Article Text

Download PDFPDF

A clinical prediction rule predicted outcome in patients with low back pain having spinal manipulation and exercise treatment
  1. Martin Underwood, MD
  1. Barts and The London
 London, UK

    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

 Q In patients with low back pain (LBP), does a clinical prediction rule predict outcome of spinal manipulation?

    Clinical impact ratings GP/FP/Primary care ★★★★★☆☆ Rheumatology ★★★★★☆☆


    Embedded ImageDesign:

    a randomised controlled trial to validate a previously developed clinical prediction rule for predicting outcome of spinal manipulation.

    Embedded ImageSetting:

    8 physical therapy clinics in 2 academic medical centres and outpatient practice settings in the US.

    Embedded ImagePatients:

    131 patients who were 18–60 years of age (mean age 34 y, 58% men), had a primary symptom of LBP, were referred to physical therapy, and had an Oswestry Disability Questionnaire (ODQ) score ⩾30%. Exclusion criteria: serious spinal condition, nerve root compression, pregnancy, or previous surgery to the lumbosacral spine or pelvis. Patients were allocated to receive spinal manipulation plus exercise (n = 70) or exercise alone (n = 61) given by a physical therapist for 4 weeks.

    Embedded ImageDescription of prediction guide:

    the clinical prediction rule criteria were (1) LBP symptom duration <16 days, (2) no symptoms distal to the knee, (3) Fear Avoidance Beliefs Questionnaire work subscale score <19 points, (4) ⩾1 hypomobile segment in the lumbar spine, and (5) ⩾1 hip with >35 degrees of internal rotation range of motion. Patients were classified as “positive” (likely to respond to spinal manipulation) if ⩾4 of the 5 criteria were met. Patients were classified as “negative” if they had positive status on ⩽3 of the 5 criteria.

    Embedded ImageOutcomes:

    disability, measured by modified ODQ score (score range 0% to 100%), and treatment “success”, defined as ⩾50% improvement in ODQ score, assessed at 1 week (100% follow up), 4 weeks (99% follow up), and 6 months (70% follow up).


    In patients who received spinal manipulation plus exercise, those who were positive on the rule (23 of 70 [33%]) had greater improvement in 1 week and 4 week disability than those who were negative on the rule (47 of 70 [67%]) (table). Positive status on the rule had a positive likelihood ratio of 13 (95% CI 3 to 52) for predicting treatment “success” at 1 week. In patients who received exercise alone, patients who were positive on the rule (24 of 61 [39%]) and those who were negative on the rule (37 of 61 [61%]) did not differ for 1 week or 4 week disability (table).

    Positive v negative status on the spinal manipulation clinical prediction rule in patients with low back pain


    A clinical prediction rule predicted outcome in patients with low back pain receiving spinal manipulation and exercise treatment but not in patients with low back pain receiving exercise treatment alone.


    This study by Childs et al presents interesting data suggesting that an identifiable subgroup of patients with back pain may benefit from adding spinal manipulation to an exercise treatment. The clinical prediction rule was developed in a previous cohort study of patients receiving manipulatory treatment after referral for physical therapy.1 Several of the prediction rule’s components identify a group of patients who are likely to do well regardless of treatment. In this new study, the prediction rule was again tested on patients referred for physical therapy.

    The current study was designed to test for a 3 way interaction between patients’ status on the rule, their treatment group allocation, and outcome at 1 week. This relationship was statistically significant, showing that both status on the prediction rule and treatment contribute to clinical outcome, which supports the hypothesis that the prediction rule specifically identifies those who will benefit from additional manipulation. However, the strength of this interaction and its statistical significance are not stated. The effect sizes presented are post hoc analyses that need cautious interpretation. As expected, the rule predicted outcome in those who received spinal manipulation. Additionally, those who were positive on the prediction rule and received manipulation did substantially better than those who received exercise only, suggesting that the rule could be used to guide treatment choices. However, using this rule to decide on referral for physical therapy including manipulation may be impractical for those unfamiliar with assessing fear avoidance beliefs or lumbar hypomobility with spring testing.

    In conclusion, this prediction rule appears to be a promising approach for identifying subgroups of patients with back pain. However, what would now be helpful is a further study that is able to replicate these findings, using a more practical version of the tool in the situation in which it will be used, with more robust analyses presented. Based on such results, a more rigorous determination on the application of this rule in routine clinical practice will be possible.


    View Abstract


    • For correspondence: Dr J D Childs, Wilford Hall Medical Center, San Antonio, TX, USA.

    • Source of funding: Foundation for Physical Therapy, Inc.