Therapies, outcomes and evidence application into practice in non-specific low back pain

Abstract

The first study of this thesis (Chapter Two) investigated the efficacy of Pilates for non-specific LBP in a Cochrane systematic review. Data from ten randomised controlled trials (n= 510 participants) were included and there was low to moderate quality evidence that Pilates is more effective than minimal intervention for reducing pain and disability. When Pilates was compared with other exercises, there was a small effect for improving function at intermediate-term follow-up. Thus, while there is some evidence for the effectiveness of Pilates for low back pain, there is no conclusive evidence that it is superior to other forms of exercises. Another common treatment modality for non-specific low back pain is pharmacological therapies. There is a large number of randomised controlled trials (RCTs) evaluating the efficacy of medicines for chronic conditions. It is also known that enriched study designs are commonly used to investigate the efficacy of medicines for specific conditions. Thus, the second study of this thesis (Chapter Three) investigated the impact of enriched and non-enriched study designs on the effect size estimates of various analgesics in trials investigating chronic musculoskeletal pain. The study included 108 trials with a total sample of 44,171 participants. Enriched designs do not influence the estimates of treatment effect for pain intensity but may reduce the estimate of the number of any adverse events. Features of methodological quality (e.g., lack of blinding of assessor and no intention-to-treat analysis) were significantly associated with greater effect sizes for pain. Still related to quality of clinical trials, the aim of the study from Chapter Four was to investigate the validity and inter-rater reliability of the PEDro scale for pharmaceutical trials. The PEDro scale has been widely used to investigate methodological quality in physiotherapy trials; however, its validity has not been tested for pharmaceutical trials. This is a secondary analysis of data from a previous study (Chapter Three). Convergent validity was evaluated by correlating the PEDro score with the summary score of the Cochrane Back and Neck risk of bias tool. Based on data from 53 trials (91 treatment effect sizes) included in the analyses, there was evidence for the convergent and construct validity for the PEDro scale when used to evaluate methodological quality of pharmacological trials. Both risk of bias tools have acceptably high inter-rater reliability. Since the PEDro database has been globally accessed around the world, Chapter Five of this thesis aimed to quantify the relationship between the number of times articles are accessed on the PEDro database and the article characteristics. A sample of 29,313 articles indexed on PEDro between August 2014 and January 2015 were used in this study. As a result, seven factors that predicted the number of accesses of an article were identified in this study: synthesis research (i.e., guidelines and reviews); recent articles, in English and that were promoted to PEDro users; being a Cochrane review; and having a higher PEDro score. Also, the musculoskeletal, neurology, orthopaedics, sports, and paediatrics subdisciplines were associated with more accesses. The number of times an article is accessed on PEDro is partly predicted by how condensed and high quality the evidence it contains is. In parallel with the idea that different designs of trials and aspects of trial quality could influence the results of the studies, Chapter Six of this thesis investigated the completeness of intervention descriptions in 200 physiotherapy trials from the PEDro database. Methodological quality was evaluated using the PEDro scale and completeness of intervention description using the Template for Intervention Description and Replication (TIDieR) checklist. This study concluded that the description of interventions in physiotherapy RCTs is typically incomplete, which prevents the replication of the treatments (by clinicians), but may also cause problems summarising similar interventions in systematic reviews. Authors and journals should aim for more complete descriptions of interventions in physiotherapy trials. Therefore, my colleagues and I contacted the International Committee of Physiotherapy Journal Editors to encourage them to adopt the TIDieR template and to publish a joint editorial emphasising the importance of complete descriptions of trial interventions. Twenty physiotherapy journals accepted to publish the editorial and 16 have already published it. To facilitate and complement the use of the TIDieR checklist, Chapter Seven of this thesis evaluated the use of the TIDieR checklist as a measure of completeness of intervention description with a summary score, conducting a secondary analysis of the data from the study in Chapter Six. In general, the Rasch analysis supported the calculation of a summary score. The use of a summary score for the TIDieR checklist will help with improving intervention descriptions in physiotherapy, as well as be an alternative to use intervention completeness as a single dependent variable in studies evaluating strategies to improve reporting and in meta-regression analyses in systematic reviews. Nevertheless, the TIDieR score requires validation in an independent data set and this could be done using data generated in other evaluations of cohorts of papers using the TIDieR checklist. The last chapter of this thesis (Chapter Eight) is dedicated to analysing the structure of the Roland-Morris Disability Questionnaire in a large sample of patients with LBP using two different statistical approaches. The Roland–Morris Disability Questionnaire is one of the most recommended back-pain specific questionnaires to assess disability, however, some recent evidence has suggested that the questionnaire should be considered as a multidimensional scale. In this study (Chapter Eight), we tested the dimensionality of the questionnaire using existing datasets from previous clinical studies (n=2,826) including patients with non-specific LBP. We used Rasch analysis and Confirmatory Factor Analysis (CFA) to test a previous hypothesis of a 3-factor solution. The Rasch analysis model showed that the questionnaire is unidimensional and results from the CFA suggested poor fit to the data on the 3-factor solution. It is recommended that the questionnaire should still be used as a unidimensional scale for measuring disability as the only construct. Overall, the studies included in this thesis have provided an important contribution to the evidence underpinning contemporary management of LBP and physiotherapy in general. The last chapter of this thesis (Chapter Nine), presents the final considerations, including a discussion on the main findings and implications of the work presented in this thesis. The main implications are: i) Pilates is effective for non-specific LBP, but it does not appear to be better than other types of exercise; ii) enriched designs do not influence estimates of treatment effect for pain intensity but may reduce estimates of adverse events in chronic pain; iii) the PEDro scale has adequate properties to be used to evaluate methodological quality of pharmacological trials; iv) the number of access of trials indexed on PEDro are predicted by synthesis research (i.e., guidelines and reviews); recent articles, written in English and that were promoted to PEDro users; being a Cochrane review; having a higher PEDro score, and the subdisciplines of musculoskeletal, neurology, orthopaedics, sports, and paediatrics; v) the description of interventions in physiotherapy trials are typically incomplete; vi) the Rasch analysis supports the calculation of a summary score for the TIDieR checklist; vii) it is recommended that the RMDQ should still be used as a unidimensional scale for measuring disability as the only construct.