Overview of research design
The study will be a randomised controlled trial comparing graded activity and motor control exercise for patients with chronic low back pain. Each exercise program will consist of 12 individually supervised hour sessions over an eight-week period, 2 additional follow-up treatments at 4 and 10 months and a home exercise program. Outcomes will be measured at 2 months, 6 months and 12 months from baseline. Putative predictor variables will be taken from the baseline clinical assessments. The study design, procedures and informed consent were approved by The University of Sydney Human Research Ethics Committee.
i) That motor control exercise will be more effective in improving pain and disability at 2, 6 and 12 months.
ii) The effect of graded activity (versus motor control) is greater in patients who are deconditioned and have negative beliefs about their pain.
iii) The effect of motor control exercise (versus graded activity) is greater in patients with impaired control of movement and stability of the spine.
A total of 172 subjects will be recruited from general practitioners, physiotherapy practices and public hospitals in Sydney and Brisbane. Contact has been made with these sites and consent letters have been obtained. Subjects will be screened to identify those who are unsuitable for exercise management of their low back pain because of significant co-morbidity such as serious spinal pathology or contraindication to exercise. To screen for serious pathology, the physiotherapist will conduct a diagnostic triage following the Royal College of General Practitioners' low back pain guidelines [19
] and will be screened for contraindications to exercise as listed in the ACSM guidelines [20
Subjects will be included if they meet all of the following inclusion criteria:
• Non-specific low back pain +/- leg pain persisting for at least 3 months duration
• Currently seeking care for low back pain
• Aged more than 18, less than 80 years
• English speaker
• Clinical assessment indicates that the subject is suitable for active exercises
• Expects to continue residing in the Sydney or Brisbane region for study duration
• Obtains a score of moderate or greater on question 7 or 8 of the SF-36.
(Question 7 – How much bodily pain have you had during the past week? None, very mild, mild, moderate, severe, very severe).
– During the past week, how much did pain interfere with your normal work, including both work outside the home and housework? Not at all, a little bit, moderately, quite a bit, extremely) [21
Subjects will be excluded if they have any of the following:
• Known or suspected serious spinal pathology (fracture, metastatic, inflammatory or infective diseases of the spine, cauda equina syndrome/widespread neurological disorder).
• Nerve root compromise (at least 2 of the following signs: weakness/reflex changes/sensation loss, associated with the same spinal nerve)
• Previous spinal surgery or scheduled for major surgery during the treatment follow-up period
• Co-morbid health conditions that would prevent active participation in the exercise programs.
Specific spinal pathology or contraindication to treatment may be suspected based upon the results of the screening questionnaire and the Physical Activity Readiness Questionnaire. If the assessor suspects the presence of any pathology or contraindication to treatment, these subjects should be further investigated by a registered physiotherapist and medical clearance obtained if necessary.
Subjects receiving workers compensation will only be included in the trial if they have written consent from the nominated treating doctor, insurer and employer agreeing to their participation in the trial. The letters should state that they agree with the condition that only the trial treatments will be administered to the patient and acknowledges that the patient will be instructed not to seek other treatments during the trial treatment period.
Assessment and Allocation
Participants will be allocated to treatment group using sealed opaque envelopes. Physiotherapists will allocate patients by drawing the next consecutive envelope allocated to their practice. A sticker containing the letter M (motor control) or G (graded activity) will be inside the envelope. After reading the sticker it will be attached to the patient's file. The therapist will not reveal the patient's allocation until the end of the study.
Following the screening consultation, personal characteristics (age, gender, ethnicity, religion, weight, height. level of education, employment status, doctor's details and contact information) and information about symptoms of low back pain will be collected. The following treatment outcomes will be measured at baseline, 2 months, 6 months and 12 months.
Primary Outcome (at 2 and 6 months)
1. Average pain intensity over the last week (0–10 scale) [22
2. Patient-generated measure of disability (Patient-Specific Functional Scale) [22
3. Patient's global impression of change (Global Change) [22
4. Condition-specific measure of disability (Roland Morris Disability Questionnaire) [24
5. Generic measure of health status (SF-36 Version I) [21
In addition subjects will be asked about their pain (average for preceding 24 hours and their average pain in the preceding week) once a month during one year. This data will be collected using a SMS (phone message) system. A SMS message will be sent monthly to all subjects reminding them to text the researcher with a number from 0–10 representing their average pain intensity over the preceding 24 hours and a number representing their average pain intensity over the preceding week. If subjects do not have a mobile phone they will be e-mailed monthly and asked to supply their average pain intensity score via return email. For patients that do not have a mobile phone or access to email, the average pain intensity will be recorded by phone. At the end of the study all participants will receive $10.00 to cover the cost of text messaging. A summary measure of the pain experienced over the 12-month period is provided by the area under the curve of the pain intensity/time curve. These measurements may better reflect the fluctuating nature of chronic pain than is possible with traditional outcome measures taken at a single time point. As the measures are novel the measurement properties are unclear and so we will regard this as a secondary outcome.
To maximise attendance at follow-ups subjects will receive a reminder letter and a phone call 24 hrs prior to their appointment. Every attempt (within ethical constraints) will be made to obtain outcome data, regardless of subject's compliance with trial protocols. Follow-up measures will be scored by an investigator who is blinded to group allocation. At 2 months, information about side effects of treatment will be collected from all subjects using open-ended questioning. At each follow-up, information on use of other pain-related treatments will be recorded as will the patient's compliance with their home exercise program.
Treatment effect modifiers
The treatment rationale for the two types of exercise suggests potential treatment effect modifiers. The rationale for motor control exercise is that people with low back pain have poor control and coordination of the trunk muscles that results in control of movement and stability of the spine that is less than optimal [25
]. This approach is based on a large body of literature that shows changes in muscle recruitment, movement and stability of the spine in patients with low back and pelvic pain. There is evidence from in-vivo and modelling studies that suggests that the muscle control strategies adopted in low back pain have the potential to prolong and prevent recovery from low back pain [25
]. These deficits in control may vary from so called 'instability' to excessive stabilisation of the spine as a result of increased muscle activity. Motor control exercise uses a motor learning approach to re-establish normal control of the deep spinal muscles and then to coordinate the entire trunk muscle system during functional tasks. Logically this treatment is expected to work best in those who have impaired control of the spinal muscles, and in particular the deep muscles. Accordingly we will take measures that reflect poor motor control of the spine e.g. trunk proprioception, trunk stiffness, trunk muscle response, tests of deep muscle control
. The rationale for graded activity is that people with chronic low back pain are generally deconditioned as a result of injury or disuse and that this deconditioning underpins their chronic symptoms. Additionally many people with chronic low back pain have beliefs about pain and injury that are unhelpful to recovery. With graded activity the physiotherapist teaches the patient how to increase their activity. Logically this treatment should work best in patients who are deconditioned and have unhelpful beliefs about their back pain. Patients with high levels of habitual activity are less likely to benefit. Accordingly we will take measures of aerobic fitness, habitual activity level, kinesiophobia and self-efficacy
We will collect baseline data about clinical/demographic data, measures of beliefs and attitudes about pain, measures of physical activity and fitness and measures of control and coordination of the lumbar spine and pelvis.
LBP screening questionnaire [28
Measures of beliefs and attitudes about pain
PASS-20 (Pain Anxiety Symptom Scale) [29
Pain Self Efficacy Questionnaire [30
Measures of Physical activity and fitness
International Physical Activity Questionnaire (IPAQ) [31
Three minute step test [32
Measures of control and coordination of the lumbar spine and pelvis
Clinical lumbar spine instability test
As no in-vivo measures of "instability" have been validated assessment of instability is currently dependent on subjective data. A recent Delphi survey of international experts [33
] has developed 15 questions where positive subject responses indicate increasing likelihood of lumbar spine instability. These questions include report of feelings of giving way, and frequent bouts or episodes of symptoms [33
Test of trunk proprioception
Proprioception is a critical component of control and determines the effectiveness of control strategies. Although studies of repositioning have provided variable results [34
], recent evidence suggests that people with low back pain have significant errors in sense of effort [36
], which contributes to sense of position [37
]. The contribution of sense of effort to proprioception will be measured using a standardised technique in which subjects are asked to generate forces aiming to achieve fluctuating targets. This test is performed first with and then without feedback of force. Studies have shown that individuals with LBP have difficulties in matching the forces in the direction of trunk extension [36
Test of trunk stiffness
Active trunk stiffening by the trunk muscles can be evaluated in a dynamic task using an indirect measure of trunk stiffness and damping (Hodges et al, unpublished data). The response of the trunk to removal of a load is modelled as a second order system and resolved for stiffness, damping and mass. People with low back pain have been shown to have increased stiffness and decreased damping (Hodges et al, unpublished data)
Test of trunk muscle response
The response of the trunk muscles to removal of a load from the trunk has been shown to predict occurrence and recurrence of pain [38
]. This measure evaluates the net response of the trunk muscles to a perturbation [39
]. During a trunk perturbation test the onset and offset of the trunk flexors and extensors are measured using electromyography with surface electrodes over 12 muscles. People with LBP and those at risk for the development of low back pain have increased co-contraction of trunk flexors and extensors, and delayed offset of activity [38
Test of deep muscle control
This test is a novel measure of a subject's ability to activate the deep abdominal muscles. Until recently recruitment of the deep abdominal muscles could only be assessed with invasive EMG recording techniques that have limited application in large clinical trials. We have recently solved this problem by developing an ultrasound test [25
] that provides a non-invasive method to investigate the automatic recruitment of the trunk muscles in people with LBP in a clinical setting [25
]. Both EMG and ultrasound measures have a good ability to discriminate patients with low back pain from pain-free controls [26
We understand that the refined laboratory measures we propose may not all be viable in clinical practice. However it is important to use the best available measure to clearly establish the predictive ability of each factor. Equipped with this information we can then focus our attention on developing clinical versions of the tests that are shown to be important predictors.
Subjects in each group will receive 12 one-hour treatments over an 8 week period, i.e. 2 sessions/week in the first month and 1 session/week in the second month and 2 additional 1 hour follow-up sessions at 4 & 10 months. Patients will be advised to do home exercises for at least half an hour per week in the first month and one hour per week in the second month. However, the type of exercises, intensity and amount done per day will be performed at the discretion of the physiotherapists. The treatment session and home exercises will add up to 20 hours of treatment. In both exercise programs an individualised program for each patient is prescribed. The treatment sessions are designed to become less frequent over time to encourage independence. Patients with continued pain following all sessions will be asked to continue their home exercise program until pain-free for a week. If a recurrence occurs, re-initiation of the home program will be advised. This is consistent with current clinical practice.
• The graded activity
program is based upon the treatment approach reported by Lindstrom [40
] and identical to the protocol we used in a previous trial [13
]. At the first visit the physical demands of subjects' work and home activities will be established and measures of mobility, strength and aerobic fitness taken. From these assessments an individualised, submaximal, progressively incremented exercise program will be developed to train those functions found to be inadequate for performance of work or leisure activities. The goal is to improve the subject's ability to complete functional activities they specify as being difficult to perform because of the low back pain. Principles underlying cognitive-behavioural therapy will be used by the physiotherapists in their training and supervision role. These principles include the encouragement of skill acquisition by modeling the exercises, providing information, setting progressively raised goals, self-monitoring of progress, and verbally reinforcing progress made by subjects towards their goals. Fear of increased pain or even possible (re)injury will be addressed by discussion of the realistic chances of exercises causing injury, by ensuring that subjects set initial goals well within their capabilities (to maximise the chances of early success experiences), and by encouraging subjects to maintain a graded increase in exercises and other activities with regular rest breaks and frequent reflection on achievements.
Each subject will carry out a form of aerobic exercise (e.g. walking or cycling), stretches, functional activities, activities to build speed, endurance and coordination and trunk and limb strengthening exercises. Subjects will be given an individualised home exercise program that will be regularly evaluated by the physiotherapist.
• The motor control
exercise program is based upon the treatment approach reported by Richardson et al [27
], O'Sullivan et al [10
] and Moseley [41
] and similar to the protocol we used in an earlier trial [12
]. The approach utilises the principles of motor learning to retrain the optimal control and coordination of the lumbar spine and pelvis. Stage one involves regaining basic control strategies. This often involves training the pre-activation of the deeper muscles of the trunk such as transversus abdominis, multifidus, pelvic floor and diaphragm muscles that are typically affected in the presence of pain. The strategies that are selected are based on the patients presenting changes in coordination. Therapists will also assess and train posture and movement dysfunction (e.g. provocative movements and poor control strategies). Basic principles of motor learning such as simplification, segmentation, augmented feedback and practice are used to retrain the ideal activation of the deep muscles. Recent data suggests that these strategies restore control of the deep muscles [42
] and lead to persistence of the change after cessation of treatment [43
]. In this stage of training therapists will have access to a range of strategies that are typical in clinical practice including ultrasound imaging for feedback of muscle contraction, EMG biofeedback and other tools [27
]. Other targets of this stage include coordination of breathing and changes at adjacent segments (e.g. hip, thoracic spine). In stage 2 participants will be progressed though to more complex static and dynamic tasks, and training of functional activities. At all progressions the therapist evaluates and corrects trunk muscle recruitment strategies, posture, movement patterns and breathing. Session 12 is a discharge session where the patient's progress will be reviewed and patients will be prescribed exercises to continue at home.
A treatment demarcation table was developed to help the trial clinicians to identify salient features from each of the interventions and to avoid unintended contamination between the two treatment approaches (see Table ).
Subjects in each treatment group will be asked not to seek other treatments for their chronic low back pain and where possible not to change current medications during the treatment period. Several mechanisms will be used to ensure that the trial protocol is consistently applied. Protocol manuals have been developed and staff will be trained to ensure that screening, assessment, randomisation and treatment procedures are conducted according to protocol. To ensure standardisation we will hold regular meetings with site visits. An independent researcher will monitor a randomly chosen subset to ensure adherence to assessment, randomisation and treatment procedures. Medical practitioners will be asked not to request the subject's allocation unless it is deemed necessary for medical care. At the completion of the exercise program, patients will be encouraged to continue the home exercise routine demonstrated at the discharge session. Subjects will be free to seek other treatment after the experimental period.
At 2 moths follow-up information about side-effects of treatment will be collected using open-ended questioning. Also at 2 months the compliance with the home exercises during the 8 weeks of treatment will be assessed. At 6 and 12 months open-ended questions about recurrence, recovery and about other treatment received for their low back pain during the study period will be sought. Reported recurrence can be compared against that recorded from the monthly SMS data.
The integrity of trial data will be monitored by regularly scrutinising data sheets for omission and errors. Data will be double entered and the source of any inconsistencies will be explored and resolved.
Treatment efficacy variables
In our primary analysis, we will use a regression model to test for the effect of treatment on outcome at 2, 6 and 12 months follow up. A treatment effect will be calculated for each of the follow-up time points and, if there is a statistically significant treatment effect at any time point, we will also calculate number needed to treat (NNT) to achieve pain recovery (pain ≤ 1 out of 10 (35)) and 95% confidence intervals.
Treatment effect modifiers
Individual variables will be tested for their association with treatment effect by adding a predictor × treatment group interaction term to the regression equation.
Sample size calculation
Brookes and colleague's [45
] simulations demonstrate that trials have the same power to detect a treatment effect that is half the size of the interaction effect so sample size needs to be dictated by the size of the interaction effect
that is to be detected. We argue that small interaction effects are not clinically significant and we have planned the study with sufficient power to detect the following clinically important interaction effects.
We have taken the SD estimates from our previous trials [12
]. With specifications of alpha = 0.05, power = 0.80 and allowing for up to 10% loss to follow up and 10% treatment non-compliance a sample size of 86 participants per group will allow us to detect an interaction effect size of 1.0 SD (the smallest effect size we have specified above) and a treatment main effect of 0.5 SD (see Table ). We understand the study is somewhat overpowered for the main treatment effect but it is crucial to adequately power the study for the interaction effect.
Power calculations for main effects and interactions
Justification of study design
The trial includes key methodological features that have been recognised as minimising bias in clinical trials: true randomisation, concealed allocation, specification of eligibility criteria, blind outcome assessment, patient blinding, blind analysis and intention-to-treat analysis. The nature of the treatments precludes blinding of patient and treatment provider.
We will use an exploratory approach and will test a wide range of potential predictors of treatment response. We acknowledge that any predictor identified in the analysis should be interpreted with caution until confirmed in subsequent independent studies.
Our choice of outcomes is consistent with best practice recommendations for clinical trials studying chronic pain [46
] and back pain [47
]. Measures of pain symptoms, disability and generic health status will be taken. We have supplemented the back-related disability measure (Roland-Morris) with a patient-generated measure of disability (Patient-Specific Functional Scale) because there is evidence that patient-generated measures of disability are more responsive than condition-specific measures [22