The Influence of Methodological Issues on the Results and Conclusions from Epidemiological Studies of Sports Injuries

Abstract

Data obtained from epidemiological studies of sports injuries are an essential requirement for developing injury prevention, treatment and rehabilitation strategies. Although many authors have discussed the strengths and weaknesses of research methods employed in epidemiology, the potential effects that variations in research design and methods of analysis can have on study conclusions have not been clearly illustrated. This article addresses a number of methodological issues and illustrates their potential effects using examples based on injury data obtained from a single, large epidemiological study in professional rugby union. The examples demonstrate that conflicting conclusions can be reached depending on how the data are collected and analysed. The pivotal roles played by injury definition (loss-of-time, missed matches, diagnostic assessment and surgery), recurrent injury definition (clinical judgement and same injury/same location/same season), method of reporting injuries (number, proportions and incidence) and method of calculating incidence (injuries per 1000 player-hours, per 1000 athlete-exposures and per 1000 matches) are highlighted and illustrated. Other examples show that if training and match injuries are combined, the incidence of injury is more likely to reflect the incidence of training injuries but the distributions of injuries are more likely to reflect the distributions of match injuries. An example is presented that demonstrates that the identification of injuries causing the greatest concern within a sport depends on whether the assessment is based on injury incidence, severity or risk. Finally, examples are presented to show that the relationships identified between sports injuries and risk factors may be dependent on whether case-control or cohort study designs are used. Although there are no simple solutions available to resolve the issues raised, the discussion demonstrates the importance, at least within a sport, of reaching consensus agreements on acceptable study designs and methods of data analysis in sports epidemiology.