Secondary Logo

Journal Logo


From Study Design and Analysis to Conclusion: New Horizons for Epidemiological Rigor in Sport Medicine

Best, Thomas M MD, PhD; Shrier, Ian MD, PhD

Author Information
Clinical Journal of Sport Medicine: May 2007 - Volume 17 - Issue 3 - p 175-176
doi: 10.1097/JSM.0b013e318061b75d
  • Free

Everyone involved in the care of athletes and their well being recognizes the importance of injury prevention and treatment. With the era of evidence-based medicine upon us, one assumes that the data and their interpretation from current epidemiological studies will help us understand more about causation and prevention. However, variations in definitions, methodologies, and analyses amongst studies can lead to differences in results and conclusions obtained. It is therefore important to be aware of these differences when interpreting the literature. The objective of this thematic issue is to help sport medicine clinicians determine if the appropriate conclusions are made in a published study and to introduce sport medicine researchers to some of the newer epidemiological developments that have implications for study design and analysis. We have specifically chosen articles that cover 3 main areas of interest: how do we define injury, methods for surveillance studies, and interpreting the analyses.

Defining injury is fundamental in the field of injury prevention, and one would think that a consensus would have already been reached. But it is not. This issue begins with a consensus statement on injury definitions from the Rugby Union. Because the overall risk for injury is related to the product of injury incidence and severity, Fuller suggests that one must also define recurrent injury, severity, and training and match exposures. His paper also provides criteria for classifying injuries in terms of location, type, diagnosis, and causation. Although there is consensus in the Rugby Union, areas for debate remain. One of the major issues still being discussed is whether an injury should be defined by time-loss or anatomical injury. We have therefore also included 2 articles (by Hodgson and colleagues, as well as Orchard and Hoskins) that debate this issue. In fact, each definition addresses different questions, and these articles will help clinicians determine if the most appropriate definition was used to answer the question posed by the authors and which study most closely addresses the clinician's own question. In a second paper, Fuller and colleagues provide a strong rationale and a theoretical recording and reporting framework, suggesting the need for recording time-loss recurrent injuries and distinguishing between exacerbation of an existing injury and a true re-injury.

Almost as fundamental as the general definition of injury is the method used to classify specific types of injuries. About 15 years ago, the Orchard Sports Injury Classification System (OSICS) was first described as a means to classify diagnoses and examine incidence of injury in Australian football players. The system has evolved over time, and OSICS-10 is presented herein by Rae and Orchard for clinicians and researchers interested in injury description and classification. One particular advantage of OSICS-10 is the ability to allow diagnoses to be quite specifically classified. For example, hamstring strains can now be retrieved by the traditional 1-3 grading system, which should permit greater accuracy and specificity for injury documentation. As an alternative approach, the Sport Medicine Diagnostic Coding System (SMDCS), first developed in 1991, is updated by Meeuwisse and Wiley and provides an interactive approach to injury documentation. A downloadable data file for both systems is provided that allows other groups to use the OSICS and SMDCS for their own purposes.

In sports injury research, it is especially important to distinguish between incidence and prevalence. As applied to sports injury epidemiology, the prevalence of injury is the proportion of athletes who have an existing injury at any given point in time, whereas injury incidence is the number of new injuries that occur over a specific period of time, such as from the start of the season. There are many ways to present the results of a study, and the method can influence the conclusions of the reader. Hopkins et al review some of the different methods in a clinician-friendly format.

Although the randomized, controlled trial (RCT) is considered by many to be the gold standard for clinical research, there are different methods of randomization. Emery provides an excellent example of how cluster randomization methods affect (1) sample size calculations, and (2) the inferences that should be made from the study. For example, if groups are randomized instead of individuals (eg, randomize different teams), a simple analysis of the number of people who got injured might suggest that the results are statistically significant when in fact they are not. Emery carefully explains the reasons why these methods are important and how authors can apply some simple adjustments to ensure the appropriate inferences are made.

Finally, all analyses are based on an underlying theoretical model. In 1992, Van Mechelen first proposed a sport injury model that related to injury prevention. A short time later, Meeuwisse and colleagues developed a model of causation to account for the interaction of both intrinsic and extrinsic risk factors. In this issue, we present 2 articles (Meeuwisse et al; Shrier) that review newer developments that may lead to a change in the way injury studies are designed and analyzed in the future. Meeuwisse and colleagues propose a new model representing a dynamic approach to risk factors and injury event during participation in sports. Built on their previous work, the model incorporates the consequences of repeated participation in sports, both with and without experiencing injury. It emphasizes a dynamic, recursive model of sports injury by painting a recursive picture where an athlete can enter a given athletic event cyclically with a differing set of risk factors, after experiencing either injury or non-injury event. Shrier reviews how we make causal inferences from studies. Recent epidemiological research demonstrates that some of the traditionally used intuitive approaches can actually lead to inappropriate conclusions, and sport medicine researchers will need to take these issues into account in the design and analysis of their studies.

We thank our authors for their time and expertise in highlighting important topics in the field of epidemiology and Sport Medicine. We hope that this thematic issue aids the busy clinician when interpreting the literature and study design and also assists clinical researchers in the design of their own studies.

Copyright © 2007 Wolters Kluwer Health, Inc. All rights reserved.