File
Almeida M1, Saragiotto B2, Maher C2, Costa L3
1Universidade Cidade de São Paulo (UNICID), Master and Doctoral Programs in Physhiotherapy, Sao Paulo, Brazil, 2University of Sydney, Sydney School of Public Health, Sydney, Australia, 3Universidade Cidade de São Paulo (UNICID), Master and Doctoral Programs in Physiotherapy, Sao Paulo, Brazil
Background: In order to identify potential sources of biases in clinical research and to understand the impact of study level characteristics in clinical trials, meta-epidemiologic studies have investigated the association between trial characteristics and the intervention effect estimates. This is the first study that investigated the influence of allocation concealment and intention-to-treat (ITT) analysis on the magnitude of the treatment effects of physical therapy interventions in low back pain trials.
Purpose: To evaluate if there is association between allocation concealment and intention-to-treat (ITT) analysis with treatment effects of physical therapy interventions in low back pain trials.
Methods: This is a meta-epidemiological study. The study was prospectively registered at PROSPERO, registration number: CRD42016052347. We included low back pain trials that compared physical therapy interventions to placebo or no intervention or minimal intervention with pain or disability outcomes. Potentially eligible trials were identified by searching in the following databases from their inception up to February 2017: PubMed, Embase, Cochrane Database of Systematic Reviews, PEDro and CINAHL. Information about allocation concealment and ITT analysis was extracted from the Physiotherapy Evidence Database. In the first stage of the analysis, we calculated the pooled treatment effect estimates using a random effects model: (i) trials with allocation concealment, (ii) trials without allocation concealment, (iii) trials with ITT analysis and (iv) trials without ITT analysis. As a post-hoc analysis, we also calculated the pooled treatment effect estimates for trials with and without both characteristics. The second stage of the analysis was a random-effect meta-regression to evaluate the association between these methodological features of interest and treatment effects. Other covariates included in the meta-regression were sample size and sequence generation.
Results: We identified 128 eligible trials (pooled n = 20555 participants) with 209 comparisons included in our analysis. 57 of the 128 trials (44.5%) included in our study performed allocation concealment; 41 of the 128 trials (32%) included in our study performed ITT analysis; and only 29 of the 128 trials (22.6%) included in our study had both characteristics. Meta regression analyses showed no influence of allocation concealment on treatment effects for pain (regression coefficient 0.009; 95% CI -2.91 to 2.91) and disability (regression coefficient 1.13; 95% CI -1.35 to 3.62), and no influence of ITT analysis for pain (regression coefficient 1.38; 95% CI -1.73 to 4.50) or disability (regression coefficient 1.27; 95% CI -1.39 to 3.64). For the other covariates, there was also no clinically significant association with the treatment effects. As a post-hoc analysis, we also investigated the association between the treatment effects and the presence of both characteristics (allocation concealment plus ITT analysis) and no significant association was found for both outcomes in the univariate analysis, as well in the multivariate analysis.
Conclusion(s): There is no influence of allocation concealment and/or ITT analysis on treatment effects of physical therapy interventions for pain and disability in low back pain trials.
Implications: Allocation concealment and ITT analysis cannot be considered as sources of bias in low back pain clinical trials, since our results from meta-regression analyses did not find statistically significant associations.
Keywords: meta-epidemiology, low back pain, physical therapy modalities
Funding acknowledgements: Matheus O Almeida was supported by São Paulo Research Foundation (FAPESP) - grant #2016/10317-0
Purpose: To evaluate if there is association between allocation concealment and intention-to-treat (ITT) analysis with treatment effects of physical therapy interventions in low back pain trials.
Methods: This is a meta-epidemiological study. The study was prospectively registered at PROSPERO, registration number: CRD42016052347. We included low back pain trials that compared physical therapy interventions to placebo or no intervention or minimal intervention with pain or disability outcomes. Potentially eligible trials were identified by searching in the following databases from their inception up to February 2017: PubMed, Embase, Cochrane Database of Systematic Reviews, PEDro and CINAHL. Information about allocation concealment and ITT analysis was extracted from the Physiotherapy Evidence Database. In the first stage of the analysis, we calculated the pooled treatment effect estimates using a random effects model: (i) trials with allocation concealment, (ii) trials without allocation concealment, (iii) trials with ITT analysis and (iv) trials without ITT analysis. As a post-hoc analysis, we also calculated the pooled treatment effect estimates for trials with and without both characteristics. The second stage of the analysis was a random-effect meta-regression to evaluate the association between these methodological features of interest and treatment effects. Other covariates included in the meta-regression were sample size and sequence generation.
Results: We identified 128 eligible trials (pooled n = 20555 participants) with 209 comparisons included in our analysis. 57 of the 128 trials (44.5%) included in our study performed allocation concealment; 41 of the 128 trials (32%) included in our study performed ITT analysis; and only 29 of the 128 trials (22.6%) included in our study had both characteristics. Meta regression analyses showed no influence of allocation concealment on treatment effects for pain (regression coefficient 0.009; 95% CI -2.91 to 2.91) and disability (regression coefficient 1.13; 95% CI -1.35 to 3.62), and no influence of ITT analysis for pain (regression coefficient 1.38; 95% CI -1.73 to 4.50) or disability (regression coefficient 1.27; 95% CI -1.39 to 3.64). For the other covariates, there was also no clinically significant association with the treatment effects. As a post-hoc analysis, we also investigated the association between the treatment effects and the presence of both characteristics (allocation concealment plus ITT analysis) and no significant association was found for both outcomes in the univariate analysis, as well in the multivariate analysis.
Conclusion(s): There is no influence of allocation concealment and/or ITT analysis on treatment effects of physical therapy interventions for pain and disability in low back pain trials.
Implications: Allocation concealment and ITT analysis cannot be considered as sources of bias in low back pain clinical trials, since our results from meta-regression analyses did not find statistically significant associations.
Keywords: meta-epidemiology, low back pain, physical therapy modalities
Funding acknowledgements: Matheus O Almeida was supported by São Paulo Research Foundation (FAPESP) - grant #2016/10317-0
Topic: Musculoskeletal: spine; Research methodology & knowledge translation
Ethics approval required: No
Institution: N/A
Ethics committee: N/A
Reason not required: This is a meta-epidemiological study
All authors, affiliations and abstracts have been published as submitted.
