Predicting a clinically important outcome in patients with low back pain following McKenzie therapy or spinal manipulation: a stratified analysis in a randomized controlled trial
© Petersen et al.; licensee BioMed Central. 2015
Received: 24 November 2014
Accepted: 11 March 2015
Published: 1 April 2015
Reports vary considerably concerning characteristics of patients who will respond to mobilizing exercises or manipulation. The objective of this prospective cohort study was to identify characteristics of patients with a changeable lumbar condition, i.e. presenting with centralization or peripheralization, that were likely to benefit the most from either the McKenzie method or spinal manipulation.
350 patients with chronic low back pain were randomized to either the McKenzie method or manipulation. The possible effect modifiers were age, severity of leg pain, pain-distribution, nerve root involvement, duration of symptoms, and centralization of symptoms. The primary outcome was the number of patients reporting success at two months follow-up. The values of the dichotomized predictors were tested according to the prespecified analysis plan.
No predictors were found to produce a statistically significant interaction effect. The McKenzie method was superior to manipulation across all subgroups, thus the probability of success was consistently in favor of this treatment independent of predictor observed. When the two strongest predictors, nerve root involvement and peripheralization, were combined, the chance of success was relative risk 10.5 (95% CI 0.71-155.43) for the McKenzie method and 1.23 (95% CI 1.03-1.46) for manipulation (P = 0.11 for interaction effect).
We did not find any baseline variables which were statistically significant effect modifiers in predicting different response to either McKenzie treatment or spinal manipulation when compared to each other. However, we did identify nerve root involvement and peripheralization to produce differences in response to McKenzie treatment compared to manipulation that appear to be clinically important. These findings need testing in larger studies.
KeywordsLow back pain McKenzie Spinal manipulation Predictive value Effect modification
The most recent published guidelines for the treatment of patients with persistent non-specific low back pain (NSLBP) recommend a program focusing on self-management after initial advice and information. These patients should also be offered structured exercises tailored to the individual patient and other modalities such as spinal manipulation [1,2].
Previous studies have compared the effect of the McKenzie-method, also known as Mechanical Diagnosis and Therapy (MDT), with that of spinal manipulation (SM) in heterogeneous populations of patients with acute and subacute NSLBP and found no difference in outcome [3,4].
Recently, the need for studies testing the effect of treatment strategies for subgroups of patients with NSLBP in primary care has been emphasized in consensus-papers [5,6] as well as the current European guidelines , based on the hypothesis that subgroup analyses, preferably complying with the recommendations of “Prognostic Factor Research”, will improve decision making towards the most effective management strategies. Although initial data show promising results, there is presently insufficient evidence to recommend specific methods of subgrouping in primary care [1,9].
Three randomized studies, comprising patients with predominantly acute or subacute low back pain (LBP), have tested the effects of MDT versus SM in a subgroup of patients that presented with centralization of symptoms or directional preference (favorable response to end range motions) during physical examination [10-12]. The conclusions drawn from these studies were not in concurrence and the usefulness was limited by a low methodological quality.
Our recent randomized study, comprising patients with predominantly chronic LBP (CLBP), found a marginally better overall effect of MDT versus SM in an equivalent group . In order to pursue the idea of subgrouping further, it was part of the study plan to explore predictors based on patient characteristics that could assist the clinician in targeting the most favorable treatment to the individual patient.
The objective of this study was to identify subgroups of patients with predominantly CLBP, presenting with centralization or peripheralization, which were likely to benefit from either MDT or SM two months after the completion of treatment.
The present study is a secondary analysis of a previously published randomized controlled trial . We recruited 350 patients from September 2003 through May 2007 at an outpatient back care centre in Copenhagen, Denmark.
Patients were referred from primary care physicians for treatment of persistent LBP. Eligible patients were between 18 and 60 years of age, suffering from LBP with or without leg pain for a period of more than 6 weeks, able to speak and understand the Danish language, and fulfilled the clinical criteria for centralization or peripheralization of symptoms during initial screening. Centralization was defined as the abolition of symptoms in the most distal body region (such as the foot, lower leg, upper leg, buttocks, or lateral low back) and peripheralization was defined as the production of symptoms in a more distal body region. These findings have previously been found to have acceptable degree of inter-tester reliability (Kappa value 0.64) . The initial screening was performed prior to randomization by a physical therapist with a diploma in the MDT examination system. Patients were excluded if they were free of symptoms at the day of inclusion, demonstrated positive non-organic signs , or if serious pathology, i.e. severe nerve root involvement (disabling back or leg pain in combination with progressive disturbances in sensibility, muscle strength, or reflexes), osteoporosis, severe spondylolisthesis, fracture, inflammatory arthritis, cancer, or referred pain from the viscera, was suspected based on physical examination and/or magnetic resonance imaging. Other exclusion criteria were application for disability pension, pending litigation, pregnancy, co-morbidity, recent back surgery, language problems, or problems with communication including abuse of drugs or alcohol.
Comparison of distribution of baseline variables between groups
Number of patients
Number of patients
RR (95% CI)
Duration of symptoms
More than a year
Sickleave past year
Six days or less¤¤¤
Nerve root involvement
Pain below the knee
Expectations to recovery
Expectations to work
After baseline measures were obtained, randomisation was carried out by a computer-generated list of random numbers in blocks of ten using sealed opaque envelopes.
Ethical approval of the study was granted by Copenhagen Research Ethics Committee, file no 01-057/03. All patients received written information about the study and gave their written consent prior to participation.
The practitioners performing the treatments had no knowledge of the results of the initial screening. The treatment programs were designed to reflect daily practice as much as possible. Detailed information on these programs have been published earlier .
The MDT treatment was planned individually following the therapist’s pre-treatment physical assessment. Specific manual vertebral mobilization techniques including high velocity thrust were not allowed. An educational booklet describing self care  or a “lumbar roll” for correction of the seated position was sometimes provided to the patient at the discretion of the therapist. In the SM treatment, high velocity thrust was used in combination with other types of manual techniques. The choice of combination of techniques was at the discretion of the chiropractor. General mobilizing exercises, i.e. self-manipulation, alternating lumbar flexion/extension movements, and stretching, were allowed but not specific exercises in the directional preference. An inclined wedged pillow for correction of the seated position was available to the patients if the chiropractor believed this to be indicated.
In both treatment groups, patients were informed thoroughly of the results of the physical assessment, the benign course of back pain, and the importance of remaining physically active. Guidance on proper back care was also given. In addition, all patients were provided with a Danish version of “The Back Book” which previously has been shown to have beneficial effect on patients’ beliefs about back pain . A maximum of 15 treatments for a period of 12 weeks were given. If considered necessary by the treating clinician, patients were educated in an individual program of self-administered mobilizing, stretching, stabilizing, and/or strengthening exercises at the end of the treatment period. Treatments were performed by clinicians with several years of experience. Patients were instructed to continue their individual exercises at home or at a gym for a minimum of two months after completion of the treatment at the back center. Because the patients suffered predominantly from CLBP we expected this period of self administered exercises to be necessary for the patients to experience the full effect of the intervention. Patients were encouraged not to seek any other kind of treatment during this two months period of self-administered exercises.
The primary outcome was the proportion of patients reporting success at follow-up two months after end of treatment. Treatment success was defined as a reduction of at least 5 points or a final score below 5 points on the 23-item modified Roland Morris Disability Questionnaire (RMDQ) . A validated Danish version of RMDQ was used . The definition of treatment success was based on the recommendations by others [21,22]. A sensitivity analysis using 30% relative improvement on RMDQ as definition of success was also performed. In accordance with the protocol , we considered a relative between-group difference of 15% in the number of patients with successful outcome to be minimal clinically important in our analysis of interaction.
Prespecified predictor variables
In order to reduce the likelihood of spurious findings , we restricted the number of candidate effect modifiers in the dataset to six. To increase the validity of our findings, a directional hypothesis was established for each variable according to the recommendations of Sun et al.  Four baseline variables have previously been suggested in randomized studies to be predictive of long term good outcome in patients with persistent LBP following MDT in comparison with strengthening training: centralization [25,26], or following SM in comparison to physiotherapy or treatment chosen by a general practitioner: age below 40 years [27,28], duration of symptoms more than 1 year , and pain below the knee . As recommended by others , another two variables were added based on the participating experienced clinicians’ judgments of which characteristics they would expect to predict good outcome from their treatment compared to the other. The additional variables prioritized by the physiotherapists in the MDT group were signs of nerve root involvement and substantial leg pain. The additional variables prioritized by the chiropractors in the SM group were no signs of nerve root involvement and not substantial leg pain.
In a supplementary analysis, we took the opportunity to explore whether the inclusion of further six baseline variables, assumed to have prognostic value for good outcome in either of the treatment groups, would appear to have an effect modifying effect as well. To our knowledge, no further variables from previous one arm studies have been reported to have prognostic value of long term good outcome in patients with persistent LBP following MDT, whereas three variables have been reported to have prognostic value following SM: male gender , mild disability , and mild back pain . Another three variables were agreed upon by the clinicians to be included in the supplementary analysis as they were assumed by experience from clinical practice to have prognostic value for good outcome regardless of treatment with MDT or SM: low number of days on sick leave past year, high patient expectations to recovery, and high patient expectations about coping with work tasks six weeks after initiation of treatment.
Dichotomization of possible predictor variables were made to allow for comparisons to be made with those of earlier studies. In cases where no cut off values could be found in the literature, dichotomization was performed above/below the median found in the sample. Definitions of variables are presented in the legend to Table 1.
The entire intention-to-treat (ITT) population was used in all the analyses. The last score was carried forward for subjects with missing two months RMDQ scores (7 patients in the MDT group and 14 patients in the SM group). In addition, a post hoc per protocol analysis was carried out comprising only those 259 patients that completed the full treatment. The analysis plan was agreed in advance by the trial management group.
The possible predictors were dichotomized and the chance of success was investigated by estimating the relative risk (RR) of success in each of the two strata. The impact of the investigated predictors was estimated by comparing the chance of success between the treatment groups when divided into the two strata. To test for treatment effect modification of the predictors we performed chi-squared tests for interaction between intervention and the two different strata for each of the predictors. This is basically the same as an interaction from a regression model. Confidence intervals were also inspected for potential clinically important effects.
Following the univariate analysis, a multivariate analysis was planned including effect modifiers with a p-value below 0.1.
Participants were similar with respect to socio-demographic and clinical characteristics at baseline in the treatment groups. An overview of the distribution of the included dichotomized variables at baseline is provided in Table 1. No differences were found between the treatment groups.
Overall, the post hoc per protocol analysis did not produce outcome results that were different from the results of the ITT analysis and therefore only the results of the ITT analysis will be reported.
None of the prognostic candidate variables explored in the supplementary analysis appeared to have any clinically important modifying effect (Additional file 1: Table S1).
The results from the sensitivity analysis using 30% relative improvement on RMDQ as definition of success were not markedly different from those presented above (Additional file 2: Table S2).
To our knowledge, this is the first study trying to identify effect modifiers when two mobilizing strategies, i.e. MDT and SM, are compared in a sample of patients with as changeable condition characterized by centralization or peripheralization.
Our study found that none of the potential effect modifiers were able to statistically significantly increase the overall effect of MDT compared to that of SM. However, the between-group difference for two of the variables exceeded our clinically important success-rate of 15% in number of patients with successful outcome, so our study is likely to have missed a true effect and, in that sense, did not have a large enough sample size.
The most apparent finding is that in our small subgroup of patients with signs of nerve root involvement, the relative chance of success appeared 1.89 times (2.31/1.22) higher than in patients with no nerve root involvement when treated with MDT, compared to those treated with SM. The difference was in the expected direction.
Although not statistically significant in our small sample, the variable peripheralization exceeded our clinically important success-rate of 15%, but was found not to be in the expected direction. No previous studies have assessed the effect modification of centralization or peripheralization in patients with CLBP. The RCT by Long et al. [25,26] concluded that patients with directional preference, including centralization, fared better 2 weeks after baseline than patients with no directional preference when treated with MDT in comparison with strengthening training. However, the outcome among peripheralizers was not reported, so the poor outcome reported in patients with no directional preference might be related to the subgroup of patients who responded with no change in symptoms during initial examination and not to those that responded with peripheralization. An alternative explanation might be that the effect modifying impact of centralization or peripheralization on MDT is dependent on the control treatment. Our findings suggest that future studies in this area need to involve predictive value of peripheralization as well as centralization.
When a composite of the two most promising predictors, peripheralization and signs of nerve root involvement, were present at baseline, the relative chance of success with MDT compared to SM appeared 8.5 times higher than for the subgroup with no centralization and nerve root involvement. The number of patients was very small and the confidence interval was wide. Therefore only a preliminary conclusion about interaction can be drawn and it calls for a validation in future studies.
In our study, there appeared to be no characteristic by which SM had better results compared to MDT. Thus, we could not support the results of two studies with a similar design as ours (two arms, sample of patients with persistent LBP, and outcome reported in terms of reduction of disability at long term follow up) [27,29]. In those studies, Nyiendo et al.  found a modifying effect of leg pain below knee on treatment by SM compared to that of the general practitioner six months after baseline, and Koes et al.  found a modifying effect of age below 40 years and symptom duration more than a year on treatment by SM compared to that of physiotherapy 12 months after baseline. However, results from those, as well as other previous RCTs comprising patients with persistent LBP, have supported our findings regarding the lack of effect modification of age [27,29,31], sex [29,31], baseline disability [27,29,31], and duration of symptoms , on SM when measured on reduction of disability 6-12 months after randomization. So, although evidence is emerging in patients with acute LBP regarding subgroup characteristics predictive of better results from SM compared to other types of treatment , we are still in the dark with respect to patients with persistent LBP.
The usefulness of choosing a criterion for success by combining an improvement of at least 5 points or an absolute score below 5 points on RMDQ is debatable. A total of 22 patients were considered successful based on score below 5 at follow up without having an improvement of at least 5 points. We therefore performed a sensitivity analysis using a relative improvement of at least 30% as criterion of success as recommended by others  (see Additional file 2: Table S2). As a result, the percentage of patients with successful outcome in the MDT group remained the same whereas 4 more patients were defined as successes in the SM group. Overall the sensitivity analysis did not produce outcome results that were markedly different from those of the primary analysis and therefore only those have been discussed above.
Strengths and limitations
This study used data from a RCT, whereas many others have used single arm designs not suitable for the purpose of evaluating treatment effect modification . In accordance with the recommendations by the PROGRESS group  we prespecified the possible predictors and also the direction of the effect. Furthermore, we limited the number of predictors included in order to minimize the chance of spurious findings.
The main limitation in secondary studies to previously conducted RCTs is that they are powered to detect overall treatment effect rather that effect modification. In recognition of the post hoc nature of our analysis, reflected in wide confidence intervals, we must emphasize that our findings are exploratory and require formal testing in a larger sample size.
In all subgroups, the probability of success with MDT was superior to that of SM. Although not statistically significant, the presence of nerve root involvement and peripheralization appear promising effect modifiers in favour of MDT. These findings need testing in larger studies.
The authors thank Jan Nordsteen and Steen Olsen for clinical expert advice, and Mark Laslett for comments and language correction.
This study was in part supported by grants from The Danish Rheumatism Association, The Danish Physiotherapy Organization, The Danish Foundation for Chiropractic Research and Continuous Education, and The Danish Institute for Mechanical Diagnosis and Therapy. RC/The Parker Institute acknowledge the funding support from the Oak Foundation. The funds were independent of the management, analyses, and interpretation of the study.
- Chou R, Qaseem A, Snow V, Casey D, Cross Jr JT, Shekelle P, et al. Diagnosis and treatment of low back pain: a joint clinical practice guideline from the American College of Physicians and the American Pain Society. Ann Intern Med. 2007;147(7):478–91.View ArticlePubMedGoogle Scholar
- NHS. Early management of persistent non-specific low back pain. NICE Clinical Guideline. 2009;88:1–30.Google Scholar
- Cherkin DC, Battie MC, Deyo RA, Street JH, Barlow W. A comparison of physical therapy, chiropractic manipulation, and provision of an educational booklet for the treatment of patients with low back pain. N Engl J Med. 1998;339(15):1021–9.View ArticlePubMedGoogle Scholar
- Paatelma M, Kilpikoski S, Simonen R, Heinonen A, Alen M, Videman T. Orthopaedic manual therapy, McKenzie method or advice only for low back pain in working adults. A randomized controlled trial with 1 year follow-up. J Rehabil Med. 2008;40(10):858–63.View ArticlePubMedGoogle Scholar
- Foster NE, Dziedzic KS, van Der Windt DA, Fritz JM, Hay EM. Research priorities for non-pharmacological therapies for common musculoskeletal problems: nationally and internationally agreed recommendations. BMC Musculoskelet Disord. 2009;10:3.View ArticlePubMedPubMed CentralGoogle Scholar
- Kamper SJ, Maher CG, Hancock MJ, Koes BW, Croft PR, Hay E. Treatment-based subgroups of low back pain: a guide to appraisal of research studies and a summary of current evidence. Best Pract Res Clin Rheumatol. 2010;24(2):181–91.View ArticlePubMedGoogle Scholar
- Airaksinen O, Brox JI, Cedraschi C, Hildebrandt J, Klaber-Moffett J, Kovacs F, et al. Chapter 4. European guidelines for the management of chronic nonspecific low back pain. Eur Spine J. 2006;15 Suppl 2:S192–300.View ArticlePubMedPubMed CentralGoogle Scholar
- Hingorani AD, Windt DA, Riley RD, Abrams K, Moons KG, Steyerberg EW, et al. Prognosis research strategy (PROGRESS) 4: Stratified medicine research. BMJ. 2013;346:e5793.View ArticlePubMedPubMed CentralGoogle Scholar
- Fersum KV, Dankaerts W, O’Sullivan PB, Maes J, Skouen JS, Bjordal JM, et al. Integration of sub-classification strategies in RCTs evaluating manual therapy treatment and exercise therapy for non-specific chronic low back pain (NSCLBP): a systematic review. Br J Sports Med. 2010;44(14):1054–62.View ArticlePubMedGoogle Scholar
- Erhard RE, Delitto A, Cibulka MT. Relative effectiveness of an extension program and a combined program of manipulation and flexion and extension exercises in patients with acute low back syndrome. Phys Ther. 1994;74(12):1093–100.PubMedGoogle Scholar
- Schenk RJ, Josefczyk C, Kopf A. A randomized trial comparing interventions in patients with lumbar posterior derangement. J Man Manipul Ther. 2003;11(2):95–102.View ArticleGoogle Scholar
- Kilpikoski S, Alen M, Paatelma M, Simonen R, Heinonen A, Videman T. Outcome comparison among working adults with centralizing low back pain: Secondary analysis of a randomized controlled trial with 1-year follow-up. Adv Physiol Educ. 2009;11:210–7.Google Scholar
- Petersen T, Larsen K, Nordsteen J, Olsen S, Fournier G, Jacobsen S. The McKenzie method compared with manipulation when used adjunctive to information and advice in low back pain patients presenting with centralization or peripheralization. A randomized controlled trial. Spine (Phila Pa 1976). 2011;36(24):1999–2010.View ArticleGoogle Scholar
- Petersen T, Olsen S, Laslett M, Thorsen H, Manniche C, Ekdahl C, et al. Inter-tester reliability of a new diagnostic classification system for patients with non-specific low back pain. Aust J Physiother. 2004;50:85–94.View ArticlePubMedGoogle Scholar
- Waddell G, McCulloch JA, Kummel E, Venner RM. Nonorganic physical signs in low-back pain. Spine. 1980;5(2):117–25.View ArticlePubMedGoogle Scholar
- Manniche C, Asmussen K, Lauritsen B, Vinterberg H, Kreiner S, Jordan A. Low Back Pain Rating scale: validation of a tool for assessment of low back pain. Pain. 1994;57(3):317–26.View ArticlePubMedGoogle Scholar
- McKenzie RA. Treat your own back. Waikanae: Spinal Publications New Zealand Ltd; 1997.Google Scholar
- Burton AK, Waddell G, Tillotson KM, Summerton N. Information and advice to patients with back pain can have a positive effect. A randomized controlled trial of a novel educational booklet in primary care. Spine. 1999;24(23):2484–91.View ArticlePubMedGoogle Scholar
- Patrick DL, Deyo RA, Atlas SJ, Singer DE, Chapin A, Keller RB. Assessing health-related quality of life in patients with sciatica. Spine. 1995;20(17):1899–908.View ArticlePubMedGoogle Scholar
- Albert H, Jensen AM, Dahl D, Rasmussen MN. Criteria validation of the Roland Morris questionnaire. A Danish translation of the international scale for the assessment of functional level in patients with low back pain and sciatica [Kriterievalidering af Roland Morris Spørgeskemaet - Et oversat internationalt skema til vurdering af ændringer i funktionsniveau hos patienter med lændesmerter og ischias]. Ugeskr Laeger. 2003;165(18):1875–80.PubMedGoogle Scholar
- Bombardier C, Hayden J, Beaton DE. Minimal clinically important difference. Low back pain: outcome measures. J Rheumatol. 2001;28(2):431–8.PubMedGoogle Scholar
- Ostelo RW, Deyo RA, Stratford P, Waddell G, Croft P, Von KM, et al. Interpreting change scores for pain and functional status in low back pain: towards international consensus regarding minimal important change. Spine. 2008;33(1):90–4.View ArticlePubMedGoogle Scholar
- Moons KG, Royston P, Vergouwe Y, Grobbee DE, Altman DG. Prognosis and prognostic research: what, why, and how? BMJ. 2009;338:1317–20.View ArticleGoogle Scholar
- Sun X, Briel M, Walter SD, Guyatt GH. Is a subgroup effect believable? Updating criteria to evaluate the credibility of subgroup analyses. BMJ. 2010;340:c117.View ArticlePubMedGoogle Scholar
- Long A, Donelson R, Fung T. Does it matter which exercise? A randomized control trial of exercise for low back pain. Spine. 2004;29(23):2593–602.View ArticlePubMedGoogle Scholar
- Long A, May S, Fung T. The comparative prognostic value of directional preference and centralization: a useful tool for front-line clinicians? J Man Manip Ther. 2008;16(4):248–54.View ArticlePubMedPubMed CentralGoogle Scholar
- Koes BW, Bouter LM, van Mameren H, Essers AH, Verstegen GJ, Hofhuizen DM, et al. A randomized clinical trial of manual therapy and physiotherapy for persistent back and neck complaints: subgroup analysis and relationship between outcome measures. J Manipulative Physiol Ther. 1993;16(4):211–9.PubMedGoogle Scholar
- Leboeuf-Yde C, Gronstvedt A, Borge JA, Lothe J, Magnesen E, Nilsson O, et al. The nordic back pain subpopulation program: demographic and clinical predictors for outcome in patients receiving chiropractic treatment for persistent low”back pain. J Manipulative Physiol Ther. 2004;27(8):493–502.View ArticlePubMedGoogle Scholar
- Nyiendo J, Haas M, Goldberg B, Sexton G. Pain, disability, and satisfaction outcomes and predictors of outcomes: a practice-based study of chronic low back pain patients attending primary care and chiropractic physicians. J Manipulative Physiol Ther. 2001;24(7):433–9.View ArticlePubMedGoogle Scholar
- Foster NE, Hill JC, Hay EM. Subgrouping patients with low back pain in primary care: are we getting any better at it? Man Ther. 2011;16(1):3–8.View ArticlePubMedGoogle Scholar
- Underwood MR, Morton V, Farrin A. Do baseline characteristics predict response to treatment for low back pain? Secondary analysis of the UK BEAM dataset. Rheumatology (Oxford). 2007;46(8):1297–302.View ArticleGoogle Scholar
- Slater SL, Ford JJ, Richards MC, Taylor NF, Surkitt LD, Hahne AJ. The effectiveness of sub-group specific manual therapy for low back pain: a systematic review. Man Ther. 2012;17(3):201–12.View ArticlePubMedGoogle Scholar
- Stanton TR, Hancock MJ, Maher CG, Koes BW. Critical appraisal of clinical prediction rules that aim to optimize treatment selection for musculoskeletal conditions. Phys Ther. 2010;90(6):843–54.View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.