J Manipulative Physiol Ther. 2005 (Mar); 28 (3): 153–158 ~ FULL TEXT
Iben Axen, DC, Annika Rosenbaum, BAppSc(Chiro), Robert Robech, MHSc(Clin Biomech),
Kristian Larsen, MPH, Charlotte Leboeuf-Yde, DC, MPH, PhD
Medical Research Unit
Ringkjøbing County, Denmark.
OBJECTIVE: To investigate whether 3 distinct patterns of reactions to chiropractic care can predict early favorable treatment outcome in patients with nonpersistent low back pain (LBP).
DESIGN: Multicenter practice-based predictive validity study.
Sixty private practice chiropractors in Sweden recruited a maximum of 20 consecutive patients with LBP each, with a duration of less than 2 weeks at the time of consultation and a maximum of 30 days totally over the past year.
METHODS: Chiropractic management was decided on by the treating chiropractor. The outcome variable was self-reported "definite improvement" at the fourth visit. The predictor variables included 3 hypothesized prognostic groups (best, intermediate, and least favorable) on the basis of clinical information collected at baseline and at the second visit. The covariates included age, sex, pain intensity during the past 24 hours, description of disability, duration and pattern of pain during the present attack, and duration and pattern of pain during the past 12 months. The 3 predictor groups were cross-tabulated against the outcome variable and the other covariates. Backward stepwise logistic regression was performed to test for confounding or modification from relevant covariates.
RESULTS: Information was provided on 708 patients, of which 674 questionnaires were valid. Of the 223 patients in the hypothesized best prognostic group, 91% (95% CI, 79-100) reported to be "definitely improved" by the fourth visit, vs 76% (72-80) of the 420 patients in the intermediate prognostic group, and 36% (19-53) of the 31 patients in the least favorable prognostic group. These results were not altered after controlling for the covariates.
CONCLUSION: For chiropractic patients with nonpersistent LBP, these findings show that it is possible to predict already by the second visit which patients may or may not report improvement at the fourth visit.
Key Indexing Terms: Low Back Pain, Manipulation, Chiropractic, Prognosis, Treatment Outcome, Predictive Validity
From the FULL TEXT Article:
There is a need to identify and describe specific subgroups of individuals with low back pain (LBP) to better understand the possibilities for prevention and treatment. Presently, in the scientific and clinical literature, emphasis is put on psychosocial factors that have been found to act as barriers against improvement. However, most clinicians probably feel frustrated about the lack of information in relation to clinical factors that may predict treatment outcome. As far as we know, there have been no large broad-based clinical studies that looked specifically at clinical predictors for treatment outcome with spinal manipulative therapy and/or chiropractic management.
In Sweden and Norway, a research program has been running over the past years, in which specific subgroups of patients with LBP are systematically studied in a concerted effort to obtain more information in this area, with focus on the duration and frequency of symptoms.
For example, in a recent Swedish multicenter study, clinical information was gathered about patients who at the time of consultation had experienced LBP for a minimum of 2 weeks and during the past year for a total of at least 30 days. In other words, these were patients with at least one longer episode of LBP or several longer or shorter episodes, adding up to at least 30 days. This group was described as having “persistent LBP” (P-LBP). In that study,  on the basis of clinical experience, 3 distinct prognostic subgroups were identified a priori and their validity tested.
The first group consisted of those who react positively directly after treatment, who report a common unpleasant reaction (or no unpleasant reaction at all) after the first treatment, and who report improvement as early as the second visit. Most of the patients with P-LBP in this hypothesized most favorable prognostic group (77% to 91%) reported considerable improvement by the fourth visit.
Another group consisted of those who report no immediate improvement but also no common unpleasant reaction (or possibly an uncommon unpleasant reaction), and also no improvement at the second visit. Only 22% to 38% of patients with P-LBP in this hypothesized least favorable prognostic group reported considerable improvement by the fourth visit. The group in between these 2 consisted of patients who fit into neither group, that is, those with a mixed pattern. In this mixed prognostic group, 58% to 68% reported considerable improvement by the fourth visit. The 3 hypothesized prognostic groups were therefore found to be clinically valid.
Interestingly, it was noted that the 2 extreme groups (hypothesized good and poor prognosis) consisted of almost equal number of subjects (115 and 116 patients, respectively), whereas the mixed group was approximately 3 times as large.
In addition, knowledge was gained relative to the baseline profile of patients with P-LBP in relation to pain (38% reported the pain to be bad), and disability in relation to the most frequently reported difficult aspects of daily life. These were in order of frequency:
(1) getting up from sitting,
(2) putting on socks and shoes,
(3) turning in bed,
(4) going for walks, and
The number of people who reported problems with these items ranged from 76% (getting up from sitting) to 38% (sleeping).
The reason for selecting this particular definition for P-LBP is that people with LBP of a total duration of at least 30 days in the preceding year in previous epidemiologic studies have been shown to have a different profile in relation to certain risk indicators for LBP, such as smoking,  obesity,  attitudes,  and heavy work. [4, 5]
It is therefore logical to assume that this group of individuals with more persistent LBP as opposed to those with nonpersistent LBP (NP-LBP) represents a specific subpopulation, also in relation to clinical matters.
After completion of the first study, our next question was, “are the baseline and improvement profiles different for people with NP-LBP who present acutely for chiropractic treatment?”
A study of near-similar design was therefore carried out to study this other specific subpopulation of patients with LBP. In addition, the outcomes were compared for patients who sought care in the first week and the second week after the onset of the problem, and the profile of the present NP-LBP study sample was compared to that of the previous P-LBP study sample.
There is a need to learn more about specific subpopulations of LBP to assist clinicians to select and manage their patients and to provide a basis for selection of patients in randomized controlled clinical trials. The present study was the second attempt in the Nordic LBP subpopulation study program to define predictors of treatment outcome in a specific LBP population and to try to describe specific subpopulations. Another 2 reports are presently underway from a similar Norwegian practice-based study.
Patients with a maximum LBP duration of 2 weeks at consultation, and, in total, a maximum of 30 days over the past year were submitted to chiropractic treatment. This group can aptly be described as “acute,” and the limited number of days with LBP in the past year indicates that their condition is more benign, in the sense that it has not been a long-lasting or frequently occurring event in the past year.
As in our previous study on P-LBP, we used a convenience-sample of chiropractors. The representativeness of our study sample (and therefore the generalizability of our results) is therefore contingent upon their patients and their recovery pattern not being linked to the chiropractors' willingness to participate in studies. In both studies, nonvalidated questionnaires were used and data were collected by the treating chiropractors. A discussion of the rationale behind this is found in our previous report on P-LBP.  Obviously, subjective data collected by the treating clinician will always carry the risk of misclassification. Attempts were made to lessen this risk by allowing only the most positive score on a 5-item list (“definitely better”) to be defined as improvement.
The present study sample, compared to that in the study on P-LBP, consisted of a larger proportion of men (58% vs 48%), but the spread of age groups was similar in the 2 studies. Not surprisingly, the pain was more often described as very bad (52% vs 38%), and a larger proportion of patients considered themselves disabled in relation to common activities of daily living (getting up from sitting, putting on socks and shoes, and turn in bed) in the NP-LBP group. Of those with NP-LBP, 79% reported “considerable improvement” vs 56% of those with P-LBP, and as a unique feature for those with NP-LBP, approximately two thirds improved considerably already before the fourth visit. The time factor was not investigated in this study, but it would be expected that this type of patient attends the first 4 treatment sessions within 2, possibly 3 weeks.
As for the patients in the past study of P-LBP, there were 3 distinct prognostic subgroups of patients also among our patients with NP-LBP. Almost all (91%) in the hypothesized best prognostic group reported definite improvement vs 36% in the poor prognosis group, whereas the mixed group fell in between. Interestingly, approximately two thirds of patients with NP-LBP “disappeared” from the practice before the fourth visit, in most cases, because they had recovered and did not require further treatment. This fact needs to be taken into account in clinical practice, as a minimum number of treatments obviously is an attractive economic argument, but it should also be considered when conducting randomized controlled clinical trials. Obviously, in this subgroup of patients, follow-up should take place already within the first week, preferably on a day-to-day basis, to verify any possible increased recovery rate with spinal manipulative therapy as compared to that of the natural course.
The OR for predicting improvement was high in NP-LBP, but further statistical “descriptions” revealed that this model was better at identifying nonresponders (99%) than responders (only 8%). Although the proportion of subjects correctly classified was high (80%), the global assessment of the performance of this model (diagnostic accuracy) is only moderate (68%),  indicating that other factors also influence the treatment outcome. However, we were unable to detect these, as none of the other factors that we tested in the multi variable analysis influenced the treatment outcome.
Interestingly, according to the bivariate analysis, patients with NP-LBP who sought care within the first week of onset, as opposed to the second week, had twice the recovery rate. This could indicate that the foundation for P-LBP is already laid in the second week, either because treatment at that time is less efficient or because LBP that persists into the second week per se is of a more long-lasting nature. The present study design, obviously, cannot provide answers to such hypotheses.
In summary, a previous study indicates that clinicians can predict treatment outcome in P-LBP at the patient's second visit. The present study tests this ability in a different LBP subpopulation (NP-LBP). The results show that the initial findings are robust but that there are differences between the 2 subpopulations. In this second study, it was also investigated if the original clinicians' simple model could be improved or taken apart by the introduction of other clinical factors. None of the covariates included in the analysis managed to alter the original results considerably.
As for patients with P-LBP, it is possible to predict at the second visit which patients with NP-LBP will not improve at the fourth visit with chiropractic care. However, chiropractic patients with NP-LBP are different from patients with P-LBP. A larger proportion are men, they are more disabled in their simple activities of daily living, recovery is more common, and it occurs at a faster rate. Speedy recovery from single LBP episodes at low treatment cost has important implications on overall health care costs, particularly in occupations that require time off work during the acute LBP stage. This aspect therefore requires more attention both in practice management and clinical research.
Axén, I., Rosenbaum, A., Röbech, R., Wren, T., and Leboeuf-Yde, C.
Can Patient Reactions to the First Chiropractic Treatment
Predict Early Favorable Treatment Outcome in Persistent Low Back Pain?
J Manipulative Physiol Ther. 2002 (Sep); 25 (7): 450–454
Leboeuf-Yde, C., Ohm Kyvik, K., and Bruun, H.
Low back pain and life-style. Part I: Smoking. Information from a population-based sample of 29,424 twin individuals.
Spine. 1998; 23: 2207–2214
Leboeuf-Yde, C., Ohm Kyvik, K., and Bruun, H.
Low back pain and life-style. Part II: Obesity. Information from a population-based sample of 29,424 twin individuals.
Spine. 1999; 24: 779–784
Leboeuf-Yde, C., Lauritsen, J., and Lauritzen, T.
Why has the search for causes of low back pain largely been nonconclusive?.
Spine. 1997; 22: 877–881
Hartvigsen, J., Bakketeig, L.S., Leboeuf-Yde, C., Engberg, M., and Lauritzen, T.
The relationship between physical work-load and low-back pain clouded by the “healthy worker” effect. A population based cross-sectional and 5 year prospective questionnaire study.
Spine. 2001; 26: 1788–1793
Leboeuf-Yde, C., Axén, I., Ahlefeldt, G., Lidefelt, P., Rosenbaum, A., and Thurnherr, T.
The types and frequencies of improved nonmusculoskeletal symptoms reported after chiropractic spinal manipulative therapy.
J Manipulative Physiol Ther. 1999; 22: 559–564
Leboeuf-Yde, C., Hennius, B., Rudberg, E., Leufvenmark, P., and Thunman, M.
Side effects of chiropractic treatment: a prospective study.
J Manipulative Physiol Ther. 1997; 20: 511–515
Senstad, O., Leboeuf-Yde, C., and Borchgrevink, C.
Frequency and characteristics of side effects of spinal manipulative therapy.
Spine. 1997; 22: 435–441
in: D.G. Altman, D. Machin, T.N. Bryant, M.J. Gardner (Eds.)
Statistics with confidence. 2nd ed.
BMJ Book, JW Arrowsmith Ltd., Bristol (UK); 2002: 111–116
Return to the LOW BACK PAIN Section