Review Article

A Systematic Review and Meta-Analysis of Efficacy, Cost-Effectiveness, and Safety of Selected Complementary and Alternative Medicine for Neck and Low-Back Pain

Table 4

Summary of findings of spinal manipulation for low-back pain (only pain and functional outcomes).

Duration and cause of painOutcomesGRADE*Findings

Manipulation versus no treatment

Acute/subacute nonspecificPain intensity score (0 to 5)Low
Design: RCT
ROB: Medium
Consistency: NA (one trial)
Directness: yes
In one trial [97], there was a significantly lower immediate posttreatment pain intensity in the manipulation group (change from 2.8 to 1.0; ) compared to “no treatment” group (change from 2.0 to 2.1, ).

Acute/subacute specificNAInsufficient
No trial
NA

Mixed nonspecificPain intensity score (VAS)Low
Design: RCT
ROB: High
Consistency: NA (one trial)
Directness: yes
In one trial [98], manipulation showed significant reduction (from baseline) in immediate/short-term posttreatment pain intensity (VAS: 12.20 versus 10.40, ), while the “no treatment” group did not experience significant reduction in pain intensity ( ).

Mixed specificNAInsufficient
No trial
NA

Chronic or Unknown (nonspecific and specific)NAInsufficient
No trial
NA

Manipulation versus placebo

Acute/subacute, nonspecificPain intensity score (VAS)Moderate
Design: RCT
ROB: Medium
Consistency: yes
Directness: yes
Four trials [97, 99, 101, 211] showed significant improvements for manipulation in reducing immediate or short-term posttreatment pain. For example, in one trial [211], manipulation was significantly superior to placebo at short-term followup (four-point VAS; ). Intermediate-term posttreatment data of the same trial showed no significant difference between the groups. In another trial [101], manipulation showed significantly better immediate-term posttreatment pain intensity (percentage of pain-free subjects: 92.0% versus 25.0%, ).
Oswestry Disability IndexLow
Design: RCT
ROB: Medium
Consistency: NA (one trial)
Directness: yes
One trial [99] showed no between-group differences in the immediate and short-term posttreatment follow-ups.

Acute/subacute specificNAInsufficient
No trial
NA

Chronic nonspecificPain intensity score (VAS)Low
Design: RCT
ROB: Medium
Consistency: no
Directness: yes
In two trials [102, 211], manipulation was significantly better than placebo. In a third trial [103], the immediate posttreatment pain intensity improved more in the manipulation group (1.3 versus 0.7) and in the short-term posttreatment (2.3 versus0.6). There was a significant change within the manipulation group but not within the placebo group. The value for between-group comparison was not reported and therefore the between-group significant difference was not assumed.
Oswestry Disability IndexLow
Design: RCT
ROB: Medium
Consistency: NA (one trial)
Directness: yes
In one trial [102], manipulation was significantly better than placebo immediately posttreatment ( versus , ), but the difference in the short-term posttreatment was not statistically significant ( versus , ).

Chronic specificNAInsufficient
No trial
NA

Mixed nonspecificPain intensity score (VAS)Low
Design: RCT
ROB: High
Consistency: NA (one trial)
Directness: yes
One trial [104] showed that immediate posttreatment improvement was numerically greater in the manipulation group (numerical data not reported, and statistical test results were not provided).

Mixed specificNAInsufficient
No trial
NA

Unknown (specific, nonspecific)NAInsufficient
No trial
NA

Manipulation versus pain medication

Acute/subacute, nonspecificPain intensity score (VAS)Low
Design: RCT
ROB: High
Consistency: NA (one trial)
Directness: yes
One trial showed a nonsignificant advantage of manipulation at the immediate posttreatment followup [211]. This advantage was not sustained at the short- and intermediate posttreatment followups (numerical data not reported, and statistical test results were not provided).

Acute/subacute specificNAInsufficient
No trial
NA

Chronic nonspecificPain intensity score (VAS)
Immediate posttreatment
Low
Design: RCT
ROB: High
Consistency: yes
Directness: yes
Two trials [69, 70] showed significantly greater pain reductions with spinal manipulation. The median (IQR) pain intensity went from 5 (4 to 8) to 3 (0 to 7) ( ) with manipulation, and from 5 (3 to 8) to 5 (2 to 7) ( ) with medication [52]. In the other trial, the change was −2.5 (95% CI: −5.0, −21) in the manipulation group and +0.3 (95% CI: −1.0, 1.7) in the medication group [70].
Pain intensity (subjective score: 5 = poor, 32 = excellent)Low
Design: RCT
ROB: High
Consistency: NA (one trial)
Directness: yes
One trial [211] showed that spinal manipulation was not significantly different from medication.
Subjective score with manipulation were 2.6 and 4.3 in the short- and intermediate-term. Subjective score with medication were 2.2 and 4.0 in the short- and intermediate-term. (Statistical test results were not provided).
Oswestry Disability IndexLow
Design: RCT
ROB: High
Consistency: yes
Directness: yes
Two trials [69, 70] showed significantly greater mean reduction in disability in the manipulation versus pain medication group immediately after treatment (50% [69] and 30.7% [70]).

Chronic specificNAInsufficient
No trial
NA

Mixed or unknown (specific, nonspecific)NAInsufficient
No trial
NA

Manipulation versus physiotherapy

Acute/subacute, nonspecificPain intensity score (VAS)Low
Design: RCT
ROB: High
Consistency: NA (one trial)
Directness: yes
One trial [211] showed better scores with manipulation at the immediate-, short- and intermediate posttreatment followups (Numerical data not reported, and statistical test results were not provided).

Acute/subacute, specificNAInsufficient
No trial
NA

Chronic nonspecificPain intensity score (VAS)Low
Design: RCT
ROB: High
Consistency: NA (one trial)
Directness: yes
One trial [211] showed better scores with physiotherapy versus manipulation at the immediate-, short- and intermediate posttreatment followups (numerical data not reported, and statistical test results were not provided).

Chronic specificNAInsufficient
No trial
NA

Mixed nonspecificPain intensity score (11-point pain scale)Low
Design: RCT
ROB: High
Consistency: NA (one trial)
Directness: yes
In one trial [105], no significant differences were found in short-term posttreatment effects between manipulation and physiotherapy (McKenzie technique based on diagnoses of derangement, dysfunction or postural syndromes).
Roland-Morris Disability scoreLow
Design: RCT
ROB: High
Consistency: NA (one trial)
Directness: yes
In one trial [105], there was no significant difference between manipulation and physiotherapy (McKenzie technique based on diagnoses of derangement, dysfunction or postural syndromes) in the short-term posttreatment effects.

Mixed specificNAInsufficient
No trial
NA

Unknown (specific, nonspecific)NAInsufficient
No trial
NA

Manipulation versus usual care

Mixed nonspecificPain intensity score (100-mm VAS score)Low
Design: RCT
ROB: Low
Consistency: NA (one trial)
Directness: yes
In one trial [106], high or low velocity spinal manipulation was not significantly different from minimal conservative medical care.
Mean VAS score difference between high velocity manipulation and usual care was 4.0 (95% CI: −4.0, 12.0), whereas this difference between low velocity manipulation and usual care was 5.8 (95% CI: −2.3 to 14.0).

Roland-Morris Disability scoreLow
Design: RCT
ROB: Low
Consistency: NA (one trial)
Directness: yes
One trial [106] showed that manipulation was significantly more effective than medical care alone in improving disability at immediate, short-, or intermediate-term posttreatment followup. The adjusted RMDQ mean change from baseline in the high and low velocity manipulation and medical care groups were 2.7 (95% CI: 2.0, 3.3), 2.9 (95% CI: 2.2, 3.6), and 1.6 (95% CI: 0.5, 2.8), respectively.

Mixed specificNAInsufficient
No trial
NA

Acute, chronic or unknown (specific, nonspecific)NAInsufficient
No trial
NA

Manipulation versus massage

Acute/subacute nonspecificPain intensity score (100-mm VAS)Low
Design: RCT
ROB: High
Consistency: NA (one trial)
Directness: yes
In one trial [107], there was no significant difference between manipulation and massage immediately posttreatment (mean difference: and , resp.).

Acute/subacute specificNAInsufficient
No trial
NA

Chronic nonspecificPain (duration of pain relief)Low
Design: RCT
ROB: Medium
Consistency: NA (one trial)
Directness: no
In one trial [108], manipulation was significantly better than massage immediately—and in the short-term after treatment. The mean (SE) duration of pain relief was with manipulation versus with massage.

Chronic specificNAInsufficient
No trial
NA

Mixed or unknown (specific, nonspecific)NAInsufficient
No trial
NA

*Precision in formal grading was applied only to pooled results.
VAS: visual analog scale; RMDQ: Roland-Morris disability scale; MPQ: McGill pain questionnaire; PDI: pain disability index; NPQ: neck pain questionnaire; NA: not applicable; ROB: risk of bias; RCT: randomized controlled trial.