercise interventions
that fall under the umbrella of core stability exercise.
In terms of retraining the control of core stability,
evidence thus far suggests that there are specific
changes in the strategies used by the CNS to control
the spine, and that this consistently involves impaired
activity of the deep muscle system, often in asso-
ciation with overactivity of one or more superficial
muscles. It is unlikely that general exercise for the
trunk such as sit-ups and back extension exercises
would restore the coordination between the trunk
muscles, and a strategy based on the principles of
motor relearning and skill acquisition is required.
This approach aims to train the skilled activation of
the deep muscles, to train the integration of the deep
and superficial systems, and to progress through a
program of tailored functional exercises in varying
environments and contexts to ensure transfer to
normal activity [4].
The nervous system has considerable potential for
plasticity and learning. A key strategy described in
the literature to retrain motor function involves prac-
tice of ‘‘parts’’ of movement rather than the ‘‘whole’’
movement [66]. When a skill is trained in parts, the
attention demand is reduced to allow attention to be
focused on a single element. Several techniques have
been described for part-task training. These include
segmentation and simplification [66]. For segmenta-
tion, the task is divided into smaller parts, to be
practiced as an independent unit, and then the prac-
ticed elements are integrated together progressively to
practice the complete skill. A key feature of this
strategy is selection of the specific features of a
movement that are impaired or dysfunctional (ie,
‘‘essential components’’), and then implementation
of strategies that optimize the performance of that
component using cognitive strategies. At a later stage,
it is important to perform the interdependent parts of
a task together, and to integrate the trained compo-
nents in a functional context [67]. From the evidence
presented above, the component of movement that is
impaired in normal function is the activity of the deep
muscle system (the activity of this system is delayed,
more phasic, and no longer independent of the
250 P.W. Hodges / Orthop Clin N Am 34 (2003) 245–254
superficial muscles). Thus, the focus of initial stages
of rehabilitation is to train this component, indepen-
dently from the superficial muscles. Simplification
refers to strategies to increase the ease of movement
performance and may be achieved by changing
parameters such as postural load (eg, training in
supported positions), reduction of attention demands,
reduction of speed, or additional strategies to aug-
ment performance (eg, improved accuracy of feed-
back with ultrasound imaging).
Motor learning occurs in three main phases:
cognitive, associative, and autonomous [68]. In the
cognitive phase, the focus is on cognitively oriented
problems. All elements of the movement performance
are organized consciously with attention to feedback,
movement sequence, performance, and instruction
during repetition and practice. In motor relearning
for LBP, the goal of the initial phase of motor
relearning is to contract the deep muscles cognitively
to increase the precision and skill of the contraction
of the local muscles. Once mastered, the goal shifts to
increased precision, increased number of repetitions
and holding time, and decreased feedback. In the
associative phase, the fundamentals of the skill have
been acquired, and the cognitive demands reduced.
The focus moves from simple elements of task
performance to consistency of performance, success,
and refinement. In this stage, many repetitions are
required in a variety of contexts to reduce the
cognitive demand of the task. In terms of trunk
muscle training, this phase involves performance of
the task in increasingly challenging positions (eg,
sitting and standing) and integration of deep and
superficial muscle function, eg, using leg-loading
tasks [18], proprioceptive neuromuscular facilitation
techniques [69], and postural challenges [3]. The final
stage of motor learning, the autonomous phase, is
achieved after considerable practice and experience.
The task becomes habitual or automatic, and the
requirement for conscious intervention is reduced.
Importantly, there is preliminary evidence that train-
ing in this manner can result in a change in the
automatic control of the spine [70,71] and peripheral
joints [72].
An important element of motor learning is the
provision of augmented feedback. Feedback can be
divided generally into two main types: knowledge of
performance and knowledge of results [73]. Put
simply, feedback that provides knowledge of per-
formance relates to ongoing information provided
during movement, whereas knowledge of results is
feedback of the outcome of the task. In the cognitive
phase, it is critical to provide accurate feedback of
contraction quality. This feedback may be intrinsic
(naturally occurring), or augmented in some way, and
may involve any of the senses including tactile
(palpation) visual (ultrasound imaging) [74], and
auditory (EMG biofeedback) information.
A critical component of skill learning is that the
performance of the skill can be transferred to different
conditions in which the environment, personal char-
acteristics, or predictability are changed. To optimize
transfer, it is considered essential to progress the task
sequentially from easy to more complex situations. If
the aim is to transfer a skilled movement to functional
tasks, then it is necessary to progress to function.
More specifically, it may be necessary to replicate
sensory characteristics (eg, limitation of visual feed-
back), environmental contexts (eg, unstable surfaces,
distractions), and personal contexts (eg, anxiety,
fatigue) [75] to ensure that the elements of the skill
can be transferred to specific contexts [66].
Another issue to consider is the importance of the
neutral position of the spine, which is variously
described as being the normal lumbar, cervical lor-
dosis, and thoracic kyphosis. Around the neutral
position, the spine exhibits the least stiffness [1].
This region is important to consider because stability
is dependent on the contribution of the trunk muscles.
In addition, it has been argued that loading through
the spine is optimal in neutral positions [5]. Clin-
ically, training trunk-muscle control in neutral posi-
tions has been argued to be important for effective
management of CLBP [4,6].
In summary, in a motor control model to progress
a patient with LBP through the normal phases of
motor learning, the basic sequence that needs to be
undertaken is: (1) skill learning, (2) precision train-
ing, (3) activation in a variety of contexts (including a
variety of postures and positions), (4) integration of
the skill into tasks that includes activation of the
superficial trunk muscles, and (5) specific functional
retraining to ensure that the appropriate coordination
of deep and superficial trunk muscles is maintained in
a functional context.
In contrast with the motor relearning approach to
retraining the control and coordination of the trunk
muscles, at the other end of the spectrum of core
stability exercise are tasks aimed at restoring the
ability of the trunk muscles themselves to meet the
demands for spinal stability. Notably, as the mag-
nitude of force acting on the trunk is increased, the
relative contribution of the superficial trunk muscles
is increased [5]. Thus, the system must be trained to
meet the demands of high-level activities. The strat-
egies employed overlap with those described above.
Though these tasks form the central component of
some core stability exercise programs, they may be
 P.W. Hodges / Orthop Clin N Am 34 (2003) 245–254
used in a motor control model as progressive training
to rehabilitate the integration of the local and global
systems for higher-level activities. Many exercise
strategies have been developed for this component
of core stability exercise. Strategies that are consistent
with this approach include Pilates exercise, dissoci-
ation exercises [18], limb movements with a neutral
lumbar spine and pelvis [4], Swiss ball programs
[76], McGill’s stability exercises [5], rhythmic stabili-
zation exercises from proprioceptive neuromuscular
facilitation [69], balance board and balance shoe
training [77], floor exercise programs [6], and so
on. The key features of exercise vary between
approaches but include features such as control of
the lumbopelvic position (particularly in neutral)
during movement of the limbs [4,6,18,76], and on
unstable surfaces [76,77], closed chain tasks [4], and
resisted movements [78].
Efficacy of core stability exercise in the
management of chronic LBP
There are an increasing number of clinical studies
that have investigated the efficacy of core stability
exercise in the management of CLBP. Although early
studies were based on audits of clinical outcome [6],
more recently, high quality randomized controlled
clinical trials have been conducted. The first study
investigated core stability exercise, from a motor
control perspective, in people with CLBP associated
with spondylolisthesis [79]. In this study, subjects
were allocated randomly to either participate in a
motor-relearning program or a nontreatment group.
The training period lasted for 10 weeks. At the
completion of training and at follow-up at 30 months,
there was a significant reduction in pain and disability
in the motor-relearning group. There was no signifi-
cant change in the nontreatment group. The second
study involved training in acute first-episode unilat-
eral LBP [80]. This group was selected because they
have a reduced cross-sectional area of multifidus
ipsilateral to their symptoms [54 ]. The intervention
involved a 4-week program of motor relearning,
focused on multifidus in conjunction with TrA. After
4 weeks, all pain and disability measures had recov-
ered in all but one participant. This is consistent with
epidemiologic data. The size of multifidus had recov-
ered only in the motor-control training group [80].
The follow-up data provide potent evidence for the
efficacy of the approach. After 3 years, people in the
control group were 12.4 times more likely to have
further episodes of pain that those in the exercise
group [81]. Although the data deal with an acute
251
population, they are important to consider here
because recurrence of pain is a major factor in CLBP.
Other studies of chronic headache [70] and patellofe-
moral pain [72], using a similar clinical paradigm,
have shown positive clinical outcomes. Additional
studies of CLBP [82] and pelvic pain [83] have
promising early results.
Another focus of clinical studies has been to
identify the mechanism of efficacy of the clinical
approach. In general, studies of muscle size [84],
muscle strength, muscle endurance, and muscle fiber
composition [85] have failed to find improvement or a
relationship with the clinical outcome. Improvements
in motor control parameters such as muscle activation
patterns [49,71,72] and performance of specific
skilled activities [70] have found a positive relation-
ship, however. Thus, there is increasing evidence of
efficacy of core stability exercise, particularly from a
motor-control perspective, and there is evidence that
the improvements are related to the factor being
addressed in the intervention, not a tertiary cause.
Summary
In conclusion, core stability exercise is an evolv-
ing process, and refinement of the clinical rehabili-
tation strategies is ongoing. Two major foci are
addressed in contemporary core stability programs:
motor control and muscle capacity. Both of these
factors have considerable foundation in the literature
and can be seen as a progression of exercise rather
than conflicting approaches. Importantly, the clinical
efficacy of these approaches is being realized in
clinical trials. Further work is required, however, to
refine and validate the approach, particularly with
reference to contemporary understanding of the neu-
robiology of chronic pain.
References
[1] Panjabi MM. The stabilizing system of the spine.
Part 1. Function, dysfunction, adaptation, and enhance-
ment. J Spinal Disord 1992;5(4):383 – 9.
[2] Butler DS. Sensitive nervous system. Adelaide, Aus-
tralia: NOIgroup Publications; 2000.
[3] Janda V. Muscles, central nervous motor regulation
and back problems. In: Korr IM, editor. The neurobio-
logic mechanisms in manipulative therapy. New York:
Plenium Press; 1978. p. 27 – 41.
[4] Richardson CA, Jull GA, Hodges PW, Hides JA. Ther-
apeutic exercise for spinal segmental stabilization in
LBP: scientific basis and clinical approach. Edinburgh:
Churchill Livingstone; 1999.
252 P.W. Hodges / Orthop Clin N Am 34 (2003) 245–254
[5] McGill S. Low back disorders: evidence based preven-
tion and rehabilitation. Champaign, IL: Human Ki-
netics Publishers, Inc.; 2002.
[6] Saal JA, Saal JS. Non-operative treatment of herniated
lumbar intervertebral disc with radiculopathy: an out-
come study. Spine 1989;14:431 – 7.
[7] Cholewicki J, McGill SM. Mechanical stability of the
in vivo lumbar spine: implications for injury and
chronic low back pain. Clin Biomech 1996;11(1):
1 – 15.
[8] Crisco JJ, Panjabi MM. The intersegmental and multi-
segmental muscles of the lumbar spine: a biomechanical
model comparing lateral stabilising potential. Spine
1991;7:793 – 9.
[9] Gardner-Morse M, Stokes IAF, Laible JP. Role of
muscles in lumbar spine stability in maximum exten-
sion efforts. J Orthopaed Res 1995;13:802 – 8.
[10] Cholewicki J, Panjabi MM, Khachatryan A. Stabilizing
function of trunk flexor-extensor muscles around a
neutral spine posture. Spine 1997;22(19):2207 – 12.
[11] Arendt-Nielsen L, Graven-Nielsen T, Svarrer H,
Svensson P. The influence of low back pain on muscle
activity and coordination during gait. Pain 1996;64(2):
231 – 40.
[12] Zedka M, Prochazka A, Knight B, Gillard D, Gauthier
M. Voluntary and reflex control of human back muscles
during induced pain. J Physiol 1999;520(Pt 2):
591 – 604.
[13] Radebold A, Cholewicki J, Panjabi MM, Patel TC.
Muscle response pattern to sudden trunk loading in
healthy individuals and in patients with chronic low
back pain. Spine 2000;25(8):947 – 54.
[14] Indahl A, Kaigle AM, Reikeras O, Holm SH. Interac-
tion between the porcine lumbar intervertebral disc,
zygapophysial joints, and paraspinal muscles. Spine
1997;22(24):2834 – 40.
[15] Moseley GL, Hodges PW, Gandevia SC. External per-
turbation of the trunk in standing humans differentially
activates components of the medial back muscles.
J Physiol 2002; in press.
[16] Bogduk N, Amevo B, Pearcy M. A biological basis for
instantaneous centers of rotation of the vertebral col-
umn. Proc Inst Mech Eng 1995;209(3):177 – 83.
[17] Bergmark A. Stability of the lumbar spine. A study in
mechanical engineering. Acta Orthoped Scand 1989;
60(suppl 230):1 – 54.
[18] Sahrman S. Diagnosis and treatment of movement
impairment syndromes. St Louis: Mosby, Inc.; 2002.
[19] Bogduk N, Pearcy M, Hadfeild G. Anatomy and bio-
mechanics of psoas major. Clin Biomech 1992;7:
109 – 19.
[20] McGill S, Juker D, Kropf P. Quantitative intramuscular
myoelectric activity of quadratus lumborum during a
wide variety of tasks. Clin Biomech 1996;11(3):
170 – 2.
[21] Snijders CJ, Vleeming A, Stoeckart R, Mens JMA,
Kleinrensink GJ. Biomechanical modelling of sacroil-
iac joint stability in different postures. Spine: State Art
Rev 1995;9:419 – 32.
[22] Cresswell AG, Grundstrom H, Thorstensson A. Obser-
vations on intra-abdominal pressure and patterns of
abdominal intra-muscular activity in man. Acta Physiol
Scand 1992;144:409 – 18.
[23] Urquhart D, Story I, Hodges P. Rotation transversus
abdominis. In: Proceedings of the 7th International
Physiotherapy Conference, Sydney, Australia. Mel-
bourne: Australian Physiotherapy Association; 2002.
p. 45.
[24] Cresswell AG, Oddsson L, Thorstensson A. The influ-
ence of sudden perturbations on trunk muscle activity
and intra-abdominal pressure while standing. Exp
Brain Res 1994;98(2):336 – 41.
[25] Hodges P, Kaigle-Holm A, Holm S, Ekstrom L, Cress-
well AG, Hansson T, et al. In vivo evidence that pos-
tural activity of the respiratory muscles increases
intervertebral stiffness of the spine: porcine studies.
Soc Neurosci Abstr 2002;28:366.
[26] Hodges PW, Cresswell AG, Daggfeldt K, Thorstensson
A. In vivo measurement of the effect of intra-abdomi-
nal pressure on the human spine. J Biomech 2001;34:
347 – 53.
[27] Gracovetsky S, Farfan H, Helleur C. The abdominal
mechanism. Spine 1985;10(4):317 – 24.
[28] Richardson CA, Snijders CJ, Hides JA, Damen L, Pas
MS, Storm J. The relation between the transversus
abdominis muscles, sacroiliac joint mechanics, and
LBP. Spine 2002;27(4):399 – 405.
[29] Macintosh JE, Bogduk N. The detailed biomechanics of
the lumbar multifidus. Clin Biomech 1986;1:205 – 31.
[30] Lewin T, Moffett B, Viidik A. The morphology of the
lumbar synovial joints. Acta Morphologica Neerlanco
Scandinav 1962;4:299 – 319.
[31] Panjabi MM, Abumi K, Duranceau J, Oxland T. Spinal
stability and intersegmental muscle forces. A biome-
chanical model. Spine 1989;14:194 – 200.
[32] Wilke HJ, Wolf S, Claes LE, Arand M, Wiesend A.
Stability increase of the lumbar spine with different
muscle groups: a biomechanical in vitro study. Spine
1995;20(2):192 – 8.
[33] Gracovetsky S, Farfan HF, Lamy C. A mathematical
model of the lumbar spine using an optimised system
to control muscles and ligaments. Orthop Clin North
Am 1977;8:135 – 53.
[34] Kaigle AM, Holm SH, Hansson TH. Experimental in-
stability in the lumbar spine. Spine 1995;20(4):421 – 30.
[35] Zetterberg C, Andersson GB, Schultz AB. The activity
of individual trunk muscles during heavy physical
loading. Spine 1987;12:1035 – 40.
[36] Gardner-Morse MG, Stokes IA. The effects of abdomi-
nal muscle co-activation on lumbar spine stability.
Spine 1998;23(1):86 – 91.
[37] Pauly J. An electromyographic analysis of certain
movements and exercises: I. Some deep muscles of
the back. Anat Rec 1966;155:223 – 34.
[38] Belenkii V, Gurfinkel VS, Paltsev Y. Elements of con-
trol of voluntary movements. Biofizika 1967;12(1):
135 – 41.
[39] Hodges PW, Richardson CA. Contraction of the ab-
 P.W. Hodges / Orthop Clin N Am 34 (2003) 245–254
dominal muscles associated with movement of the
lower limb. Phys Ther 1997;77:132 – 44.
[40] Hodges PW, Richardson CA. Feedforward contraction
of transversus abdominis in not influenced by the di-
rection of arm movement. Exp Brain Res 1997;114:
362 – 70.
[41] Moseley GL, Hodges PW, Gandevia SC. Deep and
superficial fibers of lumbar multifidus are differentially
active during voluntary arm movements. Spine 2002;
27:E29 – 36.
[42] Hodges PW, Cresswell AG, Thorstensson A. Prepara-
tory trunk motion accompanies rapid upper limb move-
ment. Exp Brain Res 1999;124:69 – 79.
[43] Hodges PW, Richardson CA. Transversus abdominis
and the superficial abdominal muscles are controlled
independently in a postural task. Neurosci Lett 1999;
265(2):91 – 4.
[44] Wilder DG, Aleksiev AR, Magnusson ML, Pope MH,
Spratt KF, Goel VK. Muscular response to sudden
load. A tool to evaluate fatigue and rehabilitation.
Spine 1996;21(22):2628 – 39.
[45] Henry SM, Fung J, Horak FB. EMG responses to
maintain stance during multidirectional surface trans-
lations. J Neurophysiol 1998;80(4):1939 – 50.
[46] Hodges PW, Richardson CA. Delayed postural con-
traction of transversus abdominis associated with
movement of the lower limb in people with LBP.
J Spinal Disord 1998;11(11):46 – 56.
[47] Hodges PW, Moseley GL, Gabrielsson A, Gandevia
SC. Experimental muscle pain changes feed forward
postural responses of the trunk muscles. Exp Brain Res
2003; in press.
[48] Hodges P, Gandevia S. Changes in intra-abdominal
pressure during postural and respiratory activation of
the human diaphragm. J Appl Physiol 2000;89(3):
967 – 76.
[49] O’Sullivan PB, Twomey L, Allison GT. Altered ab-
dominal muscle recruitment in patients with CLBP
following a specific exercise intervention. J Orthop
Sports Phys Ther 1998;27(2):114 – 24.
[50] Sihvonen T, Lindgren KA, Airaksinen O, Manninen H.
Movement disturbances of the lumbar spine and abnor-
mal back muscle electromyographic findings in recur-
rent LBP. Spine 1997;22:289 – 95.
[51] Leinonen V, Kankaanpaa M, Luukkonen M, Hanninen
O, Airaksinen O, Taimela S. Disc herniation-related
BP impairs feed-forward control of paraspinal muscles.
Spine 2001;26(16):E367 – 72.
[52] Rantanen J, Hurme M, Falck B, Alaranta H, Nykvist F,
Lehto M, et al. The lumbar multifidus muscle five
years after surgery for a lumbar intervertebral disc her-
niation. Spine 1993;18(5):568 – 74.
[53] Roy SH, DeLuca CJ, Casavant DA. Lumbar muscle
fatigue and chronic low back pain. Spine 1989; 14(9):
992 – 1001.
[54] Hides JA, Stokes MJ, Saide M, Jull GA, Cooper DH.
Evidence of lumbar multifidus muscle wasting ipsilat-
eral to symptoms in patients with acute/subacute low
back pain. Spine 1994;19:165 – 77.
253
[55] Alaranta H, Tallroth K, Soukka A, Heliaara M. Fat
content of lumbar extensor muscles in low back dis-
ability: a radiographic and clinical comparison. J Spi-
nal Disord 1993;6(2):137 – 40.
[56] Suzuki N, Ohe K. Abdominal and back muscle
strength in patients with low back pain. Orthop Surg
1978;29:325 – 8.
[57] Thorstensson A, Arvidson Å. Trunk muscle strength
and low back pain. Scand J Rehabil Med 1982;14:
69 – 75.
[58] Arena JG, Sherman RA, Bruno GM, Young TR. Elec-
tromyographic recordings of 5 types of low back pain
subjects and non-pain controls in different positions.
Pain 1989;37(1):57 – 65.
[59] Cram JR, Steger JC. EMG scanning in the diagnosis of
chronic pain. Biofeedback Self Regul 1983;8:229 – 41.
[60] Collins GA, Cohen MJ, Naliboff BD, Schandler SL.
Comparative analysis of paraspinal and frontalis EMG,
heart rate and skin conductance in chronic low back
pain patients and normals to various postures and
stresses. Scand J Rehabil Med 1982;14:39 – 46.
[61] Shirado O, Ito T, Kaneda K, Strax TE. Flexion-relaxa-
tion phenomenon in the back muscles. A comparative
study between healthy subjects and patients with
chronic low back pain. Am J Phys Med Rehabil 1995;
74(2):139 – 44.
[62] Kaigle AM, Wessberg P, Hansson TH. Muscular and
kinematic behavior of the lumbar spine during flexion-
extension. J Spinal Disord 1998;11(2):163 – 74.
[63] Gill KP, Callaghan MJ. The measurement of lumbar
proprioception in individuals with and without low
back pain. Spine 1998;23(3):371 – 7.
[64] Brumagne S, Cordo P, Lysens R, Verschueren S, Swin-
nen S. The role of paraspinal muscle spindles in lum-
bosacral position sense in individuals with and without
LBP. Spine 2000;25:989 – 94.
[65] Hodges PW, Moseley GL. Pain and motor control of
the lumbopelvic region: effect and possible mech-
anisms. Electromyogr Kinesiol 2003; in press.
[66] Magill RA. Motor learning: concepts and applications.
New York: McGraw-Hill; 2001.
[67] Carr JH, Shepherd RB. A motor relearning programme
for stroke. 2nd edition. London: Heinemann; 1987.
[68] Fitts PM, Posner MI. Human performance. Belmont,
CA: Brooks/Cole; 1967.
[69] Voss DE, Ionta MK, Myers BJ. Proprioceptive neuro-
muscular facilitation: patterns and techniques. New
York: Lippincott Williams & Wilkins Publishers; 1985.
[70] Jull G, Trott P, Potter H, Zito G, Niere K, Shirley D,
et al. A randomized controlled trial of exercise and
manipulative therapy for cervicogenic headache. Spine
2002;27:1835 – 43.
[71] Jull GA, Scott Q, Richardson C, Henry S, Hides J,
Hodges P. New concepts for the control of pain in
the lumbopelvic region. In: Vleeming A, Mooney V,
Tilscher H, Dorman T, Snijders C, editors. Proceedings
of the 3rd Interdisciplinary World Congress on Low
Back and Pelvic Pain, Vienna, Austria. Rotterdam:
European Conference Organisers; 1998. p. 128 – 31.
254 P.W. Hodges / Orthop Clin N Am 34 (2003) 245–254
[72] Cowan SM, Bennel KL, Crossley KM, Hodges PW,
McConnell J. Physical therapy alters recruitment of the
vasti in patellofemoral pain syndrome. Med Sci Sports
Exerc 2002;34:1879 – 85.
[73] Schmidt RA, Lee TD. Motor control and learning:
a behavioural emphasis. 3rd edition. Champaign, IL:
Human Kinetics Publishers; 1999.
[74] Hides JA, Richardson CA, Jull GA, Davies SE. Ultra-
sound imaging in rehabilitation. Aus J Physiother
1996;41:187 – 93.
[75] Moseley GL, Hodges PW. Chronic pain and motor
control. In: Boyling J, Jull GA, editors. Grieve’s mod-
ern manual therapy. Edinburgh: Churchill Livingstone;
2003. In press.
[76] Creager CC. Therapeutic exercises using the Swiss ball.
Berthoud (CO): Executive Physical Therapy; 1994.
[77] Bullock-Saxton JE, Janda V, Bullock MI. Reflex acti-
vation of gluteal muscles in walking with balance
shoes. Spine 1993;18(6):704 – 8.
[78] Mannion AF, Muntener M, Taimela S, Dvorak J. A
randomized clinical trial of three active therapies for
chronic low back pain. Spine 1999;24(23):2435 – 48.
[79] O’Sullivan PB, Twomey LT, Allison GT. Evaluation of
specific stabilizing exercise in the treatment of chronic
low back pain with radiologic diagnosis of spondylo-
lysis or spondylolisthesis. Spine 1997;22(24):2959 – 67.
[80] Hides JA, Richardson CA, Jull GA. Multifidus muscle
recovery is not automatic after resolution of acute, first-
episode low back pain. Spine 1996;21(23):2763 – 9.
[81] Hides JA, Jull GA, Richardson CA. Long term effects
of specific stabilizing exercises for first episode low
back pain. Spine 2001;26:243 – 8.
[82] Goldby L, Moore A, Doust J, Trew M. An RCT in-
vestigating the efficacy of manual therapy, exercises to
rehabilitate spinal stabilization and an education book-
let in the conservative treatment of chronic low back
pain. In: Proceedings of International Federation of
Manipulative Therapists. Perth, Australia: 2000.
[83] Stuge B. The efficacy of a specific stabilizing exer-
cise program in the treatment of patients with peri-
partum pelvic pain after pregnancy. A randomized
controlled trial. In: Proceedings of the 4th Interdisci-
plinary World Congress on Low Back and Pelvic
Pain, Montreal, Canada. Rotterdam: European Con-
ference Organisers; 2001.
[84] Daneels L, Cools A, Vanderstraeten G, Gambier D,
Witrouw E, Bourgois J, et al. The effect of 3 different
training modalities on the cross sectional area of para-
vertebral muscles. Scand J Med Sci Sports 2001;11(6):
335 – 41.
[85] Mannion AF, Taimela S, Muntener M, Dvorak J. Ac-
tive therapy for chronic low back pain. Part 1. Effects
on back muscle activation, fatigability, and strength.
Spine 2001;26(8):897 – 908.