SLUGGISH ADOPTION OF SIMPLE METHODS THAT REDUCE BIAS: A LONGITUDINAL ANALYSIS OF 35,653 PHYSIOTHERAPY TRIALS

G. Wallbank^1,2, A. Traeger^1,2, M. Elkins^3,2

¹The University of Sydney and Sydney Local Health District, Institute for Musculoskeletal Health, Sydney, Australia, ²The University of Sydney, Faculty of Medicine and Health, Sydney, Australia, ³Sydney Local Health District, Sydney, Australia

Background: The Physiotherapy Evidence Database (PEDro) comprehensively indexes randomised trials of physiotherapy interventions. The PEDro Scale is widely used to appraise aspects of the methodological robustness and the completeness of reporting in trials of physiotherapy interventions. Some items on the PEDro Scale are inherently difficult to achieve, such as blinding of patients and therapists with exercise interventions. However, two items are simpler to implement: concealing group allocation from those enrolling participants (ie, ‘concealed allocation’) and performing an intention-to-treat analysis. Yet even in 2022 only a minority of trials report using these features.

Purpose: Given that concealed allocation and intention-to-treat analysis were first discussed in the literature in the early 1990s, we used longitudinal analysis of trials indexed on PEDro to address the question: What is the rate of improvement in the percentage of trials on PEDro that use concealed allocation and intention-to-treat analysis?

Methods: For trials of physiotherapy interventions indexed on PEDro, year of publication and methodological quality scores (11-item PEDro scale) were extracted. The percentage of randomised trials that achieve each of the two PEDro Scale criteria of interest was regressed over time from 1980 to 2019, with data binned by decade. The outcome measured was the rate of improvement in the percentage of trials indexed on PEDro that use concealed allocation and that use intention-to-treat analysis.

Results: Data for 35,653 trials published between 1980 and 2019 and with completed PEDro ratings were downloaded and analysed. The percentage of trials that used concealed allocation increased by ~9% per decade. In the 2010s, 32% of trials used concealed allocation. The percentage of trials that used intention-to-treat analysis increased by ~10% per decade. In the 2010s, 33% of trials used concealed allocation.

Conclusions: Despite being readily achievable, concealed allocation and intention-to-treat analysis are still used by a minority of trials indexed on PEDro. At the current rate of improvement, complete uptake of these simple methods to reduce bias in trials of physiotherapy interventions will not be achieved until the 2090s for concealed allocation and the 2080s for intention-to-treat analysis.

Implications: Rapid complete uptake of these simple methods would avoid decades of physiotherapy trials being exposed to important biases. Clinical trialists who investigate the effects of physiotherapy interventions must routinely use concealed allocation throughout the period that they recruit and allocate participants to groups. Furthermore, they should ensure that they understand the definition of intention-to-treat analysis recommended by the CONSORT Statement, the Cochrane Collaboration and the PEDro Scale, and then ensure that their procedures during the conduct of the trial and at the analysis stage permit achievement of the intention-to-treat analysis criterion.

Funding acknowledgements: This work did not receive funding.

Keywords:
Randomized controlled trial
Bias
Methods

Topics:
Education
Education: continuing professional development

Did this work require ethics approval? No

Reason: Ethical approval was not required as this work is a longitudinal analysis of data from the Physiotherapy Evidence Database which is managed by our institution.