81 Unfallstatistik
In an on-going project since 2005, ADAC has been analyzing accidents documented by the ADAC air rescue service. The knowledge derived from real-life accidents serves as a basis for new test configurations and assessment criteria. In 2007, ADAC began looking into the feasibility of international data collection. The idea of Global Accident Prevention was born. Three European partner clubs have begun pioneering the project (ÖAMTC, ANWB, and RACC). The aim is to set up an international accident research network to provide a steady stream of information on road accidents. The FIA Foundation supports ADAC in developing and coordinating this initiative.
The Centre for Automotive Safety Research (formerly the Road Accident Research Unit) at the University of Adelaide in South Australia has a history of in-depth crash investigation going back to the 1970s. In recent years, our focus has been on studying factors that contribute to road crashes, with an emphasis on the role of road infrastructure. Our method involves crash notification by the South Australian Ambulance Service and detailed investigation of the crash scene usually before the crash-involved vehicles have been moved. This at-scene data collection is supplemented with police crash reports, Coroner- reports including autopsy findings for fatal crashes, case notes from hospitals for all injured persons, structured interviews with crash participants and witnesses, and computerised reconstruction of the events of the crash. One of the most notable research findings to emerge from our in-depth work has been the relationship between travelling speed and the risk of crash involvement. By comparing the calculated free speeds of crash-involved vehicles (cases) with the measured speeds of non-crash-involved vehicles travelling on the same roads at the same time of day (controls), we were able to establish that an exponential relationship exists between travelling speed and the likelihood of involvement in a casualty crash. This was the case for both metropolitan and rural areas. This research prompted the reduction of some speed limits in Australia, which has resulted in notable decreases in crash numbers. Another finding of interest in our recent investigation of 298 mostly daytime crashes in metropolitan Adelaide was that medical conditions make a sizeable contribution to the occurrence of road crashes. We found that almost half of the drivers, riders and pedestrians involved in the collisions had at least one pre-existing medical condition, and half of these individuals had two or more such conditions. We found that a medical condition was the direct causal factor in 13% of the casualty crashes investigated and accounted for 23% of all hospital admission or fatal crash outcomes. A follow-up study of all hospital admissions for road crashes in Adelaide is now going ahead to look further at this problem. The paper also describes studies looking specifically at pedestrian crashes. These include studies of the relationship between travelling speed and the risk of a fatal pedestrian crash, and studies utilising real crash data to validate headforms and test dummies used in the assessment of the safety of new vehicles in the event of a collision with a pedestrian.
In recent years special attention has been paid to reducing the number of fatalities resulting from road traffic accidents. The ambitious target to cut in half the number of road users who are killed each year by 2010 compared with the 2001 figures, as set out in the European White Paper "European Transport Policy for 2010: Time to Decide" implies a general approach covering all kinds of road users. Much has been achieved, e.g. in relation to the safety of car passengers and pedestrians but PTW accidents still represent a significant proportion of fatal road accidents. More than 6,000 motorcyclists die annually on European roads which amounts to 16% of the EU-15 road fatalities. The European Commission therefore launched in 2004 a Sub- Project dealing with motorcycle accidents within an Integrated Project called APROSYS (Advanced PROtection SYStems) forming part of the 6th Framework Programme. In a first step, the combined national statistical data collections of Germany, Italy, the Netherlands and Spain were analysed. Amongst other things parameters like accident location, road conditions, road alignment and injury severity have been explored. The main focus of the analysis was on serious and fatal motorcycle accidents and the results showed similar trends in all four countries. From these results 7 accident scenarios were selected for further investigation via such in-depth databases as the DEKRA database, the GIDAS 2002 database, the COST 327 database and the Dutch element of the MAIDS database. Three tasks, namely the study of PTW collisions with passenger cars, PTW accidents involving road infrastructure features, and motorcyclist protective devices have been assessed and these will concentrate inter alia on accident causes, rider kinematics and injury patterns. A detailed literature review together with the findings of the in-depths database analysis is presented in the paper. Conclusions are drawn and the further stages of the project are highlighted.
Empirical vehicle crashworthiness studies are usually based on national or in-depth traffic accident surveys: Data on accident-involved cars/drivers are analysed in order to quantify the chance of driver injury and to assess certain risk factors like car make and model. As the cars/drivers involved in the same accident form a "cluster", where the size of the cluster equals the number of accident-involved parties, traffic accident survey data are typical multi-level data with accidents as first-level or primary and cars/drivers as secondlevel or secondary units (car occupants in general are to be considered as third level units). Consequently, appropriate statistical multi-level models are to be used for driver injury risk estimation purposes as these models properly account for the cluster structure of traffic accident survey data. In recent years various types of regression models for clustered data have been developed in the statistical sciences. This paper presents multi-level statistical models, which are generally applicable for vehicle crashworthiness assessment in the sense that data on single and multiple car crashes can be analysed simultaneously. As a special case of multi-level modelling driver injury risk estimation based on paired-by-collision car/driver data is considered. It is demonstrated that assessment results may be seriously biased, if the cluster structure inherent in traffic accident survey data is erroneously ignored in the data analysis stage.
Internationally, the need is expressed for harmonized traffic accident data collection (PSN, PENDANT, etc.). Together with this effort of harmonization, traffic accident investigation moves more and more in the direction of accident causation. As current methods only partly address these needs, a new method was set up. The main characteristics of this method are: • Accident/injury causation (associated) factors can objectively be identified and quantified, by comparison with exposure information from a normal population. • All relevant accident and exposure data can be included: human-, vehicle-, and environmental related data for the pre-crash, crash and postcrash situation (the so-called Haddon matrix). The level of detail can be chosen depending on interest and/or budget, which makes the method very flexible. In this paper the accident collection and control group method are presented, including some of the achieved results from a pilot study on 30 truck accidents and 30 control locations. The data were analyzed by using cross-tabulations and classification-tree analysis. The method proved useful for the identification of statistically significant causational aspects.
In Germany, in-depth accident investigations are carried out in the Hannover area since 1973. In 1999 a second region was added with surveys in Dresden and the surrounding area. Internationally, the acronym GIDAS (German In-Depth Accident Study) is commonly used for these surveys. Compared to many other countries, the sample sizes of the GIDAS surveys are much larger. The goal is to collect 1.000 accidents involving personal injuries per year and region. Data collection takes place by using a sampling procedure, which can be interpreted as a two-stage process with time intervals as primary units and accidents as secondary units. An important question is, to what extend these samples are representative for the target population from which they are drawn. Analyses show, for example, that accidents with persons killed or seriously injured are overrepresented in the samples compared to accidents with slightly injured persons. This means, that these data are subject to biases due to uncontrolled variation of sample inclusion probability. Therefore, appropriate weighting and expansion methods have to be applied in order to adjust or correct for these biases. The contribution describes the statistical and methodological principles underlying the GIDAS surveys with respect to sampling procedure, data collection and expansion. In addition, some suggestions regarding potential improvements of study design are made from a methodological point of view.
Annually within the European Union, there are over 50,000 road accident fatalities and 2 million other casualties, of which the majority are either the occupants of cars or other road users in collision with a car. The European Commission now has competency for vehicle-based injury countermeasures through the Whole Vehicle Type Approval system. As a result, the Commission has recognised that casualty reduction strategies must be based on a full understanding of the real-world need under European conditions and that the effectiveness of vehicle countermeasures must be properly evaluated. The PENDANT study commenced in January 2003 in order to explore the possibility of developing a co-ordinated set of targeted, in-depth crash data resources to support European Union vehicle and road safety policy. Three main work activity areas (Work Packages) commenced to provide these resources. This paper describes some of the outcomes of Work Package 2 (WP2, In-depth Crash Investigations and Data Analysis). In WP2, some 1,100 investigations of crashes involving injured car occupants were conducted in eight EU countries to a common protocol based on that developed in the STAIRS programme. This paper describes the purposes, methodology and results of WP2. It is expected that the results will be used as a co-ordinated system to inform European vehicle safety policy in a systematic, integrated manner. Furthermore, the results of the data analyses will be exploited further to provide new directions to develop injury countermeasures and regulations.
The need for improved EU level accident information and data was identified in the EU White Paper on Transport Policy (2001)1 and detailed in the Road Safety Action Plan (2003)2. The plan specifies that the EC will develop a road safety observatory to coordinate data collection within an integrated framework.
Road safety is a major preoccupation of the European Commission and the road transport industry and depends on numerous significant factors. In order to improve road safety and to plan effective safety improvement actions for truck transport, we must first identify the problems to be addressed, i.e. what are the main causes of truck accidents. The ETAC project, initiated by the European Commission and the IRU, was launched in order to set up a heavy goods vehicle accident causation study across European countries to identify future actions which could contribute to the improvement of road safety. The results will be based on a detailed analysis of truck accident data collected in seven European countries according to a common methodology which has been elaborated through numerous national and European projects. This paper describes the common methodology used to collect the information on the scene of the accident and to analyse the data so that the reconstruction of the crash events may be carried out. CEESAR proposes a methodology using its experience gained from over 10 years of accident data collection. This methodology is based on an in-depth investigation of the parameters involved in-an accident and linked to the driver, the vehicle, the road and their environment. In-depth investigation requires accident investigator presence on the scene of the accident in order to collect volatile information such as marks on the road, weather conditions, visibility, state and equipment of the vehicle, driver interview. Later, passive and active information is gathered, either at the hospital for the driver, at the garage for the vehicle or on the spot for the road geometry. A reconstruction carried out with the help of specific software and the analysis of the data collected and calculated enables the identification of the main causes of the accident and the future actions to plan in order to improve road safety as regards truck traffic.
Data concerning accidents involving personal injury which have been collected in the context of in-depth investigations on scene in the Hannover area since 1973 and in the Dresden area since 1999 represent an important basis for empirical traffic safety research. At national and international level various analyses and comparisons are carried out on the basis of "in-depth data" from the above mentioned investigations. In-depth data play a decisive role e.g. within the validation of EuroNCAP results on secondary safety (crashworthiness) of individual passenger car models. Thus, statistically sound methods of data analysis and population parameter estimation are of high importance. Since the 1st of August 1984 the "in-depth investigations on scene" in the Hannover area have been carried out according to a sampling plan developed by HAUTZINGER in the context of a research project on behalf of BASt. In the meantime a second region of in-depth investigation on scene was added with surveys in Dresden and the surrounding area. Internationally, the acronym GIDAS (German In-Depth Accident Study) is commonly used for the two above mentioned surveys. The objective of a current research project (topic of this contribution) is, among other things, to examine and adjust the previous weighting and expansion method for the two regional accident investigations to the current general conditions.
Due to recent years accident avoidance and crashworthiness on Austrian roads were mostly developed on national statistics and on-scene investigation respectively. Identification and elimination of black spots were main targets. In fact many fatal accidents do not occur on such black spots and black-spot investigation has reached a limit. New methods are required and therefore the Austrian Road Safety Programme was introduced by the Austrian Ministry of Transport, Innovation and Technology. The primary objective is the reduction of fatalities and severe injuries. Graz University of Technology initiated the project ZEDATU (Zentrale Datenbank tödlicher Unfälle) with the goal to identify similarities in different accident configurations. A matrix was established which categorizes risk and key factors of participating parties. Based on this information countermeasures were worked out.
NASS: the glass is half full
(2007)
The National Accident Sampling System (NASS) was born in the late 1970s. It was based on a substantial amount of experience and analysis of what was needed in the United States to understand the safety challenges of our highways. This work also showed how to collect high quality and useful crash data efficiently. Unfortunately, when Ronald Reagan - a President who believed in limited government - was elected, any hope of full funding for NASS was lost. The concept of 75 teams investigating about 18,000 serious crashes in detail annually was never realized. The system got up to 50 teams, then was cut to 36, and finally to 24 teams investigating fewer than a quarter of the originally anticipated number of crashes per year. Despite this, the NASS investigations provide a rich source of data, collected according to a sophisticated statistical sampling system to facilitate detailed national estimates of road casualties on our nation- highways and their causes. In addition, changes have been made in recent years to increase the number of more serious crashes of recent model vehicles to make the results more relevant to improving vehicle safety. A recent, detailed examination of hundreds of rollovers has provided considerable insight into rollover casualties and into what can be done to reduce them. Some of these results will be presented that show the value of the NASS system. Our experience with NASS and the Fatal Accident Reporting System (FARS) suggests a number of improvements that could be made in the United States" crash data systems. It also provides justification for a doubling or tripling of our national expenditures on crash data collection.
In the context of this study, different data sources for accident research were examined regarding their possible data access and evaluated concerning the individual quality and extent of the data. Analyses of accidents require detailed and comprehensive information in particular concerning vehicle damages, injury patterns and descriptions of the accident sequence. The police documentation supplies the basic accident statistics and is amended in the context of the forensic treatment by further information, e.g. by medical and technical appraisals and witness questionings. As a new approach to the data acquisition for the analysis of fatal traffic accidents, the information was made usable which was collected by the police and by the investigations of the public prosecutor. The best strategy for obtaining reliable, extensive and complete data consists of combining the information from these two sources: the very complete, but elementary statistic data of the Niedersächsisches Landesamt für Statistik (Lower Saxony State Authority of Statistics), based on the police documentation as well as the very extensive accident information resulting from the investigation documentation of the public prosecutor after conclusion of the procedure, the so-called Court Records. Of all 715 fatal traffic accidents, which happened in the year 2003 in the German State of Lower Saxony, 238 cases were selected by means of a statistically coincidental selective procedure based on a statistically representative manner (every third accident). These cases cover the investigation documents of the 11 responsible public prosecutor- offices, which were requested and evaluated while preserving the data security. Of the 238 cases 202 cases were available, which were individually coded and stored in a data base using 160 variables. Thus a data base of a sample of representative data for fatal accidents in Lower Saxony was set up. The data base contains extensive information concerning general accident data (35 variables), concerning road and road surface data (30 variables), concerning vehicle-specific data (68 variables) as well as concerning personal and injury data (27 variables).
The accident research project in Dresden was founded in July 1999. To date over 6.000 crash investigations have been undertaken. About 10.000 vehicles have been documented and over 13.000 participants have been debriefed. But there is much more than this scientific success. Because of the interdisciplinary character between the medical and technical focus, the project affords an important contribution for the education of the involved students. Over 200 students of different fields of study have got experiences not only for the occupational career. This lecture describes the additional effects of the accident research project regarding the education of the students, the capacity for teamwork and learning about dealing with accident casualties.
Ziel dieses Forschungsvorhabens war es, das bisherige Gewichtungs- und Hochrechnungsverfahren für die örtlichen Unfallerhebungen in den Regionen Hannover und Dresden zu überprüfen und an die aktuellen Rahmenbedingungen anzupassen. Darüber hinaus sollten neue Möglichkeiten der gemeinsamen Hochrechnung von Ergebnissen aus beiden Erhebungsgebieten unter Berücksichtigung der aktuellen Datenlage, insbesondere in der amtlichen Unfallstatistik, untersucht und entsprechende statistische Verfahren entwickelt werden. Der Stichprobenplan der Erhebungen folgt einem zweistufigen Stichprobenverfahren. Tests mit einem zweistufigen Hochrechnungsverfahren der Hannover-Stichproben 2000 und 2001 auf die Gesamtheit Hannover haben jedoch ergeben, dass die theoretisch zu erwartenden Vorteile dieser zweistufigen Methode im Vergleich zur "einfachen Gewichtung" in der Praxis relativ gering sind. Unter Beibehaltung der bisherigen Gewichtungsprozedur (Anpassung an eine n-dimensionale Kontingenztabelle) wurden daher für die regionalen Hochrechnungen Dresden und Hannover alternative Methoden entwickelt, in der z.B. das Merkmal Ortslage durch die Unfallart ersetzt oder zusätzlich die Anzahl Unfallbeteiligter zur Gewichtung herangezogen wird. Leider erbringen diese Verfahren im Vergleich zur bisherigen (simultanen) Gewichtung nach Unfallschwere, Tageszeit und Ortslage nur wenige oder gar keine Verbesserungen der Anpassungsgenauigkeit bei Merkmalen der amtlichen Statistik, die nicht in die Gewichtung eingehen. Unabhängig vom Gewichtungsverfahren lässt die Abbildungsgenauigkeit bei einzelnen Variablen sehr zu wünschen übrig. Auf der Basis der Stichprobendaten 2000 wurde ferner noch ein Gewichtungsverfahren für Hochrechnungen auf das Bundesgebiet entwickelt und anhand der beiden Einzelstichproben sowie der gepoolten Stichprobe (Dresden plus Hannover) getestet. Die gepoolten Stichprobe zeigte die besten Ergebnisse. Allerdings ließen sich auch auf der Grundlage der gepoolten Daten nicht bei allen Merkmalen (gemeint sind hier Merkmale, die nicht Gewichtungsmerkmale sind) Verbesserungen des Fits erzielen. Es bleiben relativ große Abweichungen zwischen der gewichteten gepoolten Stichprobe und den bundesdeutschen Verteilungen. Es ist zu erwarten, dass auch bei vielen GIDAS Merkmale durch die Gewichtung keine essentielle Korrektur der Stichprobenverzerrungen erzielt werden kann. Insgesamt hat sich gezeigt, dass es auch mit alternativen, meist hierarchischen Gewichtungsverfahren nicht möglich ist, alle Merkmale, deren Verteilungen aus der amtlichen Statistik bekannt sind, mit hinreichender Genauigkeit an die Verhältnisse der Grundgesamtheit anzupassen. Es ist zu erwarten, dass sich dies bei den eigentlich hochzurechnenden originären GIDAS-Variablen ähnlich darstellt. Somit muss der Ansatz eines einheitlichen Verfahrens für alle Jahre und beide Erhebungsgebiete in Frage gestellt werden. Für besonders wichtige Fragestellungen sollte daher eine spezielle, auf das jeweilige Untersuchungsmerkmal abgestellte Gewichtung durchgeführt werden, wie dies exemplarisch am Beispiel der maximalen Kollisionsgeschwindigkeit gezeigt wurde. Schließlich werden im Bericht noch Empfehlungen zur Datenqualität und zu der Frage gegeben, wie sich das derzeit praktizierte Stichprobenverfahren im Hinblick auf die Gewinnung einer repräsentativen Stichprobe möglicherweise verbessern lässt.