Focus on Alternative and Complementary Therapies
www.pharmpress.com/fact
Focus Alternat Complement Ther©2005 Pharmaceutical Press
Focus Altern Complement Ther 2003; 8: 14–8
Some 80% of adults will experience low back pain (LBP) at some point in their life,1 and 40% will experience LBP in 6 months.2 The natural history of LBP is that of a self limiting yet recurrent problem.3,4 It is one of the most common reasons for visits to primary care, and the most frequent reason for visits to orthopaedic, neurosurgery and occupational medicine physicians.5 Yet, patient satisfaction with LBP visits to conventional care is known to be poor.6 Given this burden of suffering and lack of satisfaction with conventional care, it is not surprising that LBP is the most common reason for visits to complementary care providers.7 The most common complementary and alternative medicine (CAM) provider seen (in the USA) is a chiropractor,7 and the most common reason for a visit to a chiropractor is LBP.8 Spinal manipulation is the core chiropractic act.9 It is one form of manual therapy and is often used in clinical practice in combination with other forms of manual therapy, such as massage and/or mobilisation, as well as with active treatment prescriptions, such as exercise. There are several professions emphasising and/or using manual therapy and spinal manipulation therapy (SMT), including chiropractic, osteopathy, and physical therapy. There is a large body of evidence evaluating SMT, and a number of existing systematic reviews. However, these reviews come to differing conclusions. Given this seemingly conflicting body of evidence, and in the face of the strong beliefs and often passionate debates engendered by the topic, what can we conclude about the effectiveness of SMT for the treatment of LBP?
Early studies were summarised by Brunarski,10 DiFabio11 and Abenhaim,12 among others, describing most with generally positive results. However, few of the primary studies were randomised controlled trials (RCTs). The controls varied, including bed-rest, de-tuned short wave diathermy, analgesics, heat, massage, corsets, back school, mobilisation and other forms of physiotherapy, traction or various sham treatments. Other methodological issues were acknowledged by Curtis,13 including problems with subject selection, stratification, sample size estimates, standardised diagnostic categories, standardised manipulative procedures, the effects of confounding factors and appropriate statistical analysis.
Anderson et al.14 conducted an early systematic review of 23 studies and concluded SMT ‘to be consistently more effective… than were any of the ‘array’ of comparison treatments’.
A systematic review by Shekelle et al.15 in 1992 identified 58 studies, including 25 controlled trials, that were summarised to review the use, complications and efficacy of spinal manipulation for LBP. Nine trials were combined for meta-analysis to estimate effect on pain and functional outcomes. The RCTs were scored for homogeneity, comparability, follow-up, description of the intervention, type of outcome, assessment of outcome, data presentation and analysis; overall, they found the studies to be of generally poor quality. They concluded that spinal manipulation might be effective in some subgroups.
Based on a review of the primary literature, that effect is in the age group below 50 years (range 18–50 years), without evidence of neurological involvement, and the benefit appears to be in the strata 2–4 weeks of symptom onset. Compared with controls, the benefit ranges from 3% to 30% absolute improvement in symptoms. The advantage is relatively short-lived, with most groups showing similar long-term outcomes. Some studies indicate a more rapid return to work, but those data are not strong.
Interestingly, although only one of the trials (and none used for meta-analysis) was chiropractic intervention, Shekelle et al.15 go to great lengths to describe the results in the context of chiropractic care and service. This is a generalisation that they do not acknowledge, and it weakens the conclusions. It also points out an important limitation of all systematic reviews and meta-analysis of spinal manipulation, even when limited to one indication such as acute LBP: combining a wide variety of manipulative techniques may create a very general picture, but cannot answer the question of whether any particular technique or provider type is effective. It further illustrates that one must always be cautious in the acceptance of an author’s interpretation of their own results. There is often a bias to use data to support a specific profession.
An updated systematic review of the randomised clinical trials of spinal manipulation for LBP was performed by Koes et al.16 in 1996, reviewing 36 RCTs. The overall quality was considered poor. The effectiveness was compared with inconsistent controls. The most common outcome measured was pain. Acute (<6 weeks) and chronic (>6 weeks) LBP were assessed independently. The most prevalent methodological problems were similarity of baseline characteristics, randomisation, dropout description, size of population studied, clear description of intervention, inclusion of placebo group, blinding patients to treatment, blinded effect measures, and lack of an intention-to-treat analysis. Methodological rigour did not predict outcome. Unfortunately, not only was the type of manipulation applied variable, but manipulation could also have been combined with other therapeutic modalities in the intervention, making it unclear which treatment produced the results. Koes et al.’s analysis is subdivided by duration of symptoms.16
Of 12 trials, five reported positive results, four reported negative results, and three reported positive results in a subgroup only. The controls were usually physiotherapy and/or drugs.
Of seven trials, five reported positive results, and two reported negative results. The reference treatments varied: usual care, physiotherapy, back school, and analgesic use.
Of 11 trials, eight reported positive results, one positive in a subgroup only, and two reported negative results.
It should be noted that their use of the term negative is misleading, because comparison is to a standard active treatment, not a no-treatment group. Eleven trials compared manipulation with a placebo therapy (usually detuned short-wave diathermy or a sham manipulation). Seven of these reported positive results: one positive in a subgroup only, and three studies yielded negative results. Higher quality scores were associated with positive results. Criticism that limits drawing conclusions from these studies most importantly includes the poor description of the manipulation used, the lack of consistency of the manipulation used, poor inclusion-criteria definitions, and the variety of outcome measures. Overall, while there is a general trend towards a positive effect, the inconsistency of results and the poor methodology prevent conclusive statements about the effectiveness of manipulation for LBP in general.
van Tulder et al.17 performed a systematic review, all on conservative treatments of LBP, including SMT. They limited the review of manipulation to 16 RCTs for acute LBP, and nine RCTs for chronic LBP. They scored RCTs for methodology, and evaluated by comparison with placebo, and/or other conservative types of treatment. They assigned a level of evidence based on a previously used rating system and included strength of evidence measures. With stipulation for poor quality of trials, they concluded that: there is limited evidence that manipulation is more effective than a placebo treatment for acute LBP; there is no evidence that SMT is more effective than any comparison treatment for acute LBP (because of inconsistent results); there is strong evidence that manipulation is more effective than placebo for chronic LBP; there is moderate evidence that manipulation is more effective than pragmatic controls, including general medical care, bed rest, analgesics and massage. It is noteworthy that although the conclusions are different, the trials (and one of the reviewers) are the same as those reviewed by Koes et al.16 Differing criteria used for the review produces different conclusions.
More recently, Bronfort18 reviewed SMT using a ‘best evidence synthesis’ rather than meta-analysis. There were a few additional trials included in the review. Bronfort18 also subdivided by duration of symptoms, included a discussion of adverse effects, complications, cost effectiveness and provider type, and observed that for acute LBP, using magnitude of treatment effects, no chiropractic trials qualified. Similar effect sizes (medium to large) were found for osteopathic, MDs, and physical therapies (PTs) for short-term effect using standard comparison treatments such as mobilisation, heat, exercise and education. For chronic LBP, the effect sizes were of similar magnitude in the short term for PTs and chiropractors. Bronfort18 concludes that given the available evidence, there is no clear basis to suggest that SMT is more efficacious if performed by a particular provider type. There is moderate evidence of short-term efficacy in acute LBP, and moderate evidence for short-term efficacy for SMT combined with mobilisation for chronic LBP. This review methodology seems to be the most reasonable approach given the protean problems described in prior reviews.
Assendelft et al focused on only chiropractic trials of SMT for LBP.19,20 Evaluating eight RCTs, they concluded there was no convincing evidence of effectiveness for acute or chronic LBP, and they were not able to perform statistical pooling. This is consistent with Bronfort’s18 assessment.
Since the reviews mentioned above, several important trials have been published.
Carey et al.6 compared the outcomes and costs of care for acute LBP among patients seen by primary care practitioners, chiropractors and orthopaedic surgeons in both urban and rural settings. This was an observational study, collecting information on functional status, satisfaction and use of services at 2 weeks, 1, 2, 3 and 6 months. It was found that 60–70% had pain of less than 2 weeks. Estimating charges for all six groups, the chiropractic and orthopaedic charges were highest: chiropractors because of X-rays and number of visits, orthopaedists because of the cost of the visit, diagnostic procedures, and use of PT. Overall, there was rapid improvement (mean of 8 days and a median of 16 days) and only 5% had not reported functional recovery at 6 months. During the exit interviews (when patients stated they were recovered completely, or at the 6-month interview, whichever came first) patients were asked about their overall satisfaction with care. Satisfaction was higher with chiropractors than with the other practitioners. The strongest correlates of satisfaction were quality of history taking, examination, and explanation of the problem. While the functional recovery was not significantly different, cost was higher in chiropractic and orthopaedic care, and satisfaction was higher in chiropractic care. It is interesting that number of visits, need for X-rays, and prescription use varied significantly, yet outcomes were no different.
Cherkin et al.21 compared PT, chiropractic manipulation and an educational booklet for the treatment of LBP. Randomly assigning patients with pain greater than 7 days after a visit to primary care, they measured pain, level of dysfunction, and cost. The PT was provided using the McKenzie approach, and spinal manipulation was provided by four solo practitioner chiropractors. An educational booklet was distributed to the minimal intervention control group. Outcomes were measured at 1, 4 and 12 weeks, and 1–2 years. Baseline characteristics and health status were no different between groups. The chiropractic group obtained X-rays on 63% of subjects, all subjects had manipulation of the lumbar and/or lumbosacral spine and 27% had thoracic treatment, 12% had cervical treatment, 6% had pelvic treatment, and additional treatment included ice packs (20%), massage (49%) and exercises (58%). The PT group was treated by the McKenzie model with exercise only (78% self report completing the recommended exercises four out of seven previous days at the 4-week interview). The PT providers considered 55% of their patients to be complying with recommended exercises. The mean cost of chiropractic treatment was US$226, the average of PT cost was US$238 and the booklet cost US$1. There were similar outcomes in symptoms, function, satisfaction, disability, recurrences and subsequent visits for back pain between PT and chiropractic, and interestingly these were only minimally better than the outcomes when given an educational booklet but at a significantly increased cost.
In a comparison of osteopathic manipulative care and standard medical treatment for patients with benign mechanical back pain of >3 weeks but <6 months, Anderson et al.14 measured function, pain, range of motion, and straight-leg raising. Patients were randomised to either standard care or osteopathic care. Standard care included analgesics, non-steroidal anti-inflammatory drugs (NSAIDs), cyclobenzaprine, active physical therapy or ancillary PT modalities, including ultrasound, diathermy, TENS, hot or cold packs or a corset. In the osteopathic treatment group, manipulative treatment was also provided. Osteopathic treatment was individualised, including thrust, muscle energy, counter-strain, articulation and myofascial release techniques. Baseline demographics, severity of back pain, and frequency of non-musculoskeletal disease were similar. There were no statistical differences in pain scores, function questionnaires, or objective measures at the end of the study period. The use of medication was higher in the standard care group, and physical therapy was used more frequently in the standard care group. Satisfaction was high in both groups. Of note here was that standard care included a wide variety of therapies, some without evidence of effectiveness, which were applied inconsistently.
Skargen et al.22 randomised patients to physiotherapy or chiropractic as primary management for back pain, and measured pain, function, general health, recurrence rate and costs of care. There were no differences in health improvement, costs or recurrence rates. Patients presenting with 1 week of pain were modestly improved with chiropractic, and those presenting with pain longer than 1-month duration were modestly improved with physiotherapy.
Hurwitz et al.23 conducted an RCT comparing chiropractic with and without physical modalities, and medical care, with and without PT. After a 6-month follow-up period, there was no significant difference in pain or disability.
Hsieh et al.24 randomised adult patients with LBP of 3-weeks to 6-months’ duration to manipulation, myo-fascial therapy, a combination of the two or back school, and measured pain and disability at 3 weeks and 6 months. There were no significant differences between the groups.
It is clear that one can find trials to support any point of view, and that review conclusions depend on assumptions and methodology. The assumptions must be clinically relevant to the question being asked. For example, are the reference treatments sham or active controls? Are the SMTs equivalent (such as chiropractic, osteopathic, and physical therapy techniques)? Is the outcome measure comparable? Are the populations equivalent? In addition, the statistical methods of comparison for meta-analysis must recognise that fixed-effect models most commonly used assume study population homogeneity, which is rarely the case.
The problem with conducting a meta-analysis or even a systematic review of these studies is that the conclusions drawn extrapolate, at times too broadly, what information can honestly be concluded. Here, at best, we have differing treatment methods that fall into a broad category of passive movement or manual therapy/manipulation. We can only say manipulation may at times be effective for some subpopulations of patients with LBP. In addition, the study populations cannot be compared when they have varying symptom complexes. Perhaps an analogy would be that we conduct a meta-analysis on all drugs within a single class for several related conditions. The only conclusion possible is a general sense that this class of drugs may be of use for some of the conditions studied. Another analogy could be comparing differing surgical techniques for abdominal pain, done by all surgeons, and try to draw a conclusion. Until a clear diagnosis (at least a clear clinical symptom complex) is reliably made, and a specific technique is defined for study, very little can be conclusively stated, except to direct trials that are more specific.
The clinical questions are not yet reliably answered. Which manipulative technique will help a specific patient and when? We can say that there is evidence that the type of provider makes no difference in clinical outcome. There is also evidence that patient satisfaction correlates with a good history and physical examination, along with an adequate (in the patient’s terms) explanation of the problem. There is evidence that there is no long-term improvement with any form of SMT. The most reasonable statements at present seem to be that there is some evidence of effect for some subpopulations of patients with LBP. There is moderate evidence of short-term effect of SMT in the treatment of subacute LBP. There is also modest evidence of short-term effect of SMT for patients with chronic LBP, but these patients fare better with an exercise programme, with or without SMT.
Risks and costs must be adequately addressed to complete this discussion, but falls outside the scope of this review.
