Use of Statistics in Physical Therapy Over a 2-Year Period-2000-2002: Implications for Educators

Journal of Physical Therapy Education, Fall 2003 by Bandy, William D

Background and Purpose. Statistics are a tool that can be used to organize, summarize, and interpret information provided by research. An analysis of which statistical techniques are used most often in journals will help to identify those techniques most essential for understanding research. Similar studies have been performed to analyze statistics used in other disciplines such as psychology, education, and various areas of medicine, hut no such study has been undertaken in the field of physical therapy. The purpose of this content analysis is to identify the types of statistical techniques used in Physical Therapy. Methods, In order to determine the types of statistics and the frequency of techniques used, all articles from July 2000 through July 2002 were examined for the frequency of statistical techniques. Results. Findings of this study showed that the 10 most common statistical techniques used were as follows (in order of most frequent to least frequent): descriptive statistics, one-way analysis of variance (ANOVA), t test, factorial ANOVA, intraclass correlation, appropriate post hoc analyses, Pearson correlation, regression, chi square, and nonparametric tests analogous to the t test. Understanding these 10 statistical techniques would allow the reader to understand 80% of statistical techniques used in Physical Therapy during this 2-year time period. Discussion and Conclusion. Comprehension of these most common statistical techniques is very important in allowing the reader to begin to understand and critique the physical therapy literature. This article provides new and unique information, which may help educators design statistics courses for professional (entry-level) physical therapist students.

Key Words: Research, Statistics.

INTRODUCTION

Outcomes research to provide evidence for physical therapy examination and intervention has become a primary goal for the profession of physical therapy.1 In order to provide the graduating physical therapist with the tools needed to effectively and accurately examine outcomes research, a solid background in research design and statistical analysis is an advantage.2,3 Evidence of the importance of research in the profession of physical therapy can be found in the Evaluation Criteria for Accreditation of Education Program for the Preparation of Physical Therapists.4 The evaluative criteria for graduates of a professional (entry-level) physical therapy program related to research come under the heading of "Critical Inquiry and Clinical Decision-making" and include:

3.8.3.10. Evaluate published studies related to physical therapy practice, research, and education.4(p15)

3.8.3.12. Participate in scholarly activities to contribute to the body of physical therapy knowledge (eg, case reports, collaborative research).4(p15)

A case could be made that in order to most effectively meet the evaluative criteria of evaluating published studies and participating in collaborative research, a basic knowledge of statistics is needed. Portney and Watkins3 defined statistics as a means of organizing, summarizing, and analyzing a compilation of numbers representing empirical observations from a group of individuals (raw data) so that the meaning of their numbers can be communicated. Domholdt2 emphasized the importance of understanding statistics in order to provide a strong basis for understanding the data analysis and results sections of research articles in the physical therapy literature and thereby evaluate published studies. Finally, LaPier emphasized that "with the recent flood of statistical computer programming understanding of the how of a statistical test is less important than understanding the why of statistical tests."5(p19)

With these definitions and opinions in mind, to what level should physical therapy educators teach statistics as part of their research and critical inquiry component of the curriculum? More specifically, given that so many statistical tests exist-too many to include even in textbooks devoted solely to statistics-which statistical tests should he taught to entry-level physical therapists?

In a "focused" issue on research in Gerinotes, the Journal for the Geriatrics Section of the American Physical Therapy Association (APTA), LaPier and Donavan6 reviewed "common" statistical procedures. The authors included t tests, analyses of variance (ANOVA) (one-way and factorial), Pearson correlation, intraclass correlation coefficient (ICC), linear regression, and nonparametric tests analogous to the t test in their description of "common" statistical tests. The inclusion of these tests as "common" is consistent with an article published in the journal of the Cardiopulmonary Section of APTA written a year earlier.5 However, these 2 articles do not provide evidence that these statistical procedures arc truly "common" in physical therapy literature.

One method that can be used as a guide for developing a baseline for the statistical procedure that should be presented to physical therapist students in an entry-level curriculum is to survey the statistics mat are utilized in the official journal of APTA-Physical Therapy. Such an analysis will identify the statistical techniques that are used most often and, therefore, may be most essential for the understanding and evaluation of published studies related to physical therapy practice, research, and education. To date, no study has provided information as to the types and frequency of statistics utilized in Physical Therapy. Therefore, the purpose of this study was to perform content analysis to identify the types of statistics used in Physical Therapy.