13. Manipulation
Alan Breen
Chapter contents
Introduction255
What is manipulation?256
The manipulation ‘package of care’264
Researching clinical effectiveness266
Research and access to manipulation in the UK269
Manipulation and the transition to mainstream health care270
Conclusion270
Introduction
Manipulation is a mechanical treatment that has often been recommended in national evidence-based guidelines for common musculoskeletal disorders (Agency for Health Care Policy Research, 1994, Waddell et al., 1999, (NHMRC) NHaMRC, 2003, Airaksinen et al., 2006, van Tulder et al., 2006, Haldeman et al., 2008 and NICE, 2008). Such is the volume of evidence for its clinical value that it may be reasonable to ask why it is still sometimes regarded as a complementary therapy. However, part of the reason for this probably lies in gaps in the research literature in relation to its role in care. Firstly, manipulative treatment implies a rationale that links a reduction of mechanical dysfunction to improvement in health, but the relevance of mechanical dysfunction models in health remains unclear in the research literature. Secondly, manipulation is not generally practised as a monotherapy, because practitioners usually provide other interventions along with it (Burton, 1981, Huisman and Breen, 1990, Pedersen, 1994, Wilson, 2003 and Chown et al., 2008). The effects of manipulation cannot therefore be the only factors in play. This requires us to consider those of the other interventions that may often accompany it (Harvey et al. 2003). Research in this context is relatively new.
The third reason relates to its provision in health care. Manipulation and the other components of the care that surrounds it take a long time to learn if they are to be practised competently. This makes it difficult, if not impossible, to incorporate manipulation into undergraduate medical training. More importantly, the musculoskeletal conditions for which it is mainly used do not fit well into a purely biomedical model of health, even though they do constitute a huge societal cost to industrialized nations (World Health Organization 2001). Instead, they also usually call for a functional approach. Because the health, social and societal costs of musculoskeletal disorders are generally considered separately, their real impact on the gross domestic product of western countries is seldom fully appreciated. Public resources for their care are generally given relatively low priority compared to, for example, cancer and heart disease. Sufferers are also sometimes denigrated as the ‘walking wounded’ and their practitioners regarded as somewhat marginal (Coulter & Shekelle 2005).
The scope for gaining useful knowledge around this topic through research is therefore very wide and ranges from the most intrinsic to the most global of health issues. This chapter will attempt to provide an overview of these.
What is manipulation?
Manipulation is most commonly defined as: ‘a small amplitude, high velocity movement at the limit of joint range taking the joint beyond its passive range of movement, without exceeding its anatomical integrity’ (Sandoz 1969; Figure 13.1). This is therapeutically different from mobilization, which is a repetitive passive movement (i.e. within the passive range) (Harvey et al. 2003) (Figure 13.2). Manipulation is sometimes accompanied by an audible pop, or cavitation (Unsworth et al. 1971). However it has been argued that this is not necessary to qualify as a successful procedure (Flynn et al. 2003). Apparently, when it does occur, it may also come from multiple surrounding joints (Ross et al. 2004). The possible effects of cavitation manipulation on the mechanics of joints have been extensively discussed (Evans, 2002 and Breen et al., 2007), but useful experimental studies in living subjects are difficult to perform (Reeves et al. 2007).
 |
| FIGURE 13.1 |
The mechanical dysfunction model
Until relatively recently, the mechanical dysfunction model of health was portrayed as an alternative paradigm, propounded by heterodox practitioners who offered it for all manner of conditions (Hewitt & Wood 1975). However, we now know that such care is used by the public more to complement conventional practice than to replace it (Eisenberg et al., 1993 and Thomas et al., 2001), mainly for common musculoskeletal complaints and especially for low-back pain (Breen, 1977, Burton, 1981 and Pedersen, 1994). Trial evidence reviews have shown that spinal manipulation, which depends on such a model, has moderately positive effects for low-back pain (Assendelft et al. 2004). The problem is that this evidence is not consistent, which tempts a certain amount of selective reporting by reviewers. It also reflects the heterogeneity of the samples, interventions, trial designs and outcome measures. Nonetheless, its recommendation as a treatment in most back pain guidelines (European Commission CBMC, 2006, Chou and Huffman, 2007 and Hurwitz et al., 2008) is on the basis of reducing pain and disability in people who are not recovering spontaneously from acute back pain (van Tulder et al. 2006) or are suffering exacerbations of chronic back pain (Airaksinen et al. 2006). More research has been called for to identify the kinds of patients who will benefit most from manipulation (van Tulder et al. 2006).
The establishment of subgroups
One approach to identifying subgroups has been to develop and test prediction rules based on accessible clinical findings (Childs et al. 2004). For example, patients with recent-onset back pain, without leg pain, are thought to respond better to manipulation than those with longer-duration pain which radiates down the leg. However, these rules are said to require further validation (Hancock et al. 2008). In the meantime, the accepted model of care for people who are not recovering, once serious pathology has been excluded, is to adopt a psychosocial and general physical fitness approach (Waddell 1998). Despite reasonable rationales for a relationship between biomechanics and spinal pain, therefore, we lack scientific explanation of how manipulation confers benefit. Furthermore biomechanical measures themselves are seldom recommended for use for assessing outcomes (Pincus et al. 2008a).
How then could manipulation contribute to reduction in pain and disability for such conditions? It is common to refer to laboratory studies of cadaveric spines (White & Panjabi 1990) to try to understand how mechanical interventions reduce spinal pain and disability. These studies have shown, for example, that during trunk movement, stiff segments transfer mechanical stress to adjacent ones, causing abnormal force concentrations there (Bowden et al. 2008). Motion segments that become lax may repeatedly tug at their attachments during unguarded movements. It makes sense to think that this could result in pain and a variety of other neuromuscular effects (Panjabi, 1992a, Panjabi, 1992b and Panjabi et al., 1992). It is not difficult to imagine how spinal manipulation could influence this, but there is no good clinical evidence that it happens in patients and is related to pain.
How do we investigate the mechanical dysfunction model?
Part of the rationale for manipulation is to increase intervertebral motion and reduce painful deformation of intervertebral discs and ligaments (Haldeman 1996). For example, there is some evidence that manipulation may increase overall range in the cervical spine (Nansel et al., 1989, Cassidy et al., 1992 and Nilsson et al., 1996). However, increasing range of motion has not been confirmed as being associated with reduced disability or pain (Nattrass et al. 1999) and evidence that the manipulation of spinal segments actually does increase their range is rare. This is largely due to poor reliability in the measurement of intervertebral motion from the end-of-range radiographs that are traditionally used for this (Shaffer et al., 1990, Panjabi et al., 1992 and Mayer et al., 1995). The field of spine stabilization surgery suffers from similar problems (Reeves et al. 2007) and we have long needed an objective and reliable way of measuring spine kinematics in living subjects. However, this is gradually becoming possible using quantitative fluoroscopy, which is a combination of digital fluoroscopy and image processing (Teyhen et al., 2005, Breen et al., 2006 and Wong et al., 2006) where hundreds of digital X-ray images are recorded as the patient is moved and the motions of the vertebrae are tracked and recorded using computer programs (Figure 13.3). A graphical example of increased range of motion following lumbar manipulation is shown using this technology in Figure 13.4.
 |
| FIGURE 13.3 |
These quantitative fluoroscopic studies have the potential to reveal what previously only cadaveric studies could show. Since most back and neck pain is not associated with either serious pathology or nerve compression, subgrouping of ‘non-specific’ spinal pain is essential and some of these subgroups may be mechanical ones.
‘Instability’
Perhaps the most popular model of pain-producing mechanical dysfunction of the spine is segmental ‘instability’, which until recently has only been quantifiable in specimens (Reeves et al. 2007). The term ‘instability’ is probably inappropriate in relation to non-traumatic mechanical spinal disorders and investigators have traditionally used alternatives such as ‘hypermobility’ (Van Akkerveeken et al. 1979) and ‘laxity’ (Fernandez-Bermego et al. 1993). However, given the wide variation of the normal range of intervertebral motion, ‘hypermobility’, taken as increased range alone, would be difficult to associate with pain. Laxity, however, may be another matter. A lax joint does not necessarily move too far under normal physiological stresses – it moves too fast. This distinction is embedded in Panjabi’s ‘neutral zone’ theory (Panjabi, 1992a and Panjabi, 1992b), in which pain is generated when tissues fail to restrain their vertebrae near the neutral position (Figure 13.5), causing altered loading during motion and irritation at its end-range. It is a small step to imagine how manual therapies, aimed at the vertebral levels involved, or their adjacent levels, could influence this.
 |
| FIGURE 13.5 |
There are long-established theories (Gertzbein et al., 1985 and Kirkaldy-Willis, 1992), as well as cadaveric evidence (Mimura et al. 1994) that intervertebral laxity occurs with early-stage disc degeneration. Such laxity seems to be caused by loss of normal intradiscal pressure, which, under load, is a major restraining influence between vertebrae (Zhao et al. 2005). Early-stage disc degeneration is much more readily detectible with magnetic resonance than with plain X-rays (Pfirrmann et al. 2001) but there is emerging evidence, also from magnetic resonance studies, of an association between aberrant (paradoxical) intervertebral motion and disc degeneration as detected by magnetic resonance (Fowler et al. 2006), adding weight to the concept of aberrant motion as a source of mechanical pain.
Recent quantitative fluoroscopy research has moved this model of mechanical dysfunction from the realm of theory to that of biomechanical measurement and from cadaveric to patient studies. Separate from this, and using cadaveric studies, Mimura et al., 1994 and Zhao et al., 2005 established a relationship between laxity and early disc degeneration, expressing it as the proportion of overall intervertebral range that is achieved in the neutral zone. Using quantitative fluoroscopy, Wong et al., 2004 and Breen et al., 2000a then measured it in living subjects. The latter also demonstrated lax appearance in the motion patterns of lumbar motion segments adjacent to surgically stabilized ones (Figure 13.6).
 |
| FIGURE 13.6
(from quantitative fluoroscopic studies).
|
It is important to have a practical measurement of laxity to use in patient studies. Wong et al. (2004) used the slope of intervertebral motion against trunk motion to express laxity (Figure 13.7). Mellor et al. (2008) used both the slope of intervertebral to trunk motion and the proportion of the intervertebral range in the first 10o of trunk motion. Such studies are gradually accumulating normative reference information with which to compare patient data.
What this amounts to is the capability to measure intervertebral laxity and the influence that physical therapies may have upon it in living subjects. This is a new horizon in research into spinal manipulation that could lead to better trials of efficacy. It should then also be possible to establish whether manipulation does, as in Figure 13.4, restore mobility to stiff spinal joints. This old concept of the mechanism of spinal manipulation (Haldeman, 1992, Peterson and Bergmann, 2002, Liebenson, 2006 and Morris, 2006) would then at last be testable. However, this is just the beginning. The mechanical effects of manipulation could also be explained by changes in muscle activity and length (Sihvonen et al., 1991, Kaigle et al., 1998, Mannion, 1999 and Lariviere et al., 2002) and by neurological and circulatory effects (Sandoz, 1978 and Wyke, 1985). This adds further levels of complexity and calls for even more sophistication in study design.
How safe is manipulation?
Given the higher range of motion intended with manipulation, there are implications for both clinical effectiveness and patient safety. Mainly used for low-back pain, the risk of serious adverse effects is thought to be extremely low in skilled hands, therefore guidelines generally stipulate only that practitioners should be adequately trained (van Tulder et al. 2006).
The second most frequent use of manipulation is for neck pain (Pedersen 1994), where there have been reports in the research literature, plus considerable press speculation, about the (rare) occurrence of postmanipulation stroke (Ernst 2002). In a recent prospective study of 529 chiropractic patients who received neck manipulation, the main adverse effects, albeit transient, were reported to be headache or a temporary worsening of neck symptoms (Rubinstein et al. 2008). Even the most severe of these did not have long-term effects. In a subsequent large prospective cohort study of 50 275 neck manipulations (Thiel et al. 2007) no serious adverse effects were reported.
The Neck Pain Task Force of the World Health Organization found that neck mobilization and manipulation have similar degrees of effectiveness (Gross et al., 2005 and Hurwitz et al., 2008). Combined with exercise, they are reported to be more effective than the alternative strategies to which they have been compared. In the absence of any good evidence of a greater risk from manipulation, the authors recommended that the choice between the two should be guided by patient preference (Hurwitz et al. 2008). However, given the concerns that have been voiced, it is essential to try to balance the good that manual therapies do against any possible harm that might occur.
The quality of studies about harm
Researching the risk of serious harm from a treatment sometimes brings researchers into contact with stakeholders with strong ontic commitment to certain conceptions of the physical system in play (Martin 1962). Where the research question and the interpretation of evidence are also difficult, there is great pressure on investigators. It therefore makes sense to choose the best methodology to determine whether there is a significant risk of stroke as a result of manipulation and there are a range of approaches to choose from. These include: hypothetical estimates of absolute risk (Klougart et al. 1996), retrospective surveys of practices (Dvorak & Orelli 1985), reviews of case reports (Ernst 2002), case-control studies (Rothwell et al., 2001, Smith et al., 2003 and Cassidy et al., 2008), prospective cohort studies without controls (Thiel et al., 2007 and Rubinstein et al., 2008) and prospective cohort studies with controls. Whichever technique is chosen, the rigour of both the research design and its execution will determine the authority with which the result is ultimately regarded (Levine et al. 1994). At the low end are estimates from numbers of patients treated against incidence from other studies. Then come estimates from case reports. Then probably come retrospective surveys of practices and unblinded prospective cohort studies without controls. Close to the top, as far as quality is concerned, are case-control studies that allow the inference of causation (and which give us our current best evidence) and, finally, prospective cohort studies with controls (of which none yet exist).
To study risk in a prospective cohort would probably be most feasible in a community stroke study, where strokes and transient ischaemic attacks, reported in primary care and confirmed by a neurologist, are investigated for exposure, or otherwise, to manipulation. (This would be superior to a study based on hospital discharge because many patients with mild strokes are never hospitalized.) These would then be compared with a matched population from the community, who have not had strokes, for exposure to manipulation. This helps to avoid errors in case ascertainment and failure to include all relevant cases (such as carotid strokes) as in retrospective hospital discharge studies (Cassidy et al. 2008). It also ensures that study populations are relevant and avoids having to use controls who have had other kinds of strokes (Smith et al. 2003). Finally, and importantly, it helps to avoid comparing cases and controls with different risk factors for cerebrobasilar stroke (Rothwell et al. 2001).
Such an opportunity did once present itself in a 10-year follow-up to the Oxford Community Stroke Study (Bamford et al. 1990), but was not taken up. In the meantime, although case-control studies do seem consistently to suggest that there is a relationship between manipulation and stroke, especially in younger patients with neck pain or headaches (Rothwell et al., 2001, Smith et al., 2003 and Cassidy et al., 2008); although it seems to have been established that the manipulation is unlikely to be causal. This was found by concurrently investigating the relationship to general practitioner visits and finding it to be similar. In other words (and as suggested 6 years before: Breen 2002), patients who are about to have a cerebrobasilar stroke may visit either their manipulation practitioner or their general practitioner and thereafter suffer the same event as part of natural history and not as a result of seeing either practitioner.
Nevertheless, this does not make the problem go away. Some people may not discriminate between manipulation causing a stroke and manipulation being associated with it for another reason. Others may ignore the evidence base and remain bound to their own preconceptions. The question may, therefore, continue to be raised in the popular press. Replication of the Cassidy study (Cassidy et al. 2008) may therefore be necessary before minds are set at rest and we can focus more on research that would help the patients who are at risk. This research should address the question of what good it would do to be able to recognize these patients in primary care practice. Are there interventions that could minimize progression to infarction if the risks were recognized in time? Are the risk factors sufficiently well known and accessible to be detected by practitioners? Could they be? A systematic review of risk factors for cervical artery dissection (Rubinstein et al. 2005) (Table 13.1) suggests that the main ones (increased aortic root diameter and change in carotid arterial diameter) could not be detected without vascular imaging studies, although the lesser ones (minor trauma, classical migraine and recent infection) certainly could.
| CI, confidence interval. | |
| Risk factor | Odds ratio (95% CI) |
|---|---|
| Aortic root diameter >34mm | 14.2 (3.2–63.6) |
| Carotid artery diameter change | 10.0 (1.8–54.2) |
| during cardiac cycle | |
| Migraine | 3.6 (1.5–8.6) |
| Trivial trauma (neck manipulation) | 3.8 (1.3–11.0) |
| Recent infection | 1.6 (0.7–3.8) |
| Homocysteine | (unknown) (1.1–1.5) |
The manipulation ‘package of care’
For people with spinal problems, the reduction of pain and the return to normal activity may also benefit any concurrent cardiovascular, mental and general health problems (Ong et al., 2004 and Waddell and Burton, 2004). This approach in rehabilitation implies that the practitioner who uses manipulation should also engage with the psychological and social factors that influence the course of chronic pain (Truchon 2001). This is very much in keeping with the wider biopsychosocial understanding of health that has come to be expected of most practitioners who deal with musculoskeletal conditions (Weigl et al. 2006). Thus, manipulation is not the mainstay of treatment for either of these conditions, but a useful addition to evidence-based care. This evidence-based management comes with a few simple messages:
• A case history and physical examination should be carried out.
• People should be given information and person-specific advice about how to manage pain and recover.
• Progress should be reviewed and management revised as needed.
• Most people are best managed in primary care.
If a person consults early in the course of the problem, the main consideration is to control pain while avoiding withdrawal and inactivity. If the patient’s condition is chronic, then practitioners are advised to assess prognostic factors and provide a consistent, cognitive behavioural approach.
Adherence to the evidence and doing it better
The practice styles of many manipulation practitioners already reflect this closely, but manipulation may be overutilized by them while pain control is underutilized. This may be rooted in the natural healing tradition of the professions and an in-principle aversion to drugs. However, there is evidence that many manipulation practitioners already have evidence-based care within their traditions. Indeed, before guidelines began to be published, a survey of 1012 chiropractic patients throughout Europe in the early 1990s (Pedersen 1994) had found that virtually all had received a comprehensive case history and examination and an explanation of their symptoms. Within 3 months of their first attendance, 65% of them had had reassessments recorded in their clinical records. Eighty-seven per cent were advised to stay active and only 2% and 5% (respectively) were advised to have bed rest and/or stay off work. However, only 4% were advised to take analgesics to control pain while almost all were given a manual treatment at most attendances. Explanation of symptoms, ergonomic advice and psychological and social advice were consistently provided for patients across the first five visits. In a subsequent study of UK chiropractors (Lauridsen 1999), 98% of respondents were found to use some form of exercises for chronic low-back pain. The use of X-rays, which is not recommended for back and neck pain unless pathology is suspected, has reduced sharply in chiropractic from an estimated 5112 new films annually in 1973 (Breen 1977) to only 158 in 2000 (Breen et al. 2000b).
Taking account of societal factors
With a few exceptions, therefore, the care traditionally offered by many practitioners of manipulation already reflects the evidence base. This assumes, however, that our biopsychosocial understanding of common health problems is wide enough. The overall cost of back pain alone is thought to be 2–3% of the gross domestic product of western European economies (Nachemson et al. 2000). In the UK in 2007, this was estimated at £1333 billion (National Statistics 2007), which would put this cost in the region of £27bn ($40bn). These are apparently mainly from lost production and other societal costs, especially in chronic cases (Clinical Standards Advisory Group Epidemiology Review 1994), while the total UK health care bill for back pain has been estimated at around £1bn (Maniadakis & Gray 2000). One explanation may be that it is the social and societal spectrum of problems associated with chronic back pain that is generating most of the cost and not just the back pain itself. If this is the case, a biomedical approach to back pain is not only inadequate (Foster et al. 2003), but fundamentally inappropriate. The social and societal factors are:
• work problems (unemployment, job dissatisfaction)
• financial problems (low economic status, low education)
• legal limitations (compensation pending and insurance rules)
• health care system (low priority, poor integration)
• social problems (poor housing, low support, low sense of control) (Pincus et al. 2008b).
These, together with other health problems, multisite pain and psychological problems (distress and depression), help to explain the high impact of back pain-related disability, which is not solvable by physical treatment alone. The sheer size of this impact means that the attention of policy-makers is urgently required and a multidisciplinary, biopsychosocial approach needs to be expected of all who attempt to help the sufferer. Among those who do offer such help are, or course, practitioners who use manipulation. Research has already shown, and may continue to reveal, to what extent practice remains aligned to the evidence, or even improves by addressing this wider spectrum of prognostic factors. Central to this will be whether manipulation remains the focus of their approach or becomes a tool within it.
Regular surveys of practice are crucial if the manipulating professions are to develop themselves optimally. Studies of patient cohorts could reveal the relationship between these new issues, such as the engagement of the practitioner with the wider context of health and its effect on clinical outcomes. However, this also calls for new outcome measures. How do we measure the impact of social and societal factors? To what extent do practitioners who use manipulation in primary care make an explicit attempt to improve them as part of their clinical management? There is reasonable international consensus that the above factors are the main ones (Pincus et al. 2008b) and that the appropriate method for measuring their effect sizes is inception cohort studies. However, we do not have instruments that measure, for example, ‘the influence of compensation litigation’ or ‘employer intransigence’. It is also worth finding out if practice in the private and public sectors is equally effective for patients.
Researching clinical effectiveness
The randomized trial is generally considered to be the best study design for comparing the effectiveness of treatments, but a truly valid comparison is sometimes very difficult, to the extent that some regard the randomized controlled trial as being of dubious value and potentially misleading (Bolton 1997). Koes (2004) outlined the main problems as:
• being able to compare populations that otherwise have the same prognosis
• standardizing the interventions
• blinding patients, therapists and outcome measurement
• recruiting an adequate sample size
• drop-outs and loss to follow-up.
Guidelines
The importance of having high-quality trials is that they form a large part of the basis of clinical guidelines. The consequences of adopting misleading conclusions from low-quality clinical trials surface particularly when guidelines based on them are monodisciplinary or based on trial evidence alone, to the exclusion of clinical expertise and patient preference (Sackett et al. 1996). In some back pain guidelines, for example, manipulation is not even mentioned (Philadelphia Panel, 2001 and Bekkering et al., 2003), despite the amount of evidence that surrounds it.
The potential consequences of low-quality guidelines are:
• poor patient care
• inflexibility
• disruption of practice
• threat to livelihood
• medicolegal liability (if not following) (Breen 2006).
Buy Membership for Complementary Medicine Category to continue reading. Learn more here
