Computer assisted telephone interviewing (CATI) is an increasingly common survey methodology in the public health arena (Szuster 2003). CATI methodology has been applied both in highly specific areas of health behaviours research (Robertson et al. 2000) and broad-scale national health surveys (Bolen et al. 1999). In recent years, the CATI Technical Reference Group (TRG)-a subcommittee of the National Public Health Information Working Group-have been developing topic-specific question modules for population health surveys in conjunction with key informants (Szuster 2003, Wilson et al. 2001).

In 2004 the National Injury Surveillance Unit (NISU) published a discussion paper regarding the development of an injury-related question module, in consultation with the CATI TRG (Bradley and Harrison 2004). The report discussed the application of CATI surveys in injury surveillance both in Australia and internationally. In summary, the report advocated the development of a CATI question module addressing behaviours, attitudes and knowledge regarding injury in order to complement existing national data on serious injury.

The development of CATI question modules is a rigorous process, including a cognitive testing phase, where the proposed question module is presented to a small number of respondents and analysed by cognitive psychologists, and a wider field- testing phase, which addresses reliability and validity issues using a test-retest protocol (CATI TRG 2003, Szuster 2003). Cognitive testing of the proposed injury question module was conducted at the ABS cognitive laboratory in September 2003 (ABS 2003), presenting questions addressing safety behaviours and practices, home and community injury concerns, injury preventability beliefs, alcohol and injury, and exercise participation in older people. Problems with interpretation and terminology were identified and response category refinements were also recommended. Questions regarding alcohol consumption and injury were identified as particularly problematic and it was recommended that this item be discontinued (ABS 2003).

Field testing of the injury question module, among others, was conducted in mid-2004 by the NSW Health Survey Program on behalf of the CATI TRG. The module used in the field tests was significantly truncated from that applied in the cognitive testing, retaining only questions relating to the presence of smoke alarms and first aid kits in the home and questions relating to falls in the elderly (more specifically, exercise in the elderly as exercise is known to reduce falls risk, e.g. Chang et al. 2004). Field testing of this version confirmed the issues identified in the cognitive testing phase, especially the importance of careful selection of wording and terminology in the questions. The field testing also demonstrated the limited utility of asking questions about falls injury incidence (unless the sample size is large) and questions regarding specific forms of exercise. From this we conclude that the injury module as tested here is of limited utility for injury prevention researchers and requires further development.

As both NISU and the CATI TRG felt that further development of the module was necessary in order to produce a question set which will be more relevant for both injury prevention research and inclusion in a standardised national question set, a joint workshop between the TRG and the Strategic Injury Prevention Partnership (SIPP) was hosted by the Public Health Information Development Unit (PHIDU) in Adelaide in October 2004. Discussion involving members of these groups highlighted a structural difficulty confronting development of a module that would meet design constraints required by the CATI TRG (chiefly a very small number of items) yet provide information that is meaningful and useful to injury prevention policy makers and researchers. A result of this workshop was the establishment of the Injury Prevention Population Surveys Interest Group (IPPSIG). IPPSIG is a web-based discussion list, intended to be a forum to promote and enable discussion and collaboration on the development and use of population surveys, and with the primary purpose being to provide expert advice to the CATI TRG on the further development of the injury question module.

Some discussion between SIPP and CATI TRG members was conducted via IPPSIG but the forum could not solve the conundrum of how to present questions which would be useful for injury prevention researchers in the scope stipulated by the CATI TRG's mandate. The CATI TRG required a very small set of closed-response questions however injury is a diverse topic with characteristics which complicate adequate coverage by such a question set. Prevalence of recent non-trivial injury is low, resulting in low positive response proportions for questions focusing on injury experience of respondents. Questions focusing on exposure to injury risk factors can avoid this constraint, but risk factors are diverse, requiring numerous questions if adequate coverage of the topic is to be achieved. Questions dealing with knowledge and attitudes concerning injury risk and injury prevention have potential for adequate coverage in a small question module but were outside the preferred scope of the CATI TRG. Further work needs to be conducted in both the development of a CATI injury module suitable for inclusion in the National Health Data Dictionary (NHDD) and, outside the mandate of the TRG, in the development of key indicator questions which generate useful and timely data for injury prevention professionals.