Volume 24, Issue 1 , Pages 63-66, January 2001
Low back pain: Time to get off the treadmill☆
Article Outline
- Introduction
- The 4 basic causal models
- A proposed fifth causal model: The subgroup concept
- A strategy for identifying subgroups of LBP
- Closing the gap
- Acknowledgements
- References
- Copyright
Introduction
The present conceptual framework pertaining to low back pain (LBP) is useless for the individual practitioner. Granted, LBP is difficult to study and understand. It is often labeled “nonspecific,” “idiopathic,” “mechanical,” or “activity-related,” because it is usually impossible to make a specific diagnosis on the basis of obvious pathologic findings. Clearly, there have been some fundamental changes in the conceptual framework surrounding nonspecific LBP. The pathoanatomical and monocausal explanatory model (eg, spondylolisthesis) has largely given way to the biopsychosocial, and hence multicausal, concept.1
However, the literature in this area is vast and confusing, and though we have some knowledge about the causes of LBP from a public health perspective, we cannot use much of this information on an individual level. For example, it is still not possible to predict who will develop LBP or to determine what the reasons for such development are. Obviously, this renders successful primary prevention impossible.
It is also not known why some people recover fully from an attack of acute LBP and never or rarely experience it again whereas other people have numerous bouts of LBP and may even never recover at all. This renders prediction and secondary prevention impossible.
On the therapeutic side, we have lately seen a paradigm shift toward fitness, exercise therapy, spinal manipulation, and even no treatment at all (merely providing fear-reducing information and patient activation). However, although these methods of approach have been shown to produce positive results in some clinical studies, it is obvious that they do not suit all patients. In other words, we do not know, on an individual level, which patients can be helped, and of those who can be helped we do not know whether the choice of therapy will play any role in the treatment outcome. This renders prescription of treatment uncertain, inasmuch as the process is based on personal experience and intuition.
It is tempting to conclude that to get more streamlined information on the causes and cures of LBP, we simply require more research. However, before embarking on yet another epidemiologic or clinical study on LBP, we must consider the conceptual frame within which we operate. For that reason, it is interesting to look at this issue in some detail.
The 4 basic causal models
The way that we look at LBP, as a disease entity and in relation to its development, will determine our research strategy. Most of the research on LBP can be seen in terms of 4 basic causal models: (1) a single cause, a single disease; (2) a single cause, several diseases; (3) several causes, a single disease; (4) several causes, several diseases. Examples of how these models are used in epidemiologic research are given in Table 1.
Table 1. Examples of common causal models in relation to epidemiologic study design
| Causal model | Research question | Study design |
|---|---|---|
| Model 1 (one cause, one disease) | Does smoking cause LBP? | The presence of LBP is compared in smokers and nonsmokers; if it is more common in smokers (after controlling for various suspected confounders/modifiers) it is suggested that smoking may cause LBP. |
| Model 2 (several causes, one disease) | Do smoking and obesity cause LBP? | As in Model 1; however, both suspected risk factors must be positively associated with LBP. |
| Model 3 (one cause, several diseases) | Does smoking cause LBP and sciatica? | As in Model 1; however, the suspected risk factor must be positively associated with LBP and sciatica. |
| Model 4 (several causes, several diseases) | Do smoking and obesity cause LBP and sciatica? | As in Model 1; however, both suspected risk factors must be associated with LBP and sciatica. |
Clinical trials conducted by researchers subscribing to any of the single-disease models would use wide inclusion criteria and probably not stratify the study sample at the time of randomization or analysis. On the other hand, researchers who believe that there are several diseases (such as sciatica and LBP, or acute LBP and chronic LBP) are more likely to use stringent inclusion criteria because they wish to capture a specific patient group; alternatively, they attempt to divide the different diagnostic groups equally among the treatment groups.
A proposed fifth causal model: The subgroup concept
It is rare that researchers consider a fifth causal model— namely, that nonspecific LBP is made up of several distinct subentities, each with its own set of causes and thus with its own set of cures, as illustrated in Fig 1.
Fig. 1.
Nonspecific low back pain (large circle) may consist of a number of largely unidentified subentities (smaller circles LBP1, LBP2, etc), each having its own set of causal mechanisms (C1, C2, etc).
Fig. 2.
Sampling method in one study may result in a disproportionately large number of people from one or several specific subentities of LBP.
Fig. 3.
Sampling method in another study may result in a disproportionately large number of people from some other specific subentities of LBP
More studies based on any of the 4 main causal models would therefore simply add to the confusion by producing different outcomes depending on which particular subgroups of LBP happened to be included in the individual studies. Such a disproportionality of subgroups is likely to be the result when different target populations are used (such as different specific occupational groups or different particular clinical populations), when the sampling method of general populations is inadequate, or when the sample size is too small to give all of the subentities a chance to participate.
The following example is provided: when the therapeutic value of physical training is being tested, the results can be expected to be excellent if the study sample largely consists of a subgroup of patients who developed their LBP from physical inactivity; however, the results would probably go in the opposite direction if the study sample happened to contain a large proportion of people who developed their LBP from physical overactivity. Hence, the study might test not the usefulness of the therapy per se but rather the usefulness of the study sample!
From this example we can infer that if these LBP subgroups exist, they should be identified. Otherwise, we will continue to produce nonsense instead of information and fail to understand why. However, identification of the subgroups is rather difficult, because we are dealing with several “unknowns”: we do not know the cause of LBP for any subgroup, and we do not know the best treatment for any subgroup. Nonetheless, it is possible to make a systematic search for specific subgroups of nonspecific LBP.
A strategy for identifying subgroups of LBP
Subgroups could be identified through (1) studies of their association with risk factors, (2) observations of their clinical picture, recovery pattern and natural course, (3) experimental studies, (4) outcome studies, and (5) randomized controlled clinical trials. For some examples, see Table 2.
Table 2. Suggested methods of identifying subgroups of LBP
| Method | Description |
|---|---|
| Identifying subgroups through their association with nonspecific LBP. | It might be possible to identify specific subgroups of LBP by studying specific risk factors in groups of people Perhaps it can be shown that this previously amorphous group contains specific with risk factors subentities having features such as “inactivity-induced” LBP, “overactivity-induced” LBP, or “posture-induced” LBP. |
| Identifying subgroups by observing the clinical picture or objective findings | More complex analyses, such as cluster analysis, could help identify specific subgroups of LBP in relation to pain patterns or the symptomatic picture, such as “worse with rest,” “worse with activities,” or “worse with certain postures/tasks.” By following specific cohorts on the basis of well-defined events, it might be possible to identify prognostic subgroups, such as the circumstances surrounding first-time attacks of LBP—eg, “severe injury,” “multiple severe injuries,” “minor injury,” “minor multiple injuries,” “injury when not warmed up,” or “injury when tired.” The definition of subgroups could also be attempted through experimental studies, for example by investigating specific biochemical findings, pain response, or tissue changes. |
| Identifying subgroups through outcome studies | Another method of identifying subgroups would be through the systematic use of outcome studies for various therapies. Here it would be possible not only to (a) distinguish among those who recover quickly, those who recover slowly, and those who recover not at all and (b) describe their characteristics, but also to see if some of these characteristics are specific for specific therapies. |
| Identifying subgroups through randomized controlled clinical trials | It would be possible to test whether outcomes vary for several presumed subentities of LBP by submitting these to different therapeutic approaches—eg, to determine whether physical training controlled clinical trials and spinal manipulation work differently for different subgroups. It would also be possible to do subgroup analysis in randomized controlled clinical trials with large study samples. |
It is unusual to find any of these study approaches in the LBP literature, but some examples of research carried out in accordance with the fifth model have been listed in Table 3.
Table 3. Examples of studies that attempted to identify subgroups of LBP other than through use of the usual 4 causal models
| Identifying subgroups through their association with risk factors | In a prospective study, it was shown that heavy lifting and much standing was a predictor of low back pain in smokers but not in nonsmokers. It was hypothesized that reduced perfusion and malnutrition of tissues around the spine render these vulnerable to mechanical stress.2 A monotonic dose-response curve was found between body weight and LBP for people with longer-lasting or recurrent LBP in comparison with those with shorter or fewer attacks.3 Investigation of the associations between nonspecific LBP and various potential risk factors showed that there was usually no positive association between these and LBP. However, when nonspecific LBP was divided into subgroups, previously unidentified specific profiles emerged.4 |
| Identifying subgroups by observing the clinical picture or objective findings | Through use of interviews and physical examinations, it was possible to identify 7 clinical subgroups of “organic” localized LBP after those with a diagnosis of psychiatric disorder were excluded.5 Pain behavior subgroups were identified through pain ratings and observations of pain behavior.6 By identifying aggravating, relieving, and cyclic features in case histories of patients with back pain, it was possible to group these into 2 main categories (one with mechanical features and one that was cyclic).7 A subset of patients with chronic LBP having internal disk disruption was identified through the use of computed tomography diskography.8 |
| Identifying subgroups through outcome studies | A number of nonorganic signs were shown to be particularly common in “problem” patients and absent in normal subjects. These findings appeared to be completely independent of the conventional symptoms and signs of pathologic conditions of the spine.9 From a 1-year follow-up study of patients in a subacute phase after a first episode of back pain, it was possible to identify different recovery patterns.10 |
| Identifying subgroups through controlled clinical trials | A separate analysis was made of responders and nonresponders participating in a clinical trial on intensive dynamic back exercises for chronic LBP. The outcome of treatment was not influenced by age, sex, duration, degree of severity of LBP, preexisting sciatica, or pathological radiographic findings.11 Subgroup analysis of a randomized clinical trial between chiropractic and physiotherapy indicated that patients with acute, uncomplicated problems may have had more benefit from chiropractic treatment whereas patients who had had their problems for >1 month may have had more benefit from physiotherapy.12 |
Closing the gap
Ideally, when a number of logical, reproducible, and clinically relevant subgroups of LBP have been identified, they should be tested against suitable treatment methods, such as exercise for the inactivity-induced LBP, rest for the overactivity-induced LBP, ergonomic changes for the posture-induced LBP, spinal manipulation for the mechanical-features LBP, anti-inflammatory treatment for the cyclic LBP, and pain behavior modification for patients who exhibit inappropriate pain behavior. Perhaps there are also a “no-apparent-reasons” LBP and a “bad-luck” (genetic) LBP that are best left alone.
In this way, it might be possible to establish the links between the cause, the disease, and the cure!
Acknowledgements
A. Jordan, DC, PhD, is gratefully acknowledged for editorial comments.
References
- . Biopsychosocial analysis of low back pain. Baillieres Clin Rheumatol. 1992;6:533–557
- . Smoking, heavy physical work and low back pain: a four-year prospective study. Occup Med. 1999;49:155–160
- . Low back pain and lifestyle, II: obesity—information from a population-based sample of 29,424 twin subjects. Spine. 1999;24:779–784
- . Why has the search for causes of low back pain largely been nonconclusive?. Spine. 1997;22:877–881
- . Classification of nonspecific low back pain, II: clinical diversity of organic forms. Spine. 1992;17:1038–1042
- . Behavioral assessment of low back pain: identification of pain behavior subgroups. Pain. 1990;40:153–160
- . Rationalizing back pain: the development of a classification system through cluster analysis. J Manipulative Physiol Ther. 1997;20:303–310
- . The prevalence and clinical features of internal disc disruption in patients with chronic low back pain. Spine. 1995;17:1878–1883
- . Nonorganic physical signs in low-back pain. Spine. 1980;6:117–125
- One-year follow-up of first onset low back pain. Pain. 1997;73:213–221
- . Intensive dynamic back exercises for chronic low back pain: a clinical trial. Pain. 1991;47:53–61
- . Cost and effectiveness analysis of chiropractic and physiotherapy treatment for low back and neck pain: six-month follow-up. Spine. 1997;22:2167–2177
☆ bBack Clinic, Ringe Hospital and University of Southern Denmark.
PII: S0161-4754(01)25939-9
doi:10.1067/mmt.2001.112009
© 2001 JMPT. Published by Elsevier Inc. All rights reserved.
Volume 24, Issue 1 , Pages 63-66, January 2001
