To evaluate the effectiveness of spinal manipulation in the treatment of low back pain (LBP).

MEDLINE, CINAHL, and BIDS were searched for the period of 1985 to 1997 using the keywords 'low back pain', 'spinal manipulation', 'randomized clinical trials', 'effectiveness', 'chiropractic', 'musculoskeletal disease', 'osteopathic', 'comparison study' and 'blind study'. In addition, references given in relevant publications were screened. Unpublished studies and abstracts were not selected. Only English language publications were considered for inclusion.

Study designs of evaluations included in the review

Randomised controlled trials (RCTs). Duration of included studies, where specified, ranged from 2 weeks to 24 months.

Specific interventions included in the review

Manipulation of the lumbar spine delivered alone or along with other interventions. Specific types of manipulation reported in the studies included orthopaedic, chiropractic, Lewitt, forceful manipulation, rotational, cyriax, Janda and Lewitt, SI joint mobilisation, (with or without thrust technique), manipulation (with or without mobilisation), and high velocity with low amplitude. Reference treatments were mainly physiotherapy interventions e.g. short wave diathermy (SWD), ultrasound (US), exercise, infra-red radiation (IRR), corset, massage, non-steroidal anti-inflammatory drugs (NSAIDs), analgesia or placebo therapy.

Participants included in the review

Individuals with LBP (subacute, acute and chronic), that was defined as an identified area between the lowest ribs and the inferior gluteal fold. The stage of LBP reported by included studies were acute or subacute (duration of less than 12 weeks), and chronic (lasting more than 12 weeks).

Outcomes assessed in the review

There were no a-priori outcome measures considered for inclusion in the review. The type of outcomes reported by included studies were reduction in pain, level of pain on visual analog scale, mobility, differences in posture, gait, active spinal movement, manual examination of spine, neural examination, range of motion, improvement in Oswestry Disability, other physical tests, days on sick leave, Scale and analgesic consumption. Outcome measurements during or just after the intervention were defined as short-term effects, while long-term effects were defined as those outcomes measured at least 3 months after randomisation.

How were decisions on the relevance of primary studies made?

The authors do not state how the papers were selected for the review, or how many of the reviewers performed the selection.

The following criteria were used to assess methodological quality:

1. Study population included clear definition of inclusion and exclusion criteria (homogeneity), similarity of relevant baseline characteristics, randomised procedure, drop outs described for each study group separately, loss of follow-up described for each group separately and study size.

2. Interventions included explicit description of experimental treatment, comparison with another intervention, co-intervention avoided except exercise education, placebo controlled, qualified manipulative therapist.

3. Measurements of effect included patients blinded, relevant outcome measures (pain, disability, cost, and return to work), blinded outcome assessment, and duration of follow-up.

4. Data presentation and analysis included intention to treat analysis and data presented for most important outcome measures.

The authors do not state how the papers were assessed for quality, or how many of the reviewers performed the quality assessment.

The type of data extracted from individual studies included in the review were: reference details, type of manipulation and the number of subjects in the intervention group, type of comparison used and number of subjects in the control group(s), type of back pain, and a summary of the results and conclusions. It was not stated how many of the authors were involved in this process.

How were the studies combined?

The studies were combined in a narrative. The outcome of individual studies was classified as positive (if there was a difference in effect between the compared treatment in favour of manipulation total subjects or in a subgroup of the population) or negative (if there was no reported difference between the compared treatments, or if there was a difference in effect in favour of one or more of the reference groups).

How were differences between studies investigated?

Differences between the studies (e.g. variation in manipulative techniques and treatment environment) were discussed in a narrative. No formal statistical test for heterogeneity was performed.

Twenty-five RCTs with a minimum of 3,187 participants.

There were 17 (68%) trials in which the authors reported better results in favour of manipulation compared with the reference treatment. Of these 17, there were 12 (48%) studies in which the authors presented positive effect in favour of manipulation in all subjects and five (20%) studies that showed positive results in one or more subgroups. In six (24%) studies, manipulation was no more effective than the reference treatment, and in two (8%) trials the authors did not report a conclusion.

Of the 18 trials which included patients with an acute or sub-acute condition (total group or subgroup), 12 (67%) trials reported positive results in favour of manipulation. In five (28%) of these studies there were no superior results in favour of manipulation, and in one (5%) study there was no conclusion. Seven studies measured outcomes for long-term effects. Of these, five studies reported results in favour of manipulation, and two showed negative results.

Nine studies compared manipulation with placebo therapy. Of these, six (67%) studies reported a positive effect in all participants, and two (22%) found a positive effect only in a subgroup. Only one study presented a negative result.

Methodological flaws were seen in both positive and negative trials. The most important flaws were small sample size of the study population (6 studies), lack of description for drop-out (4 studies), lack of real placebo group trusted by patents (3 studies), unblinded assessor (4 studies), co-intervention unavoided (3 studies), and lack of naive patients (4 studies).

Although there were methodological flaws, which precluded strong conclusions, it appears that manipulation may be more effective than other interventions in the treatment of LBP, both in short- and long- terms effects. This finding is based on limited data at this time.

The review stated the objective and was based on a-priori inclusion/exclusion criteria as well as a fairly thorough search of the literature. However, abstracts and unpublished studies were excluded and therefore the presence of publication bias cannot be ruled out. In addition, only English language trials were considered for inclusion, which may cause bias due to the exclusion of RCTs published in any other language. The quality of included studies was assessed using a structured validity checklist and study details were clearly presented in tables, although a summary of the effect size of individual studies was not presented. Information about the methodology of the review process (such as how decisions on the relevancy of primary studies were made, whether more than one reviewer conducted data extraction, and how discrepancies were resolved) was very limited.

The authors' conclusions seem to follow from the results.

Practice: The authors do not make any recommendations for practice.

Research: The authors note that better research methodology is needed and that longer follow-up is required for a more convincing estimate of cost-effectiveness.

Islamic Republic of Iran Ministry of Health and Medical Education (Mazandaran University of Medical Sciences)

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.