UNDERSTANDING ACCIDENT MECHANISMS - AN ANALYSIS OF THE COMPONENTS OF 2516 ACCIDENTS COLLECTED IN A MAIM DATABASE

MAIM is an acronym for the Merseyside Accident Information Model, the current version of which is an intelligent, knowledge-based software s ystem. This paper describes a full scale trial using MAIM to record an d categorise data on the causes of injuries. The subjects studied were 2516 patients attending the Royal Liverpool University Hospital for d iagnosis and treatment of injuries between September 1992 and Septembe r 1993. The aims were to test the MAIM software, to show that it is po ssible to collect high quality accident information from hospital pati ents without writing, typing or coding, and to find methods of analysi ng the database to provide information that can be applied to accident prevention. Subsidiary aims were to show the extent of accidents whic h occur in sequences of more than one event and to confirm that it is possible to collect routinely the first and final events in accidents. No single description from the accident database could provide a pers pective of how the population was injured. The database project has pr ovided evidence on the complexity of accidents and the rare occurrence of identical combinations of all components and there were no two ide ntical accidents. This illustrates the difficulties of preventing acci dents. To assist the analysis and to focus attention on information us eful for accident prevention, an analysis method has been developed to identify objects and event verbs associated with both the causes of a ccidents and the causes of injuries. A coefficient can be computed whi ch links events either to the start or end of an accident. The coeffic ient allows accidents to be grouped so that typical or average acciden ts can be formulated and accidents with common features can be analyse d to show the course of average or typical events reported by patients . This allows a detailed examination of common causes of similar accid ents. Difficulties in classifying accidents have been highlighted; thi s is especially true where accident recording systems attempt to class ify accidents into broad groups. The analysis method provides an insig ht into the mechanisms causing accidents and injuries. (C) 1998 Elsevi er Science Ltd. All rights reserved.