The objective of this study was to specify of a limited data coding protocol and transformation procedures for development of comparable datasets. WP3 is concerned with hospital registration of traffic casualties. In the first task of this WP, the three different registrations (DUHAT from Spain; ARVAC from France; LMR from the Netherlands) have been described and compared (Deliverable D3). According to the plan, this task is to be followed by activities concerning the production of a common data protocol (task 2, D6), and activities concerning the production of a common analysis protocol (task 3, D7), while also activities concerning the linking/matching of datasets have to be carried out. All these activities are being applied to ultimately carry out relevant crash-injury analyses on the three separate datasets in order to help improve traffic safety in the EU, especially car safety (task 4). In order to be able to produce a Data protocol (Task 2.2, Deliverable D6), activities during the first half of 2004 (M13-M18) aimed at establishing comparable datasets. First results of this work pointed to a number of practical barriers, ranging from differences of definition of variables, to differences of content of the data due to their different scopes. In fact, it was found that the strict separation of activities described in the contract with respect to Task 3.2 (Data protocol), Task 3.3 (Analysis protocol), and Task 3.4 (the actual analyses) could not be maintained. Therefore, instead of the sequential approach as originally planned, a more parallel approach of these activities was chosen. This is due to the following: As to be expected, the three sets of data to be used primarily for this WP (ARVAC from Lyon/France; DUHAT from Barcelona/Spain; and LMR from the Netherlands) are different both with respect to definitions of variables used, and with respect to background of the area in which the data are registered. This has already been described in Deliverable D3. Even after adjustment for the most relevant differences, and using the same injury and injury- severity descriptors (such as AIS-predotcodes, AIS-severity, and Barell type and location of injury), some unexplained differences continue to exist. This was the preliminary result of an analysis based on selected tables from the three different registrations. Some of these differences may well be linked to the inherent differences between the traffic situations in the three areas of registration, others may be due to the fact that two of the registration cover a far broader range of traffic casualties (ARVAC and DUHAT) than the third (LMR) which solely covers data from hospital in-patients. It is thought to be worthwhile to look more in depth into the background of these differences before proceeding according to the original plan (which includes the data protocol, the analysis protocol, as well as carry out linking procedures as planned within each of the partners' other available datasets). This deliverable will concentrate on one aspect, data selection, because it will be the basis on which all further activities and analyses will be dealt with. (Author/publisher)
Abstract