COMMONLY MISUNDERSTOOD STATISTICS IN PHYSIOTHERAPY RESEARCH
Q louw1, C Ardern2, M Elkins3, R Zambelli Pinto4
1Stellenbosch University, Division of Physiotherapy in Medicine and Health Sciences, Stellenbosch, South Africa, 2Karolinska Institute, Division of Physiotherapy, Stockholm, Sweden, 3University of Sydney, Faculty of Medicine and Health, Sydney, Australia, 4Federal University of Minas Gerais (UFMG), Department of Physical Therapy, Belo Horizonte, Brazil
Learning objectives:
- Recognise common misunderstandings of statistical procedures.
- Clarify the correct meaning and appropriate uses of statistical analyses.
- Identify plain-English explanations of statistical results to facilitate involvement of patients in shared decision-making.
Description: In a survey of 1064 physiotherapists, the two most-common barriers to using evidence from clinical research in their management of patients were limited understanding of (i) research methods and (ii) statistical analysis, respectively [1]. This inadequate understanding of research and statistics amongst physiotherapists was confirmed internationally in a recent systematic review [2]. It is also evident in surveys of physiotherapy research, where ideal methods and analyses are not always used [3, 4, 5].
This symposium will explore some common misuses of statistical analyses and some common misunderstandings of statistical results. For each of these, a plain-English explanation of the correct use and interpretation will be provided. Attendees will then be referred to online resources that can provide further reference. For example, a recommended approach to statistical analysis that has been very slow to be adopted in physiotherapy research is the use of confidence intervals to estimate treatment effects instead of hypothesis testing with p values [3]. Here, the presentation will: explain the situations where these analyses might be relevant [6]; dispel some common misinterpretations of confidence intervals [7]; explain confidence intervals in plain English; discuss why estimation using confidence intervals is superior to hypothesis testing with p-values [8]; and list resources for further reading [9]. In other examples, such extra resources might include widely endorsed but accessible research reporting guidelines (such as the Explanation & Elaboration document for the CONSORT Statement), the Evidence in Practice series in J Orthop Sports Phys Ther, Masterclass papers in the Braz J Phys Ther, Research Notes in J Physiother, or Statistics Notes in the BMJ.
The range of statistical manoeuvres addressed by this symposium will be wide-ranging: assessing baseline comparability in trials, sample size calculation, comparing groups in trials, acknowledging uncertainty in estimates generated by research, positing the smallest worthwhile effect, gauging the effect of adherence, prognostic model research, and case-control studies.
This symposium provides ample opportunity to help rectify the current situation where clinicians report that they are more confident to read the Introduction and Discussion sections of a research paper than they are to read the Methods and Results sections [10]. If clinicians are more familiar with statistical concepts and can interpret statistical results in plain English, this should also facilitate shared decision-making with patients. The symposium should also assist physiotherapy researchers to make their research more scientifically rigorous and their research reports more acceptable for publication.
Implications/conclusions: It is crucial that the physiotherapy profession addresses the problems that some statistical procedures are commonly misused, and some statistical results are commonly misunderstood. Clinical physiotherapists can use the explanations in this symposium to improve their appraisal of published research. Research physiotherapists can use the advice and resources mentioned in this session to improve the scientific rigour of their research. By helping physiotherapists to strive to achieve these improvements, we stand to make a positive impact on the evidence that we use to move the profession forward.
References:
[1] Ramírez-Vélez R, Correa-Bautista JE, Muñoz-Rodríguez DI, et al. Evidence-based practice: beliefs, attitudes, knowledge, and skills among Colombian physical therapists. Colomb Med. 2015; 46:33–40.
[2] Da Silva TM, Costa C, Garcia AN, et al. What do physical therapists think about evidence-based practice? A systematic review. Man Ther. 2015; 20:388–401.
[3] Moseley AM, Elkins MR, Janer-Duncan L, Hush JM. The quality of reports of randomized controlled trials varies between subdisciplines of physiotherapy. Physiother Can. 2014 Jan;66(1):36-43.
[4] Freire AP, Elkins MR, Ramos EM, Moseley AM. Use of 95% confidence intervals in the reporting of between-group differences in randomized controlled trials: analysis of a representative sample of 200 physical therapy trials. Braz J Phys Ther. 2019 Jul 1;23(4):302-10.
[5] Moseley AM, Elkins MR, Herbert RD, Maher CG, Sherrington C. Cochrane reviews used more rigorous methods than non-Cochrane reviews: survey of systematic reviews in physiotherapy. J Clin Epidemiol. 2009;62(10):1021-30.
[6] Herbert RD. How to estimate treatment effects from reports of clinical trials. I: Continuous outcomes. Aust J Physiother. 2000;46(3):229-235.
[7] Hoekstra R, Morey RD, Rouder JN, Wagenmakers EJ. Robust misinterpretation of confidence intervals. Psychonomic Bull Rev. 2014;21(5):1157-64.
[8] Wasserstein RL, Schirm AL, Lazar NA. Moving to a world beyond “p< 0.05”. Am Stat. 2019:1-19.
[9] Kamper SJ. Showing confidence (intervals). Braz J Phys Ther. 2019;23(4):277.
[10] Lai NM, Teng CL, Lee ML. The place and barriers of evidence based practice: knowledge and perceptions of medical, nursing and allied health practitioners in Malaysia. BMC research notes. 2010;3(1):279.
[2] Da Silva TM, Costa C, Garcia AN, et al. What do physical therapists think about evidence-based practice? A systematic review. Man Ther. 2015; 20:388–401.
[3] Moseley AM, Elkins MR, Janer-Duncan L, Hush JM. The quality of reports of randomized controlled trials varies between subdisciplines of physiotherapy. Physiother Can. 2014 Jan;66(1):36-43.
[4] Freire AP, Elkins MR, Ramos EM, Moseley AM. Use of 95% confidence intervals in the reporting of between-group differences in randomized controlled trials: analysis of a representative sample of 200 physical therapy trials. Braz J Phys Ther. 2019 Jul 1;23(4):302-10.
[5] Moseley AM, Elkins MR, Herbert RD, Maher CG, Sherrington C. Cochrane reviews used more rigorous methods than non-Cochrane reviews: survey of systematic reviews in physiotherapy. J Clin Epidemiol. 2009;62(10):1021-30.
[6] Herbert RD. How to estimate treatment effects from reports of clinical trials. I: Continuous outcomes. Aust J Physiother. 2000;46(3):229-235.
[7] Hoekstra R, Morey RD, Rouder JN, Wagenmakers EJ. Robust misinterpretation of confidence intervals. Psychonomic Bull Rev. 2014;21(5):1157-64.
[8] Wasserstein RL, Schirm AL, Lazar NA. Moving to a world beyond “p< 0.05”. Am Stat. 2019:1-19.
[9] Kamper SJ. Showing confidence (intervals). Braz J Phys Ther. 2019;23(4):277.
[10] Lai NM, Teng CL, Lee ML. The place and barriers of evidence based practice: knowledge and perceptions of medical, nursing and allied health practitioners in Malaysia. BMC research notes. 2010;3(1):279.
Key-words: 1. Evidence-based physiotherapy 2. Research methods 3. Statistical analysis
Funding acknowledgements: Nil.
All authors, affiliations and abstracts have been published as submitted.
