Clinical question: what is the effectiveness of general supervised exercise programmes or specific exercise training programmes (individual and group) compared with usual care on pain, functional disability or psychological distress?
6.3.1. Clinical evidence
Eight studies were included for this question: 1 Cochrane review, 1 RCT on yoga, 1 on hydrotherapy/spa therapy and 5 on exercise programmes.
A systematic review (Hayden, J. A., van Tulder, M. W., Malmivaara, A. et al, 2005) evaluated the effectiveness of exercise therapy in adult nonspecific acute, subacute and chronic low back pain versus no treatment and other conservative treatments. The Cochrane Central Register of Controlled Trials, MEDLINE, EMBASE, PsychInfo, and CINAHL databases to October 2004 were searched, alongside citation searches and bibliographic reviews of previous systematic reviews. The aim was to identify randomised controlled trials involving participants with nonspecific low back pain comparing exercise therapy to no treatment/placebo/sham, another conservative therapy or another exercise group. Outcomes of interest were self-reported pain intensity, function, global improvement and return-to-work. Pooled analysis of four trials of sub-acute patient populations suggest that there is insufficient evidence to support or refute the effectiveness of exercise therapy for reducing pain intensity and improving function. Meta analysis of functional and pain outcomes from 20 and 23 studies respectively involving chronic low back pain patient populations suggests exercise therapy is slightly effective at decreasing pain and improving function relative to other comparisons (no-treatment, sham, placebo or other conservative treatment). People involved in the studies on chronic low back pain may have had co-interventions during the study period.
This was a high quality systematic review with a very low risk of bias.
The United Kingdom back pain exercise and manipulation (UK BEAM) trial (UK Back pain exercise and manipulation (UKBEAM) Trial Team., 2004) aimed to estimate the effectiveness of adding exercise, spinal manipulation or a combination of both to the standard care in general practice. Patients recruited from participating centres had to be aged 18–65 and have had pain everyday for the 28 days before randomisation (or 21 out of 28 days before randomisation and 21 out of 28 days before that). They also had to agree to avoid physical treatment other than trial treatments for 3 months. Exclusion criteria included cancer, osteoporosis, ankylosing spondylitis, cauda equina compression, previous spinal surgery, anticoagulant treatment and cardiovascular disease or hypertension.
A total of 1334 patients were included in the study, with 310 randomised to the exercise group and 338 were randomised to a ‘Best Usual Care’ control group. All patients received advice to continuing normal activities and avoiding rest, and were provided with copies of ‘The Back Book’. Following an initial individual assessment participants randomised to the Exercise programme attended group classes incorporating cognitive behavioural principles. The programme was delivered by trained physiotherapists, and the participants were invited to attend up to eight 60-minute sessions over four to eight weeks, and a “refresher” class at 12 weeks after randomisation.
Results showed that compared to Best Care, the exercise programme produced statistically significant improvements in mean RMDQ score at three months only (P <0.01), in mean Von Korff disability and pain scores and back beliefs score at both three and 12 months (P <0.05 at both follow-ups), and in mean SF-36 physical score and fear avoidance beliefs physical score at three months only (P <0.001). Mean SF-36 mental score did not differ.
This was a high quality RCT with a very low risk of bias
One randomised controlled trial (Kuukkanen, T. and Mälkiä, E., 2000; Kuukkanen, Tiina, Mälkiä, Esko, Kautiainen, Hannu et al, 2007) assessed the effectiveness of a home exercise programme on patients with nonspecific low back pain. Patients were recruited from eight regional occupational healthcare centres in central Finland and referred to physicians in a hospital in central Finland. Inclusion criteria included a local place of residence, age between 20 and 55, employment and no sick leave exceeding a total of three months during the previous year, disabling LBP over three years, pain at rest or with stress and localisation to lumbar area or buttocks. Exclusions included need for surgery, pregnancy, history of back disease (cancer, fracture, spondylarthritis ancylopoetica or infection), substance abuse and somatic or psychiatric disorder preventing patients from exercising.
A total of 57 patients were randomly allocated to a home exercise programme group (n=29) or a control group (n=28). Patients in the home exercise programme received a three month programme consisting of three progressive monthly programmes. The physiotherapist instructed the patients on the exercises, which aimed to improve the function of abdominals, back extensors, upper and lower limbs muscles, and established the optimal function of the spine. The progression of the programme was based on weekly tests, which the home exercise group performed independently. A physiotherapist supervised the exercise programmes once a month in an exercise room. The programmes were carried out at home, without extra equipment, with 10min warm-up and cool-down periods. The load of each exercise movement was individually adjusted according to the repetition maximum. The exercises were performed as three to four sets of 15–20 repetitions. The goal was for subjects to attempt exercises every day, and to record this in their diaries. Patients in the control group did not alter physical activity levels or participate in any exercise programme during the study.
Results showed that pain intensity and functioning decreased significantly in all subjects during the study period, and that for patients in the home exercise group those values remained below baseline values in the 12 months follow-up. After five years pain intensity was significantly lower (P <0.01) in the home exercise group. Functioning also decreased in that group over the five year period, but there were no statistical difference between the groups (P <0.27). The overall conclusion is that the study indicates that supervised controlled home exercises lead to reduced LBP and that positive effects were preserved over five years.
This was a RCT with a high risk of bias
One randomised controlled trial assessed the effectiveness of Alexander technique lessons, exercise prescription and massage for chronic and recurrent back pain (Little, P., Lewith, G., Webley, F. et al, 2008). Participants were recruited from 64 general practices in the UK. Participants (aged 18 to 65) had to have presented in primary care with low back pain more than 3 months previously, score 4 or more on the RMDQ, have current low back pain for more than 3 weeks. Exclusion criteria included previous experience of Alexander Technique, clinical indicators of serious spinal disease, current nerve root pain, previous spinal surgery, pending litigation, history of psychosis or major alcohol misuse, and perceived inability to walk 100m.
A total of 579 participants were included in the study: of these 72 received normal care; 73 received six lessons in Alexander Technique; 73 received 24 lessons in Alexander Technique; 72 received exercise prescription; 72 received exercise prescription and massage; 71 received exercise prescription and 6 lessons of Alexander Technique; 71 received exercise prescription and 24 lessons in Alexander Technique. The Alexander Technique and Exercise prescription treatments were compared to each other and to normal care. Outcomes were the RMDQ, number of days of pain in the past four weeks, quality of life, Von Korff scale and the Deyo ‘troublesomeness’ scale. These outcomes were measured at baseline, 3 months and 1 year. Lessons in Alexander Technique lasted 30–40 minutes and each participant was encouraged to record the time between lessons dedicated to practicing the Alexander Technique.
Results showed significant changes in the RMDQ score and days in pain at three months for all groups compared to the control group. Exercise prescription and lessons in the Alexander Technique were still effective at one year compared to the control group (P =0.045, P <0.001 and P =0.008 for 6, 24 lessons of Alexander Technique and exercise prescription respectively). The overall conclusion was that structured programmes of Alexander Technique and exercise prescription compared to usual care were effective at reducing pain and functional disability. Additionally, six lessons in Alexander Technique followed by exercise prescription were nearly as effective as 24 lessons.
This was a well conducted RCT with a low risk of bias.
One randomised controlled trial involved hospital employees with chronic low back pain (Maul, I., Läubli, T., Oliveri, M. et al, 2005). Potential candidates were recruited amongst employees of a large university hospital (Switzerland) who returned a modified version of the Nordic Questionnaire on LBP. Inclusion criteria included over 30 days of low back pain in the previous 12 months, an age between 20 and 55 and the ability to read and write German or Italian. Exclusions included cardiovascular or metabolic diseases, progressive radicular neurological defects, inflammatory disease of the spine, previous spinal surgery, pregnancy and regular strength training within the last six months.
A total of 97 patients were allocated to the Exercise group, and 86 were allocated to the comparison group. All patients attended a back school which consisted of three sessions, each lasting one hour and giving information about functional anatomy of the spine, correct lifting techniques, how to use mental stress coping strategies and giving advice on sports activities. Additionally, patients in the exercise programme groups received exercises based on concepts of medical training therapy and sequence exercise training. The programme consisted of three phases of individual training, each lasting four weeks with sessions two or three times a week. Each training session was supervised by a physiotherapist.
Results showed that in addition to back school, supervised physical training effectively improved functional capacity in terms of muscular endurance and isokinetic strength during a six months follow-up. Furthermore, self-rated pain and disability significantly decreased during a one-year follow-up.
This was a RCT with a high risk of bias
One randomised controlled trial aimed to determine the effectiveness of graded activity as part of a multistage return-to-work (RTW) programme (Steenstra, I. A., Anema, J. R., Bongers, P. M. et al, 2006). A total of 112 workers absent from work for >8weeks due to LBP were randomised to either graded activity (n=55) or usual (n=57). Inclusion criteria were sick leave for >8 weeks and no plans to return to work within a week, inclusion in the multistage RTW back pain management programme at two to six weeks of sick-leave, age between 18 and 65 and ability to read and write in Dutch. Exclusion criteria were specific cause to the LBP, coexisting cardiovascular, psychiatric contraindications or juridical procedures pregnancy, sick leave due to LBP less than a month prior to current episode. Outcomes were return-to-work, pain intensity and functional status.
Workers in the graded activity group received an individual, submaximal, gradually increasing exercise programme, with an operant-conditioning behavioural approach. This was based on findings from patient history, physical examination, functional capacity evaluation, the demands from the patients’ work and the patients’ expectations on time to return to work. The entire programme consisted of 26 one-hour sessions maximum, with a frequency of 2 sessions a week. Workers in the usual care group received care following the Dutch occupational physician guidelines for low back pain. Patients were followed-up at 12 weeks and 26 weeks. Results showed that graded activity did not improve pain or functional status clinically significantly.
This was a RCT with a high risk of bias
Hydrotherapy/Spa therapy studies
One randomised controlled trial investigated the claimed benefits of group hydrotherapy for subjects with chronic low back pain (McIlveen, B. and Robertson, V. J., 1998). Following publication of an article about the study in the local newspaper, subjects referred for hydrotherapy by their GP or physiotherapist contacted a large community care centre in Australia. Patients were then assessed for suitability and were excluded if they couldn’t read or write in English, had spondylolisthesis, had had previous lower limb joint replacement surgery or were receiving work or traffic injury-related compensation insurance. Other exclusion criteria were uncontrolled hypertension, severe postural hypotension, left heart failure, exercise induced angina, lung vital capacity of less than 1.5 litres, faecal or urinary incontinence, an allergy to chlorine, severe limiting airways disease, early pregnancy (i.e. 1st trimester), and a tendency to antisocial behaviour such as can occur with a head injury,.
A total of 56 subjects were randomly assigned to the hydrotherapy group, and 53 were assigned to a control group (delayed hydrotherapy). Patients in the hydrotherapy group participated in 60-min group hydrotherapy sessions twice weekly for 4 weeks. Each session was led by experienced pool volunteers with additional training in delivering the prescribed 20 spinal exercises. Ten repetitions of each prescribed exercise were included in each session. Prescribed exercises included walking in water, marching on the spot, swinging the legs backwards and forward in the water, bicycling the legs and pushing and pulling a kickboard with the hands. Patients in the control group were placed on the existing 4-week waiting list for hydrotherapy. Both groups were reminded not to start any other treatment, medication or exercise programmed for their low back pain during this period. Outcomes were range of flexion, extension, pain, and function.
Results showed that patients in hydrotherapy group significantly improved in function (measured by the Oswestry Disability Index, P <0.05). However, the differences between subjects in the experimental and control groups were not significant for the other measures of pain or the ranges of flexion and extension.
This was a RCT with a high risk of bias
Yoga therapies
One randomised controlled trial aimed to determine whether yoga is more effective than conventional exercise or a self-care book for patients with chronic low back pain (Sherman, Karen J., Cherkin, Daniel C., Erro, Janet et al, 2005). Patients from a non-profit integrated healthcare system in the USA were recruited. Letters describing the study were mailed to patients matching the inclusion criteria (based on the available electronic records). The study was also advertised in the consumer magazine. Patients had to be aged between 20 and 64, have visited a primary care provider for treatment for back pain 3–15 months before the study (according to electronic records), and have the ability to read and understand English. Exclusion criteria were sciatica, previous back surgery, spinal stenosis, pregnancy, cancer, spondylolisthesis, fractured bones, dislocated joints, concurrent treatment for back pain, participation in yoga or exercise training for back pain in the previous year, current litigation, unstable medical or severe psychiatric conditions and contraindications or schedules that preclude class participation.
A total of 101 patients were randomly assigned to the yoga group (n=36), the exercise group (n=35) or a self-care booklet group (n=30). The yoga and exercise classes were developed specifically for the study and consisted of 12 weekly 75min classes designed to benefit people with chronic low back pain. Participants were also asked to practice daily at home. Patients in the yoga group performed vini yoga, which emphasises safety and is relatively easy to learn. All sessions emphasised the use of postures and breathing, and each session had a specific focus: relaxation; strength-building, flexibility, and large-muscle movement; asymmetric poses; strengthening the hip muscles; lateral bending; integration; and customising personal practice. The postures were selected from a core of 17 relatively simple postures. Each class included a question and answer period, an initial and final breathing exercise, five-12 postures, and a guided deep relaxation. Patients in the exercise group followed a specifically-designed 12-session class series. Each session consisted of an educational talk, a warm-up to increase the heart rate, repetitions of a series of seven aerobics exercises and 10 strengthening exercises that emphasised leg, hip, abdominal and back muscles. Over the course of the 12-weeks series, the number of reps of each aerobic and strength exercise increased from eight to 30 in increments of two. The strengthening exercises were followed by 12 stretches for the same muscle groups. Classes ended with a short, unguided period of deep slow breathing. Patients in the self-care book group were mailed a copy of the Back Pain Helpbook, an evidence-based book that emphasised such self-care strategies as adoption of comprehensive fitness and strength programme, appropriate lifestyle modification and guidelines for managing flare-ups.
Results showed that after adjustment for baseline values, back-related function in the yoga group was superior to the book and exercise groups at 12 weeks (P <0.001). No significant difference in “bothersomeness” of pain was found between any two groups at 12 weeks. At 26 weeks, back-related function in the yoga group was superior to the book group (P <0.001). At 26 weeks, pain bothersomeness was also better in the yoga group than in the book group (P <0.001). Overall, yoga was more effective than a self-care book for improving function and reducing chronic low back pain and the benefits persisted for at least several months.
This was a well conducted RCT with a low risk of bias
6.3.2. Health economics
Two studies were included. One was a UK-based cost-effectiveness study of four interventions for treatment of low back pain, two of which included exercise programmes. The second was a UK-based economic evaluation of the Alexander technique.
The first study aimed to assess the cost-effectiveness of adding exercise, spinal manipulation or a combination of both to standard care in general practice. An economic evaluation was conducted alongside the UK Back pain Exercise And Manipulation trial. (UK Back pain exercise and manipulation (UKBEAM) Trial Team, 2004) Patients recruited from participating centres had to be aged 18–65 and have had pain everyday for the 28 days before randomisation (or 21 out of 28 days before randomisation and 21 out of 28 days before that).
The four treatment groups were 1) best care, which included active management and providing ’The Back Book’ to patients, 2) best care + an exercise programme of up to nine classes over 12 weeks, 3) best care + spinal manipulation package of eight sessions over 12 weeks and 4) combined treatment, which included best care + six weeks of manipulation followed by six weeks of exercise. The main outcome measures were healthcare costs, quality adjusted life years (QALYs), and cost per QALY over 12 months. The number of QALYs gained over 12 months was estimated using EQ-5D questionnaire responses which were collected as part of the trial. The costing perspective was that of the UK health service. Healthcare resources included those for: the spinal manipulation package, the exercise programme, hospital inpatient stays, outpatient attendances, and general practice consultations. These resources were costed using published national averages for England. Private care was costed using information from a major insurance provider. Costs were reported in pounds sterling at 2000/2001 prices. Costs were not discounted because the focus was on effects over only one year.
To cover scenarios in which either exercise or manipulation was not available ICERs were calculated to compare best care with manipulation alone or exercise alone.
Results (base case)
The mean cost (Standard Deviation) of best care was £346 (£602). Best care+exercise cost £140 more than best care. Relative to best care, best care+exercise generated an additional 0.017 (−0.017 to 0.051) QALYs.
At base case, best care + exercise was dominated by combined therapy: it cost more and generated fewer QALYs over the 12 month period. With all options available, the combination package was the most cost effective strategy. However, if manipulation was not available (n=668) exercise generated 0.017 more QALYs per patient than best care at an additional cost of £140 per patient, yielding an ICER of £8,235 per QALY.
Sensitivity analysis
Sensitivity analysis examined the impact on costs if the NHS purchased private care for some or all of the patients. The justification for this was that in the short term it might be difficult to make all manipulation or combined treatment available within the NHS: there are insufficient numbers of trained practitioners in the NHS to meet demand and it would take a few years to train people up within the NHS. The results did not change the finding of the base case analysis.
To conclude, this analysis suggested that the cost-effectiveness of the included exercise programme, when added to best care had an ICER of £8,300 compared to best care alone. Furthermore, there was about a 60% chance that the estimated ICER was less than £20,000 per QALY.
A 12 month cost effectiveness study compared the Alexander technique (AT), with normal care, with massage and with an exercise programme, in patients with chronic and recurrent back pain (See section 1.2.1 for a description of the RCT). (Hollinghurst, S, Sharp, D., Ballard, K. et al, 2008)
The 4 main treatment groups were AT-6 lessons, AT-24 lessons, normal care (control group) and massage. Half of the participants in each group were prescribed a home based exercise programme and nurse behavioural counselling by their GP (from hereon this will be referred to as the exercise prescription), resulting in 8 groups altogether. The study size was 579.
The study took a societal perspective but reported NHS costs separately. NHS resources included those for primary care, outpatient and inpatient contacts as well as medication. NHS resources were costed using national published estimates, in 2005 prices. Main health outcomes were Roland-Morris disability score, days in pain and QALYs derived from EQ-5D questionnaire data collected at baseline and 3 monthly intervals.
NHS mean cost per patient (Standard deviation) for each of the 4 groups which did not have an Exercise prescription component were as follows: normal care £54 (100); massage £258 (204); AT- 6 lessons £218 (146); and AT-24 lessons £610 (262). In the four groups which included an exercise prescription, costs were as follows: normal care £154 (523); massage £267 (363); AT-6 lessons £239 (107); AT-24 lessons £661 (328).
The authors performed incremental analysis for a selection of interventions based on what they considered the appropriate comparator groups to be. However, for the purposes of this guideline the objective was to compare the cost-effectiveness of each of the interventions with each other. Therefore, in a separate exercise the single interventions of exercise prescription, AT-6 lessons, massage and AT-24 lessons were assessed using normal care alone (control group) as the main comparator, by using data from the published study. The study performed two further analyses. In one of these the exercise prescription was taken out of the analysis to investigate the cost-effectiveness of AT-24 lessons compared to AT-6 lessons and massage. In the second, the focus of the cost-effectiveness analysis was on the addition of AT and massage to the exercise prescription. It should be noted that the latter two types of analysis are presented here for completeness and for illustrative purposes. The validity of the approach, where the exercise prescription option is excluded from the analysis, is questionable given that the exercise prescription turned out to be the most cost-effective single intervention. Similarly, it is unclear why AT-6 lessons or massage would be added to the exercise prescription when the latter two interventions were dominated by the exercise prescription in the cost-effectiveness analysis of the single interventions.
Single interventions
The incremental cost of AT-6 lessons compared to normal care alone was £163 and the incremental QALY gain was 0.03. Therefore, AT-6 lessons resulted in a cost per QALY of £5,400 compared to normal care alone. However, when AT-6 lessons was compared to normal care plus the exercise prescription the incremental cost of AT-6 lessons was £63 and the incremental QALY gain was −0.01 which meant that the AT-6 lessons intervention was dominated by the exercise prescription. That is, it cost more and produced fewer benefits compared to the exercise prescription.
It was not possible to calculate the incremental cost effectiveness of AT-24 lessons compared to normal care alone due to lack of QALY data reported in the study. However, AT-24 lessons was £456 more costly than the exercise prescription and the incremental QALY gain from the AT-24 lessons intervention was 0.01. This meant that the cost per QALY gained with AT-24 lessons was £45,600 compared to the exercise prescription.
Cost-effectiveness analysis using the Roland scores showed that massage and AT-6 lessons should both be excluded because of dominance by the exercise prescription. For the AT-24 lessons intervention, the cost per one point improvement in the Roland score was £168 compared to the exercise prescription.
In terms of pain-free days the exercise prescription is the least costly at £9 per pain-free-day gained compared with normal care alone. AT-6 lessons cost £31 per pain-free-day gained relative to the exercise prescription, while AT-24 lessons cost £56 per pain-free-day gained compared to AT-6 lessons.
Excluding exercise from the analysis
When the exercise prescription as a single intervention is excluded from the analysis there remained three single interventions to be compared with normal care alone: that is, AT-24 lessons, AT-6 lessons and massage. Incremental cost-effectiveness analysis shows that massage was dominated by AT-6 lessons. The cost per QALY of AT-6 lessons was £5,704, cost per point reduction in Roland disability score was £89, and cost per pain-free-day gained was £12, compared to normal care. The incremental cost-effectiveness of AT-24 lessons was £17,454 per QALY, £203 per one point improvement in the Roland disability score, and £51 per pain free day gained, compared to AT-6 lessons.
Double or two-stage therapies
This analysis considered the addition of AT-6 lessons, AT-24 lessons or massage to the exercise prescription. When the cost-effectiveness analysis used the Roland disability score and pain-free days as the main outcomes, the addition of massage was dominated by the addition of AT-6 lessons. The cost per QALY gained from adding AT-6 lessons was £915, compared to AT-6 lessons alone. However, there was a very small QALY gain associated with adding massage over adding AT-6 lessons. This resulted in a cost per QALY gain of £5,217 for the addition of massage compared to the addition of AT-6 lessons. When AT-24 lessons was added to the exercise prescription the cost per QALY gained was £13,914 compared to the addition of AT-6 lessons.
It should be noted that the results of the economic analysis in this study are fairly unstable due to the wide confidence intervals around costs and outcomes.
