Clubs volunteering for sports injury prevention research - is there any selection bias?

Fundamental to any epidemiological research is the recruitment of a representative sample of participants. Selection bias can occur if volunteers self-select if they are worried about the specific exposure or injury outcome. Minimising selection bias is a particular concern when recruiting participants into a randomised controlled trial (RCT) of the effectiveness of a new intervention. Group RCTs, which randomly allocate a ‘pre-formed’ group rather than individuals to each study arm, can help to remove bias and eliminate contamination effects. Group randomisation of teams of players is an appropriate strategy for sports injury studies since there is a natural grouping of players into teams under the wider control of a single coaching team. However, teams of players are still required to volunteer for such studies. Ideally this should be done before randomisation to intervention arms. The Australian Football Injury Prevention Project (AFIPP) was an RCT to examine the effectiveness of headgear and mouthguards in Australian football. Community football teams from clubs in metropolitan Melbourne, Victoria were allocated to four intervention arms: (a) soft-shell headgear, (b) custom-made, tri-laminate mouthguards, (c) headgear and mouthguards, and (d) control. The sampling process required clubs to nominate whole teams of players to the same intervention arm. Clubs were not aware a priori of what protective equipment, if any, their players would be required to wear during the study. The aim of this study reported here was to identify any selection bias in the volunteering of clubs for AFIPP. Methods and procedures