Double coding and mapping using Abbreviated Injury Scale 1998 and 2005: Identifying issues for trauma data

IntroductionThe 2005 version of the Abbreviated Injury Scale (AIS05) potentially represents a significant change in injury spectrum classification, due to a substantial increase in the codeset size and alterations to the agreed severity of many injuries compared to the previous version (AIS98). Whilst many trauma registries around the world are moving to adopt AIS05 or its 2008 update (AIS08), its effect on patient classification in existing registries, and the optimum method of comparing existing data collections with new AIS05 collections are unknown. The present study aimed to assess the potential impact of adopting the AIS05 codeset in an established trauma system, and to identify issues associated with this change.MethodsA current subset of consecutive major trauma patients admitted to two large hospitals in the Australian state of Victoria were double-coded in AIS98 and AIS05. Assigned codesets were also mapped to the other AIS version using code lists supplied in the AIS05 manual, giving up to four AIS codes per injury sustained. Resulting codesets were assessed for agreement in codes used, injury severity and calculated severity scores.Results602 injuries sustained by 109 patients were compared. Adopting AIS05 would lead to a decrease in the number of designated major trauma patients in Victoria, estimated at 22% (95% confidence interval, 15–31%). Differences in AIS level between versions were significantly more likely to occur amongst head and chest injuries. Data mapped to a different codeset performed better in paired comparisons than raw AIS98 and AIS05 codesets, with data mapping of AIS05 codes back to AIS98 giving significantly higher levels of agreement in AIS level, ISS and NISS than other potential comparisons, and resulting in significantly fewer conversion problems than attempting to map AIS98 codes to AIS05.ConclusionsThis study provides new insights into AIS codeset change impact. Adoption of AIS05 or AIS08 in established registries will decrease major trauma patient numbers. Code mapping between AIS versions can improve comparisons between datasets in different AIS versions, although the injury profile of a trauma population will affect the degree of comparability. At present, mapping AIS05 data back to AIS98 is recommended.