Back to presentation overview

Title:  Effect sizes for treatment of non-specific low back pain.

Authors and affiliation: Keller Anne1,2, Hayden J2, Bombardier C2, van Tulder M3

Email address: anne.keller@medisin.uio.no

Introduction: Several treatments are available for non-specific low back pain and numerous randomized trials have been published investigating the effectiveness of treatments by comparing different interventions or by comparing interventions with no treatment/placebo. However, several trials have examined combinations of different interventions and compared these combinations with either one intervention or another combination of interventions. Hence, it is impossible to unravel which of the interventions had effect, and this raises questions of the basic benefit of a each treatment. Many systematic reviews have been published on the effectiveness of interventions for non-specific low back pain. Most of these are based on qualitative analysis, where the evidence is expressed as a level of evidence. Another approach is a quantitative analysis where studies are statistically pooled and expressed as pooled effect size either as Standardized Mean Difference (SMD) or as Weighted Mean Difference.

Objective: The objective of the present study was to evaluate which intervention for non-specific low back pain has the largest effect size and to compare them with each other.

Methods: Effect sizes were extracted from systematic reviews of treatment of non-specific low back pain published in the latest issue of the Cochrane Library, issue 2, 2005. Reviews were selected if they assessed a commonly used treatment and included at least two studies for each treatment. Inclusion criteria for individual studies within the included reviews were: The interventions were compared to placebo, sham treatment, no treatment or waiting list, and the study population consisted of a general population. Outcome measurements were pain and function for short- and long term follow-up. A quantitative meta-analysis was performed in which the effect sizes were pooled using the random effects model. For continuous variables the effect size was calculated as standadized mean difference (SMD), which is defined as the differences in outcome measures between two groups divided with the standard deviation for the difference. For dichotomous data the effect size was calculated as relative risk (RR), where RR is the risk of an event in the treatment group divided by the risk of the event in the comparison group. The effect size was evaluated according to Cohen, who defined effect sizes between 0.2 and 0.5 as small, effect sizes between 0.5 and 0.8 as moderate and effect sizes > 0.8 as large. Statistical heterogeneity was assessed using I2 statistics and confidence interval. For the continuous variables the pooled effect sizes for the treatments of the acute and chronic conditions were presented in a forest plot.

Results: Twenty systematic reviews were identified in the Cochrane Library 2005, issue 2, of which 7 fulfilled the inclusion criteria and 13 were excluded. For acute low back pain the following treatments were included: exercise, manipulation, non-steroidal anti-inflammatory drugs (NSAID) and non-benzo­diazpines, where only exercise had short- and long term follow-up. The pooled effect size (SMD), for short-term follow-up of pain was moderate for NSAID (0.55) and non-benzodiapines (RR=0.58) and small for manipulation (0.38) and exercise (0.01). For chronic conditions the following treatments were included: acupuncture, behavioural therapy, benzo­diazepines, exercise, manipulation and TENS (Transcutaneous electric nerve stimulation). The pooled effect size (SMD) for short-term follow-up of pain was moderate for acupuncture (0.61), behavioural therapy (0.57) and exercise (0.52), large for benzo­diazepines (RR 0.82) and small for manipulation (0.32) and TENS (0.22). For exercise the pooled effect sizes were small for long-term pain relief (0.25) and for short- and long-term improvement in function (0.22-0.13).

Conclusion: The effect sizes for the pure benefit of treatments of low back pain were small to moderate estimated by SMD. Long-term follow up for pain and function was lacking. For the acute condition NSAID had the largest effect size and behavioural treatment and acupuncture had the highest effect sizes for the chronic condition.