Low Back Pain (LBP) is one of the major concerns of current health care [^1-5]. Only 10% of LBP cases can be attributed to specific disorders like nerve root compression, vertebral fracture, tumour, infection, inflammatory diseases, spondylolisthesis or spinal stenosis. Consequently, NSLBP, in which the cause of symptoms is unknown, is diagnosed in about 90% of all patients and is a health problem of high economic importance [⁶].

Patients with NSLBP present diverse clinical findings, courses of disease and prognoses. They therefore make up a heterogeneous group of patients, which may explain why treatment effects in numerous studies looking at specific physiotherapy treatments in the NSLBP group are often discouraging. Identifying defined subgroups of patients within the NSLBP population has been a major focus in recent research [^7-12]. Current European guidelines encourage outcome studies in subgroups of patients with a shared diagnostic pattern or prognosis that might benefit from specific treatments [¹]. This research agenda is expected to reveal further evidence for the effect of treatments designed for specific subgroups [¹³]. Of 767 RCTs about the effect of conservative treatment on chronic LBP performed and published between 1982 and 2008, 68 publications examined manual or exercise therapy, of which five studies had an additional subclassification and matched treatments [¹⁴].

There is no evidence for the effectiveness of exercise in patients with acute LBP of a less than 6 week duration [¹,^15-17], but exercise therapy is effective in chronic and subacute LBP [¹⁶]. There is evidence that home exercise may be effective in decreasing pain and disability, but results have shown no significant difference between exercise types on work disability [¹,¹⁸]. There is moderate evidence suggesting that exercise therapy may prevent recurrences of LBP, but there is no evidence for a difference in effect between types of exercise [¹⁹]. Individually designed exercise programs are recommended [¹⁷] but the question remains as to which types of exercise are effective for which subgroups of patients.

Within the framework of a new NSLBP classification system developed by O'Sullivan, one of the subgroups of patients that can be distinguished contains those suffering from MCI [¹¹]. In a first step, this classification of NSLBP distinguishes between patients with non-mechanical disorders, and patients with mechanical disorders. Whereas in patients with non-mechanical NSLBP, psychosocial factors, fear and catastrophising play central roles, pain in relation to posture and movement is predominant in patients with mechanical NSLBP. Patients with mechanical NSLBP are further divided into those with movement impairment (MI) and movement control impairment (MCI). Patients with MI may suffer movement restrictions in single or multiple directions. MCI is defined as a deficit in the control of movements during functional daily activities. The range of the movement is not restricted in the MCI group.

Clinical tests to identify MCI were developed in recent research [¹¹,^20-26]. Further evaluation revealed six tests which reliably detect MCI in patients [²⁷]. (Additional files 1) These tests will be used to select patients with MCI for this study. Movement control tests are easy to perform in clinical practice. Tests and clinical presentation allow a further classification of movement control dysfunctions according to direction in extension, flexion, frontal plane and multi-directional MCIs. (Table 1)

Two mechanisms are proposed to explain the impaired movement control behaviour. One relates to conditioning and habituation, which are important factors in motor learning. Patients use postures and movements that are potentially harmful due to maladaptive processes, like avoidance or overuse during the acute pain phase [²⁸]. Another mechanism is non-awareness of the posture's pain provocation. Altered cortical representation of the lumbar spine in the presence of pain may play an important role [²⁹]. Two point discrimination is decreased in patients with NSLBP, indicating changes in cortical representation because of NSLBP [³⁰]. Both mechanisms can either be induced by pain or be the cause of pain.

It is hypothesised that, once movement control is impaired, it results in repetitive mechanical deformation of innervated tissue and leads to increased nociceptive input to the central nervous system and, therefore, pain [³¹]. All joint capsules, ligaments, tendons and muscles are possible pain sources, especially due to continuous strain or longstanding repetitive movements [³¹,³²]. Repeated misuse of these tissues can also initiate the inflammatory cascade, a further cause of pain [³³].

The inter-rater reliability of the clinical classification of NSLBP in MI and MCI is very high; in experts and raters with less experience, k = 0.85 and k = 0.6 respectively [³⁴,³⁵]. The inter- and intra-rater reliability of the previously described 6 active tests performed by the patient was evaluated as good to substantial [²⁷]. Validity of MCI tests is supported by significantly different results in healthy subjects, with 0.75 positive MCI tests (95% CI 0.55-0.95) and low back pain patients with 2.21 positive MCI tests (95%CI 1.94-2.48, effect size d = 1.18) [³⁶].

In a preceding prior controlled case series, 38 preselected patients with positive MCI tests were treated with an individualised movement control exercise program and showed improvements in MCI test performance associated with improvements in patient specific functional complaints and disability [³⁷]. The present study will compare movement control training with general exercise in a randomised controlled trial.

MCI exercises are often and falsely referred to as motor control exercises, spinal stabilisation or core stability exercises. While MCI exercises aim to improve function through repetitive normal use, the latter retrain delayed muscle activity first in order to improve control of the spine. Several trials and systematic reviews have evaluated the effect of stabilising exercises with conflicting results [^38-42]. A recent randomised placebo-controlled trial with a specific stabilisation program in a population with chronic low back pain found a significant improvement in activity in both the short and long term when measured with a patient specific functional scale (PSFS), and a short term improvement for global impression of recovery and disability [⁴³]. However, they did not subgroup the patients and the effects were not beyond the smallest minimal meaningful clinical change.

The general exercise treatment is based on treatments used in previous studies. Exercises were developed to improve endurance, strength and flexibility of the spinal tissues [⁴⁴]. Experimental studies have shown that these exercises involve many global and local muscles and improve the stability of the spine [⁴⁵,⁴⁶]. In an RCT of patients with subacute and chronic NSLBP, a course of only general exercise was compared with a combination of general exercise and stabilisation exercise. Immediately after treatment, disability was significantly lower in the general exercise group. No differences were found in other outcomes or at follow-up time points [⁴⁷].

Psychosocial factors have an important impact on treatment outcome in NSLBP patients who also show non-acute [⁴⁸] depression, pain catastrophising, fear of pain and avoidance. These aspects are part of the fear-avoidance model of chronic pain, and previous data has reported its various effects on treatment outcome [⁴⁹]. Pain-related fear and subsequent avoidance of movements that patients believe to be harmful can also lead to disability and physical deconditioning [⁵⁰]. In our study patients with a high risk of psychosocial problems are excluded based on an assessment with the Örebrö Musculoskeletal Pain Questionnaire (ÖMPQ) [⁵¹,⁵²]. This influential variable will be observed using the Fear-avoidance beliefs questionnaire.

This randomised controlled trial will, to our knowledge, be the first to compare two exercise programs for a well-defined subgroup of patients with non-specific low back pain and movement control impairment. We will evaluate the effect of individualised movement control exercise versus general exercise on disability during a one year follow-up period.

Inclusion of patients began in July 2010 and is expected to last until the end of 2011. Results are expected in 2013.

This randomised controlled trial will compare the effectiveness of two exercise-based physiotherapy treatment protocols in a well-defined subgroup of NSLP. Inclusion is based on the clinical diagnosis of MCI with clinical tests shown to be reliable. The MCI tests allow easily applicable selection of participants. We will evaluate whether a treatment protocol addressing movement control problems is more beneficial than a general exercise protocol.

The multicentre design of the study allows treatment by different physiotherapists and improves generalisability. The selection of patients reflects the population usually found in different clinical settings. Randomisation is organised centrally and prevents selection bias.

Both treatment protocols are widely used and well-established in physiotherapy. The physiotherapists treating each group are equally instructed and experienced in applying the respective treatments. This prevents a disadvantage for the participating patients regarding group allocation.

Blinding the therapists is generally not possible. To minimise measurement bias, the measurements taken after the treatment will be taken by an assessor not involved in the treatment procedures.

The results of this study will provide evidence to improve the selection of exercise treatments for patients with NSLBP and MCI. Results will also contribute to the understanding of the mechanisms behind MCI and its relation to low back pain.

CPG: Chronic pain grade; GCPS: Graded chronic pain scale; MCI: Movement control impairment; MI: Movement impairment; NSLBP: Non-specific low back pain; ÖMPQ: Örebrö musculoskeletal pain questionnaire; PSFS: Patient specific functional scale; RCT: Randomised controlled trial; RMQ: Roland Morris disability questionnaire; TPD: Two point discrimination

The authors declare that they have no competing interests.

HL originated the idea of the study, which is based on his previous work. HL, JK, JS, JMS designed the trial protocol. JS drafted the manuscript and the other authors revised it critically, corrected draft versions and approved the final manuscript.

The pre-publication history for this paper can be accessed here:

http://www.biomedcentral.com/1471-2474/12/207/prepub

The authors would like to acknowledge the main input of the physiotherapists and doctors in clinics and practices in Switzerland, which are: Bethesda Spital, Basel; Kantonsspital Winterthur, Winterthur; Klinik Schulthess, Zurich; Medbase Physiotherapie, Winterthur; Physiotherapie Seen, Winterthur; Physiotherapie Bellaria, Zurich; Physiotherapie Erlenbach, Erlenbach; Physiotherapie Reinach, Reinach; Physiotherapie Würenlingen, Würenlingen; Provital Physiotherapie, Egg; Rheumaklinik, Universitätsspital Zurich, Zurich; Segeten Physiotherapie, Zurich; Uniklinik Balgrist, Zurich.

We'd like to express our gratitude to Dr. med. Andreas Klipstein and Dr. med. Michael Gengenbacher for their contribution in getting ethical approval.

Further we would like to acknowledge Caroline HG Bastiaenen for her input to the research design, and Daryl Snell for his assistance in preparing the final manuscript.

The project was supported by Swiss National Science Foundation (SNSF). (Project no. 127240) HL, JK, JS are funded by SNSF; JMS and RAB are funded by CAPHRI School for Public Health and Primary Care, Maastricht University.