A Meta-Epidemiological Appraisal of the Effects of Interdisciplinary Multimodal Pain Therapy Dosing for Chronic Low Back Pain

Abstract

Using a meta-analysis, meta-regression, and a meta-epidemiological approach, we conducted a systematic review to examine the influence of interdisciplinary multimodal pain therapy (IMPT) dosage on pain, disability, return to work, quality of life, depression, and anxiety in published randomised controlled trials (RCTs) in patients with non-specific chronic low back pain (CLBP). We considered all RCTs of IMPT from a Cochrane review and searched PubMed for additional RCTs through 30 September 2018. A subgroup random-effects meta-analysis by length, contact, and intensity of treatment was performed followed by a meta-regression analysis. Using random and fixed-effect models and a summary relative odds ratio (ROR), we compared the effect sizes (ES) from short-length, non-daily contact, and low-intensity RCTs with long-length, daily contact, and high-intensity RCTs. Heterogeneity was quantified with the I² metric. A total of 47 RCTs were selected. Subgroup meta-analysis showed that there were larger ES for pain and disability in RCTs with long-length, non-daily contact, and low intensity of treatment. Larger ES were also observed for quality of life in RCTs with short-length, non-daily contact, and low intensity treatment. However, these findings were not confirmed by the meta-regression analysis. Likewise, the summary RORs were not significant, indicating that the length, contact, and intensity of treatment did not have an overall effect on the investigated outcomes. For the outcomes investigated here, IMPT dosage is not generally associated with better ES, and an optimal dosage was not determined.

Keywords: interdisciplinary multimodal pain therapy; low back pain; meta-analysis; pain rehabilitation; programme dosage.

Figure 1

Risk of bias graph: assessments for seven risk of bias criteria presented as percentages across all included studies.

Figure 2

The relative odds ratios (RORs) and 95% confidence intervals (CIs) for each outcome, and the summary RORs and their 95% CIs at short term of a short-length treatment vs. long-length treatment. The RORs were calculated with a random-effects model. A ROR >1 favours long length; an ROR <1 favours short length.

Figure 3

The relative odds ratios (RORs) and 95% confidence intervals (CIs) for each outcome, and the summary RORs and their 95% CIs at short term of non-daily contact vs. daily contact. The RORs were calculated with a random-effects model. A ROR >1 favours daily contact; an ROR <1 favours non-daily contact.

Figure 4

The relative odds ratios (RORs) and 95% confidence intervals (CIs) for each outcome, and the summary RORs and their 95% CIs at short term of low intensity vs. high intensity. The RORs were calculated with a random-effects model. A ROR >1 favours high intensity (i.e. >30 h per week); an ROR <1 favours low intensity (i.e. <30 h per week).