Low back pain (LBP) is the most common musculo-skeletal disorder causing huge humanitarian and economical costs (Andersson, 1999; Aromaa et al., 2002). The pain may arise from injured structures within the lumbar spine, however the anatomical cause and functional consequences of the LBP often remain undefined. In chronic pain, the precise aetiology is even less well understood. Imaging studies such as MRI visualise the macroscopic structural changes but similar changes are often found in healthy subjects and frequently the correlation with the clinical condition seems to be rather low (Jensen et al., 1994; Jarvik et al., 2001).

The chronic LBP is a multidimensional problem including pain and functional disability with its associated socioeconomical consequences. Impairments in neuromuscular control have often been associated with chronic LBP and are considered a probable link between pain and disability (Luoto, 1999; Ebenbichler et al., 2001; Holm et al., 2002; Hodges and Moseley, 2003; van Dieen et al., 2003b). There is increasing evidence that the impaired functions can recover with treatment and be restored by active rehabilitation (Ebenbichler et al., 2001).
The appropriate muscular control and movement perception is of vital importance in preventing low back injury. The protection against injury requires anticipation of events and adequate muscular responses. Both abnormal and missing protective reflexes could possibly lead to trauma or microtrauma of muscles, nerves, intervertebral discs and ligamentous spine during loading (McGill, 1997).

The impaired neuromuscular control becomes particularly emphasised in conditions of chronic pain, which are also the most expensive for the society and the most demanding for the health care system. The etiological research is important for developing more effective treatment modalities especially in prolonged and chronic pain conditions. Unfortunately we have only a limited understanding of the time-course, appearance and clinical relevance of changes in motor control in LBP attributable to different pathophysiological mechanisms.

This study has investigated the postural control and lumbar motor and proprioceptive functions in specific lumbar disorders of disc herniation and lumbar spinal stenosis. The purpose was to evaluate the phenomena involved in motor control in LBP and patient groups with specific back disorders were used as experimental models.

Classification of lumbar disorders
Classification of the LBP is still a challenging problem. Lumbar diseases are often divided into specific and non-specific conditions. In non-specific (idiopathic) cases the origin of the pain is unknown, however, there are several potential sources of pain like muscles, tendons, ligaments, nerves, discs etc. The (rare) specific spinal diseases are attributable to fractures, tumours and infections etc. The more common specific conditions are disc herniation and lumbar spinal stenosis, which are the main causes of sciatica (Frymoyer, 1988; Deyo and Weinstein, 2001). The LBP is also divided by its duration, usually the pain lasting less than 6 weeks is called acute, 6-12 weeks is subacute and over 3 months is chronic LBP (Bigos et al., 1994).

Low back pain

Pathophysiology
The underlying mechanisms of LBP are usually unknown but there are several potential sources and causes of pain. Currently, there seems to be only concepts but little hard evidence for the development of the LBP. One potential course of the syndrome could be as follows: first the micro-injury e.g. to the muscles, tendons, ligaments, intervertebral disc or endplate. Injury is followed by macrophage and neutrophil accumulation and inflammation (Ahn et al., 2002; Burke et al., 2002). Static lumbar flexion has been indicated to cause the development of creep and micro-injuries of the collageneous structures in the lumbar ligaments. This elicits acute inflammation and hyperexcitability of the multifidus muscle (Claude et al., 2003; Solomonow et al., 2003a). This reflexive muscular stiffness is thought to protect the injured structures and enabling tissue healing (Claude et al., 2003; van Dieen et al., 2003a; Solomonow et al., 2003a).

Reactive oxygen species (ROS) seem to contribute to muscle fatigue (Reid, 2000) and therefore they may have role also in the pathophysiology of LBP. Dorsal ramus neuropathy is thought to be one potential cause of LBP (Sihvonen, 1995). Experimental studies have indicated that pain arising from the passive or active structures can lead to pathological activation of the paraspinal muscles at the same and adjacent segments (Indahl, 1999; Solomonow et al., 2003a). Despite this important finding, one still requires unequivocal clinical evidence about these protective reflexes. In the clinical course of a long-lasting spinal disorder, ultimately, the chronic pain may lead to deconditioning of the muscles and neural modulation of the central nervous system (Flor et al., 1997; Kankaanpää, 1999; Luoto, 1999).

Symptoms and findings
The primary diagnostic evaluation includes perusal of the medical history and a physical examination. This aims to exclude or diagnose serious causes of pain and nerve root compression. In addition, the social and psychological factors causing distress possibly complicating the recovery should be clarified (Bigos et al., 1994; Waddell et al., 1996; Deyo and Weinstein, 2001). Patients suffering prolonged pain and when there is suspicion of a serious illness or persisting radicular symptoms need further evaluation.

In the acute condition, imaging is often unnecessary (Deyo and Weinstein, 2001). The plain radiograph is currently the basic examination procedure for prolonged LBP lasting over six weeks (Jarvik and Deyo, 2002). The primary supplementary examination is MRI, which reveals also macroscopic soft tissue changes (Herzog et al., 1995) and can even be considered as the primary procedure in the patients with radicular symptoms when surgery is potentially indicated. Disc degeneration is a usual finding also in asymptomatic subjects (Boden et al., 1990; Tertti et al., 1991; Jensen et al., 1994; Jarvik et al., 2001) and does not appear to have a cause-effect relationship with pain. Disc disruptions seems to have clinical significance (Kuslich et al., 1991; Moneta et al., 1994), but the clinical relevance of the indicative MRI findings of the disc disruptions is rather weak (Videman et al., 2003) and the need for their invasive evaluation (discography) and treatment is not clear. Therefore, evaluation of the functional ability is often more important than anatomic findings (Bigos et al., 1994; Waddell et al., 1996; Deyo and Weinstein, 2001).

Treatment and prognosis
LBP is not a self-limiting condition (Hestbaek et al., 2003). Effective pain medication is essential in the acute condition, and it may prevent the pain from becoming chronic (Bigos et al., 1994; Waddell, 1996; van Tulder et al., 1997). It is important to inform the patient about the benign nature of the syndrome, the need to avoid bed rest and to encourage the patient to maintain normal daily activities (Malmivaara et al., 1995). In acute benign pain it is crucial that the diagnosis and interpretation of the radiological findings are undertaken to ensure that the patients do not progress to illness behaviour and fear-avoidance (Koes et al., 2001), which are often complicating factors promoting chronic LBP and disability.

In prolonged pain, activating procedures are recommended (van Tulder, 1997), and in the case of radiating pain surgeon should be consulted at the latest after six weeks of observation (Deyo and Weinstein, 2001). The patient should be encouraged to exercise and make active use of the back. Spinal manipulation and physical therapy are potential alternatives to treatment (Deyo and Weinstein, 2001). If not started before, then an active rehabilitation process including guided and self-motivated exercises should be added to the treatment of pain lasting over six weeks. In chronic pain, a comprehensive evaluation of the condition and an intensive integrated rehabilitation should be undertaken (Bigos et al., 1994; Guzman et al., 2002).

The good prognosis of the acute pain should be emphasised from the beginning but also the patient should be informed about the frequent recurrence of the pain. Only five to 10 percent of the cases become chronic, lasting over three months but those cases result in the most severe medical and economical hardship.

Disc herniation

Pathophysiology
In the herniated disc, the nucleus pulposus has gone through the ruptured annulus fibrosus. Disc herniation causes mechanical compression of the nerve roots and chemical irritation mainly by activation of the inflammatory processes (Kuslich et al., 1991, Olmarker et al., 1993). The tumor necrosis factor-alpha (Olmarker and Larsson, 1998, Olmarker and Rydevik, 2001) and several other cytokines (Watkins and Maier, 2002) appear to be clearly associated with the inflammatory process in sciatica. The use of specific anti-TNF- treatment may alter the treatment strategy of acute sciatica in the future (Watkins and Maier, 2002; Karppinen et al., 2003).

Symptoms and findings
The disc rupture itself can be painful and, in addition to the local LBP, they can also mimic the radicular symptoms evoked by a herniated disc (Karppinen et al., 2001). The radicular symptoms include pain, numbness, sensory disturbances, paresthesia and motor weakness. In severe cases (cauda equina syndrome) urinary retention, anal incontinence or sensory loss in a saddle distribution may be present (Frymoyer, 1988). The disc herniation is not visible in the plain radiograph but can be detected by MRI (Herzog et al., 1995) or optionally by CT.

Treatment and prognosis
The radicular symptoms usually disappear in three months by conservative treatment. In those cases with persistant severe symptoms over six weeks with a clear radiological finding, surgery is usually considered. Intolerable pain, cauda equina syndrome or acute paresis/paralysis may require emergency surgery (Kostuik et al., 1986; Bigos et al., 1994). Due to the good spontaneous recovery, only approximately ten percent of the patients need to undergo surgery (Frymoyer, 1988), however, the result of surgery is better if it is initiated within three months from the beginning of the symptoms (Hurme and Alaranta, 1987). One year after surgery 50-85% of the operated patients are free of sciatica (Hoffman et al., 1993; Junge et al., 1995). However, there is a considerable recurrence of the syndrome. The need for re-operation may be up to 20% during 13 years follow-up (Nykvist et al., 1995), according to Finnish Hospital Discharge Register 14% of patients needed re-operation during a 10-year follow-up (Österman et al., 2003).

According to an early randomised trial comparing surgery and conservative treatment, the surgery was more effective for two years follow-up, but from two until ten years the results were statistically equivalent (Weber, 1983), though a later reanalysis has found a slight benefit favouring surgery also during a longer follow-up. The recent RCT showed faster recovery after surgery but only minor differences between conservative care and surgery at two-year follow-up (Österman et al., 2002). The current conservative treatment includes pain medication and patient information. Bed rest should be avoided and the patient should be encouraged to resume normal daily activities (within the limits imposed by the pain) as in non-specific LBP. New promising treatments, such as therapy with TNF-alpha antibodies, are also on the horizon (Karppinen et al., 2003).

Lumbar spinal stenosis
LSS is a notable degenerative disorder of the lumbar spine, responsible for low back and lower extremity pain. The symptoms of the disorder arise from compression of the cauda equina due to degenerative processes leading to anatomical stenosis of the lumbar spinal canal and intervertebral foramina (Verbiest, 1954).

Pathophysiology
The anatomical narrowness of spinal canal is the basic aetiology for LSS but the syndrome is caused by degeneration of lumbar spine leading to facet arthrosis and disc degeneration with osteophytes and thickening and calcification of ligamentum flavum (Rauschning, 1993; Fritz et al., 1998; Spivak, 1998). The pathophysiology of the symptoms is poorly known, but it is thought that the mechanical compression irritates the nerve roots and causes the symptoms of the disease. However, the symptoms do not clearly correlate with the degree of the stenosis, and there is a considerable fluctuation in the symptoms (Amundsen et al., 1995; Jönsson et al., 1997). One potential explanation is vascular compression theory based on the venous congestion caused by the mechanical compression of the cauda equina veins, leading to decreased blood flow and subsequent dysfunction of the lumbar nerve roots (Porter, 2000). Recent experimental studies have supported that theory by indicating ectopic firing caused by venous stasis of lumbar nerve roots and inhibition of ectopic firing after decompression (Ikawa et al., 2003).

Symptoms and findings
The symptoms include LBP, intermittent claudication and radicular symptoms such as pain, numbness, weakness and loss of sensation in the lower limbs. Claudication and radicular symptoms are often provocated by spine extension and revealed by flexion and the claudication eases slowly, when the subject stops walking (Spivak, 1998). The plain radiograph can indicate LSS but the diagnosis is based on MRI or CT findings correlating with the symptoms (Katz et al., 1994).

Treatment and prognosis
After diagnosis of LSS, the options are follow-up, conservative treatment or surgery. The mild symptoms are usually treated conservatively, but severe stenosis often requires decompressive surgery. Conservative treatment includes analgesic medication, physical therapy, spinal support and calcitonin therapy (Eskola et al., 1992), however, there is still need for the RCT based evidence on the effect of conservative treatment modalities. If the symptoms and disability are intolerable despite conservative treatment, then surgery is indicated. The symptoms are eased, and the extent of disability decreased after surgery (Atlas et al., 2000), but the long-term outcome shows large variation (Herno, 1995; Hurri et al., 1998). The natural course of LSS is unclear but according to a recent RCT, surgery seems to be better than conservative care in moderately severe LSS at one-year follow-up (Slätis et al., 2002).

Motor control

Classification of motor control to voluntary and automatic actions
Human motor control includes schematically three different co-operating levels (Figure 2). The highest level is planning of the movement, the lowest level is spinal reflexes, and the intermediate level connects the highest and lowest levels for the execution of the task (Kalaska and Drew, 1993). The terms voluntary and reflex are widely used in scientific papers and in everyday conversation, but the exact meaning of these terms is not clear and need some definition (Prochazka et al., 2000). This study defines focused attention encompassing conscious (voluntary) control and subsidiary activation encompassing automatic motor programs (including reflexes). Isolated voluntary and automatic actions are rarely seen in real life, the normal functions are combinations of different levels of control. Reflexes assist in voluntary functions and cortical activation may regulate reflexes. The primary motor cortex is involved in reflex inhibition via the descending pathways, which vary their activity level according to the attention, arousal and emotional state of the subject.

Focused attention
Voluntary movements are programmed in the brain. Several cortical areas e.g. supplementary motor cortex, basal ganglia, cerebellum, thalamus, anterior gyrus cinguli and primary and supplementary somatosensory cortex are linked to the primary motor cortex. These centres are involved in the planning of the movement. The final command to perform the motion is initiated by the pyramidal cells of primary motor cortex and transmitted through the corticospinal pathway to the ventral horn of the spinal cord. At a specific spinal level, the upper motor neuron synapses with the alpha motor neuron which then transmits the command to muscle fibers of its own motor unit (Rothwell, 1994).

Subsidiary movement
The subsidiary muscle activations are often associated with voluntary movements and are function of servo actions needed e.g. in postural control (Rothwell, 1994). These actions include anticipatory postural adjustments and are usually automatic i.e. unconsciously controlled but are more complex than simple reflexes. The responsible level controlling these fast actions is intermediate, also called the extrapyramidal system. In addition to thalamus and basal nerve nuclei, this involve the cerebellum and basal ganglia, which are involved in the tuning of the fine movements, coordination, timing, motor learning and muscle tone.

Automatic motor programs
Reflexes are the fastest way to control movements. They are categorically divided into mono-, di- and polysynaptic reflexes. The monosynaptic reflex is the simplest and fastest of the three reflex types. The electrical reflex latency varies from ~10 ms in spinal muscles to ~50 ms in distal lower limb muscles depending on the distance of respective muscle from the spinal cord or cranial nerve nuclei. The latency consists of the sensory impulse conduction in sensory neurons, synaptic transmission and motor impulse conduction in the motor neurons (Rothwell, 1994; Schmidt and Lee, 1999). In addition, there is a notable time lag from the electrical activation of the muscle cells to the mechanical force produced by the muscle (Rothwell, 1994; Schmidt and Lee, 1999). In complex responses, the relative time for information processing is longer.

In the simple monosynaptic stretch reflex, muscle spindles sense the stretching of the muscle fibres. The information is mediated by the Ia sensory afferent nerves into the dorsal horn at the respective level of the spinal cord. These sensory nerve cells synapse with the motor neurons activating the corresponding muscle. In more complex reflexes, an increasing number of interneurons are involved in mediating the reflex (Rothwell, 1994). The muscle activation is modulated also by the gamma motor neuron system. These types of reflexes are usuall involved in voluntary and subsidiary movements e.g. permitting the fluency of movements by controlling the agonist-antagonist muscle activation.

Postural control
The equilibrium of the body is essential for locomotion and performing other types of limb movements. The purpose of the postural control system is to support, stabilize and balance with the aim of maintaining postural stability. Postural control functions at three tightly connected levels i.e. postural reflexes, triggered reactions and voluntary movements these being listed from fastest to slowest level of action and from the lowest to highest level of cognition, respectively. The somatosensory, vestibular and visual receptors provide the sensory information needed in this process (Rothwell, 1994). The sensory information is transmitted by the ascending pathways of the spinal cord and the motor commands are transmitted by the descending pathways which synapse with the lower motor neurons. The major descending pathways controlling body posture are the vestibulospinal tracts. The neurons of the vestibulospinal tracts synapse with the same alpha motor neuron as the corticospinal tract (Rothwell, 1994).

Postural responses are modulated by feedback information but also pre-programmed functions are needed (Schmidt and Lee, 1999; Hodges, 2003). The cerebellum is important in maintaining the balance and adaptation and learning of postural reflexes and feed-forward responses (Horak and Diener, 1994; Rothwell, 1994). Postural control is a complex procedure and vulnerable to disruption by a wide variety of disorders (Horak and Nashner, 1986; Rothwell, 1994). In addition to the neurological and vestibular diseases, postural control plays also a significant role in several musculo-skeletal impairments including back and joint disorders (Alaranta et al., 1994). Increasing age starts to have detrimental effects on postural control after the fifth decade of life (Alaranta et al., 1994; Aalto, 1997).

Motor control of the lumbar spine
Lumbar spinal anatomy has been described in detail by Bogduk and Twomey (1991) and reviewed in the association with LBP by Indahl (1999) and Kankaanpää (1999). According to Panjabi (1992), the spinal stability system consists of three subsystems, which are the passive spinal column, the active spinal muscles and the neural control unit. An overloading or dysfunction of any of these subsystems may lead to injury if there is a failure of compensation mechanisms (Panjabi, 1992). Intra-abdominal pressure, diaphragm, pelvic floor and abdominal muscle activation have been shown to have a significant role in spinal control (Hodges, 2003).

The simple stability and instability concept in the low back syndrome has recently been challenged and the importance of neural control system has been emphasised. Hodges (2003) has shown that CNS does not simply stiffen the spine and restrict the spinal motion, but actively uses movements to maintain equilibrium in the posture. Lund (2003) has indicated that mechanical LBP is associated with restricted but painful motion rather than mechanical instability. This indicates that a potential pathomechanism of "the instability" is dysfunction of neuromuscular control system.

Segmental function of the lumbar spine
Two vertebrae, their intervertebral disc and facet joints form the functional spinal unit (FSU, Pope et al., 1993). The ligamento-muscular reflex between multifidus muscles, zygapophysial joints and intervertebral ligaments has been demonstrated in the control of the intricate neuromuscular balance in the lumbar motion segment (Solomonow et al., 1998; Indahl, 1999). The possible interactive responses between injured or diseased spinal structures, i.e., disc or facet joints, and the paraspinal musculature may lead to segmental dysfunction of the lumbar spine (Sihvonen et al., 1991; Indahl et al., 1995; 1997; Kaigle et al., 1998).

Kinetic chain
Thoracolumbar fascia connects the spine, upper limbs and pelvis and via the sacrotuberous ligament the pelvis to the lower limbs. This allows dynamic load transfer from spine to legs (Vleeming et al., 1995). Paraspinal, gluteal and hamstring muscles exhibit sequential activation during sagittal trunk flexion and extension (Paquet et al., 1994; Leinonen et al., 2000). Dysfunction of any part of the integrated system may lead to abnormal load transfer between low back and pelvis. Therefore, it seems rational that the hip extensor muscles, in addition to the spinal and abdominal musculature, will be subjected to deconditioning in CLBP patients, if the use of these muscles is avoided (Kankaanpää et al., 1998).

Feedback control
The appropriate proprioceptive information is essential for motor control. In unpredictable postural perturbation, the control strategy is based on sensory information transmitted from the receptive structures and during intended movements, feedback information is needed for error correction (Schmidt and Lee, 1999). The proprioceptive information of the body and limb movements originate from the muscle spindles, the Golgi tendon organs, joint and the cutaneous receptors (Rothwell, 1994; Schmidt and Lee, 1999). Muscle receptors seem to play a major role in joint position sense (McCloskey, 1978; Pedersen et al., 1998). The increased repositioning error after paraspinal muscle vibration and impaired lumbar position sense after lumbar paraspinal muscle fatigue emphasize the importance of muscle spindles in the positioning of lumbosacral spine (Brumagne et al., 1999; 2000; Taimela et al., 1999).

Tactile sensations and conscious proprioceptive information about the body are mediated to the somatosensory cortex via the dorsal columns, the fasciculus gracilis from lower limbs and the lower part of the body and by the fasciculus cuneatus from the upper limbs and the upper part of the body. The dorsal spinocerebellar and probably also the spinothalamic tracts transmit the proprioceptive information needed in the maintenance of postural stability (Rothwell, 1994). The precise neurophysiological mechanisms processing the proprioceptive input in conscious sensory perception remain to be determined but probably involve coordinated inputs from cerebellum, thalamus and somatosensory cortex.

Feed-forward control
In predictable postural perturbation, the CNS can plan control strategies in advance. According to the recognition schema theory, muscle activation can be pre-programmed based on initial condition, environmental outcome and expected sensory consequences (Schmidt, 1975; Schmidt and Lee, 1999).

Upper (Cordo and Nashner, 1982, Zattara and Bouisset, 1988, Hodges and Richardson, 1996) or lower (Hodges and Richardson, 1998) limb voluntary movements evoke non-conscious muscle activation in the trunk muscles via a feed-forward mechanism. This refers to the activities of the central movement control system, which maintains postural stability and prepares the trunk to bear a potentially increasing load by activating certain trunk muscles. These trunk muscles which maintain the dynamic spine stability, are activated before the activation of the prime muscles responsible for the gross limb movement without an afferent input from the respective trunk movement (Belen'kii et al., 1967; Cordo and Nashner, 1982; Friedli et al., 1988; Zattara and Bouisset, 1988; Aruin and Latash, 1995; Hodges et al., 1999).

Measurements of lumbar function
The clinical examination include inspection, palpation, range of spinal motion, testing of lower limb sensory function, muscles and reflexes. These examinations with the functional and provocation tests form the basics of the evaluation of lumbar function in LBP (Deyo and Weinstein, 2001). The test results are usually non-specific and evaluated subjectively, but their repeatability is acceptable and they are very useful, especially when they are correlating with the subjective symptoms. Neurophysiologic assessments include neurography, electromyographic examination and motor and somatosensory evoked potential measurements. They are used in the assessment of nerve injury and in the diagnosis of the level of nerve injury (Dvorak et al., 2000).

Several functional measurements can be used in the study of LBP. Surface electromyography measurements are used in the assessment of muscle function (Sihvonen, 1995) and endurance (Kankaanpää, 1999). Muscle strength (Rantanen, 2001) and lumbar kinematics are measured with a wide variety of methods (Lund, 2003). Postural control can be measured on a force-platform either with or without simultaneous EMG and trunk motion measurements (Luoto, 1999). They have been mainly used in assessing physiological and pathophysiological mechanisms of lumbar function (Hodges, 2003). The test results usually overlap with values obtained from back healthy subjects. Therefore they are not of any great benefit in the diagnosis but can be used for example in the planning of rehabilitation procedures and in the objective evaluation of the effectiveness of the rehabilitation. The diagnostic value of the functional measurement is often rather poor and their clinical use is also hindered by a lack of standardisation.

Pain and sensory-motor control

Pain and proprioception
Impaired lumbar proprioception is associated with LBP (Taimela et al., 1999). An increased repositioning error during trunk movement has been observed in LBP patients (Gill and Callaghan, 1998; Brumagne et al., 2000; Newcomer et al., 2000). Paraspinal muscle spindles seem to be important in the correct positioning of the lumbosacral spine and the muscle spindle input may be decreased in lumbar pain (Brumagne et al., 1999; Taimela et al., 1999). In addition, experimental muscle pain has been shown to affect the central modulation of proprioceptive signals of jaw muscle spindles (Capra and Ro, 2000). However, there is experimental evidence that the nociceptive input can actually enhance the central sensitivity to the mechanoreceptor input (Torebjörk et al., 1992).

Pain and trunk muscle function
Pain has clear effects on motor control. LBP can induce changes in neuromuscular control and motor performance. The function and co-ordination of the muscles stabilising the lumbar spine seem to be impaired in patients suffering from LBP (Hodges et al., 1996; Wilder et al., 1996). Delayed or absent trunk muscle activation has been observed in low back pain patients during upper (Hodges and Richardson, 1996) and lower limb movements (Hodges and Richardson, 1998) and after experimental pain (Hodges et al., 2003). A different muscle response pattern to sudden load release (Radebold et al., 2000; 2001) and during expected and unexpected upper limb and trunk loading (Wilder et al., 1996; Magnusson et al., 1996) has been found in CLBP patients compared with healthy controls.

Psychosocial stress can change spinal muscle activation and loading. However, notable individual differences have been observed in its importance (Marras et al., 2000; Davis et al., 2002). CLBP patients have been claimed to exhibit an increased paraspinal (Roy et al., 1989) and gluteal (Kankaanpää et al., 1998) muscle fatigability. During trunk flexion-extension, CLBP patients had increased paraspinal muscle activity when acting as antagonist (Ahern et al., 1988; Sihvonen et al., 1991; Sihvonen, 1995) and decreased activity when acting as agonist (Sihvonen et al., 1991), with similar activation patterns being observed during experimental pain (Zedka et al., 1999). The CLBP and experimental muscle pain induced increased paraspinal muscle activity in the normally silent period and decreased activity in the normally active period of the gait (Arendt-Nielsen et al., 1996).

LBP and general motor control
The impaired postural control is associated with chronic lumbar disorders (Byl and Sinnot, 1991; Luoto et al., 1996). A slow psychomotor reaction time is also associated with back pain (Taimela et al., 1993). Poor postural and psychomotor control (Taimela, 1992) is a known risk factor for accidental injuries, but in lumbar disorders it is not yet known whether the poor postural control is a consequence of pain or neuromuscular impairment or a potential major source of the disorder. Thus, there appears to be little prospective evidence on whether the impaired motor control is pre-existing (risk factor) or a secondary phenomenon (consequence) of the musculoskeletal disorder. The muscle dysfunction and deconditioning seem to be rather a consequence of the tissue injury and pain. However, this claim needs further prospective evaluation.

Pain seems to decrease the muscle activation amplitude during voluntary contractions and increase the muscle activation during automatic contractions (Zedka et al., 1999). One potential explanation for this is the hypothesis based on the model of fear-avoidance behaviour, where the decrease in central output is revealed as decreased primary motor activation and decreased inhibition of reflexes.

Active physical rehabilitation has been claimed to restore the impaired neuromuscular functions (Wilder et al., 1996; Kankaanpää et al., 1999). However, impaired postural control failed to improve after rehabilitation and after unsuccessful treatment the condition may even deteriorate (Luoto et al., 1998). Despite the widespread acceptance of the importance of stabilising functions of the spine, there is still a need of evidence based on randomised controlled trials on the effect of stabilising exercises in LBP.

In conclusion extensive variety of changes in motor and sensory control and trunk muscle function have been reported. However, they occurrence and clinical relevance in LBP with different pathophysiological causes remains undefined. In addition, the time-course of the changes and the effect of interventions are rarely evaluated.

Patients with lumbar spinal stenosis
The study included 26 LSS patients (11 men, 15 women; study 4, 5). The diagnosis was based on the patients' symptoms and signs and was confirmed by computed tomography (CT) or magnetic resonance imaging (MRI).

Assessment of paraspinal muscle function

Surface EMG
Bipolar surface electromyography (EMG) was recorded bilaterally over the paraspinal muscles at T12-L1 and L5-S1 levels by a four channel ME 3000P EMG system (Mega Electronics Ltd, Kuopio, Finland) with disposable Ag/AgCl surface electrodes (Medicotest, Olstykke, Denmark). The electrodes were placed on the erector spinae (ES, T12-L1 level) and multifidus (MF, L5-S1 level) muscles as recommended by Biedermann et al., (1991). The raw EMG signal was recorded at the sampling rate of 2 kHz and band-pass filtered between 7-500 Hz with an analogue filter, amplified (differential amplifier, CMRR>110 dB, gain 1000, noise<1µV), analogue-to-digital converted (12-bit), and stored in a personal computer for later analysis. (study 1-3, 5).