Data collection and reporting
Road crash and injury data systems
Numerous groups are interested in road safety and need to use road crash data. They include road safety professionals, highway engineers, the police, lawyers, research groups, politicians, teachers, statisticians, motor manufacturing companies, vehicle fleet operators, insurance companies and even members of the public (perhaps in relation to an insurance claim). They all tend to have different needs and reasons for wanting the data and, since there are practical limitations on the amount of information that can be collected, the content of crash databases has to balance the data wanted against the burden of collecting and entering data. The database should nevertheless be as reliable and comprehensive as possible. Some of the key reasons for collecting crash and injury data are to:
1) overview the problem,
2) monitor trends,
3) identify high risk/problem groups,
4) identify high risk, hazardous locations,
5) enable objective planning and resource management,
6) evaluate effectiveness and monitor achievement of targets,
7) make international comparisons and
8) provide evidence for prosecution.
Information sources
Potential sources of data include:
1) Police crash data
2) Hospital and medical data
3) Insurance data
4) Vehicle operators
5) Special surveys
Most countries have made it a legal requirement that road crashes, at least those involving casualties, be reported to the police. This requirement is often reinforced by insurance company rules which require claimants to follow this law.
However, in many countries a significant number of road crashes are not reported to the police. The level of under-reporting varies considerably from country to country, but even where there is a legal requirement to report crashes involving personal injury, studies of hospital data have demonstrated considerable under-reporting (as high as 97 per cent missing for some categories - Aeron-Thomas, 2000). Under reporting is less of a problem for the more serious crashes but even for fatal crashes there are problems with definition of death. For example some countries define death as "dead on the spot" whereas others use "dead within 30 days" which is the recommended standard.
Hospital data are an important source of information, especially about injury and treatment, but information relating to the scene of the crash is often missing in medical data. Also, in poorer countries, hospitals cannot yet cope with the burden of collecting accurate data on road crash victims. Nevertheless a minimum requirement should be the collection of reliable data about causes of death.
For trend analyses and overviews of the scale of road death and injury it is recommended that both health and transport (usually police) records be examined. Whilst direct matching of these two sources is difficult, comparisons of aggregate data from both sources can help to confirm trends or identify problems with the data where there are differences between the two sources.
The other data sources such as insurance have more limitations and it is usually the police who provide the main source of crash information for national systems.
Content of the database
The police have limited time to spend on recording crash information. Therefore they need to strike the right balance between the amount of detail required by all stakeholders and the ease with which they can collect the data according to their expertise and the tools and support available. Ideally the police should not have to record the same information twice and therefore the data required for evidence and for the database should be made as compatible as possible. This can be done by making the form used for data entry acceptable as evidence in the courts.
For the road safety professional the database should at least be able to answer the following questions:
| Where did crashes occur: | location by map coordinates, road name and km post, class |
| When did crashes occur: | by year, month, day of month, week, time of day |
| Who was involved and who was injured: | People in relation to vehicles, vehicles, animals, roadside objects |
| What was result of the crashes: | worst severity of injury or property damage |
| What environmental conditions: | poor light, weather, road surface condition |
| How or why did the crashes occur: | collision type, driver fault type |
The exact details to be collected for the database are usually decided by a multisectoral committee to ensure that all stakeholders are consulted and that a reasonable spread of data are collected within the capacity of the police.
Ideally, the required information should be completed at the scene on a user-friendly form, or booklet. This form should be easy to use and be acceptable as evidence in the courts. The form is usually structured for easy data entry which means grouping information by:
- general details,
- vehicle and driver details,
- casualty details,
- location information and sketch map, and
- simple summary description.
Examples of relatively easy reporting forms, source TRL
The identification of location is often not an easy task. Modern geographic positioning systems (GPS) can help pinpoint the locations but without this technology the police will need to use maps or other ways of coding location such as km posts along highways or link/node systems in urban areas. A sketch diagram is also important to show the movements of the involved parties just before the crash.
Example sketch of crash scene
As the amount of data is quite large and often collected under difficult circumstances it is essential that the quality of the data collected is effectively controlled. This means that missing data and errors are quickly detected by a supervisor and then corrected by the original investigating officer.
An example of a questionnaire form (designed and used by TRL/DFID) for a road safety community analysis in Bangalore India can be viewed by clicking here: road safety survey.pdf
For more information on data collection, sources, injury surveillance systems, analyses and indicators a good source is Chapter 2: The Global Impact in "The World Report on Road Traffic Injury Prevention" by WHO and the World Bank, p52 -
http://www.who.int/world-health-day/2004/infomaterials/world_report/en/chapter2.pdf
Software Packages for Database Management and Analysis
Crash databases are complex in structure but powerful and easy to use software is available. The packages chosen should be simple to use, with easy data entry, full editing and back-up facilities, and logical internal checking routines to ensure that the data is as accurate as possible when entered. Validation of data should be done as close to the data entry as possible, to enable rapid follow up and correction of errors in the data records.
Key software facilities usually include:
Data entry, data management and validation checks: enable electronic file management and provides standard logical checks within fields and between fields which are usually automatically applied on data entry.
Example of data entry form, Fiji, source TRL
Database searches: enable the user to find quickly an individual record or group of records satisfying user-specified conditions. Provide a fast means of identifying problems by people or location characteristics and enable the production of sub sets of files for special analysis, such as a road section or urban area.
Tabulations: provide cross tabulations of fields by crash frequencies, or casualties, or the vehicles involved. Tables can be set up as standard for regular and automatic output.
Presentation graphics: display figures and tabular results graphically in the form of bar charts or pie charts.
Example of graphic illustration, Bangalore, Source TRL
Hazardous location identification and investigation: provides the user with a list of the "worst" sites, however defined, in any area of interest. Now usually uses electronic maps and powerful onscreen tools for selecting sites and areas and carrying out on line analysis.
Example of graphic illustration of high-risk road crash sites. Fiji, Source TRL
FIJI Crash location analysis, Source TRL
Stick diagrams: provide engineers with a simple visual tool for searching for crash patterns at selected sites and areas. Key features of a crash (e.g. type of collision, hour of day, vehicle types) can be selected and these are shown using symbols where each column represents one crash.
Collision diagrams: are another tool for engineers. These show in a simple way all the types of crash and related road user movements over a period of time.
Example of crash diagram, Malaysia. Source: TRL and University Putra Malaysia.
Performance indicators: are important for assessing risk and monitoring changes. This facility requires traffic, demographic and geographic data to be linked to the crash database.
The system should enable the production of a number of useful outputs. These should include regular reports varying from simple 24 hour reports often required by the police to comprehensive annual reports available as a public document.
The system also needs to provide widespread access and it is important that local road safety teams can analyse their own data. Where there is strong decentralisation of the data collection and entry and road safety responsibility, then the national system should meet the local needs but also have a standard and simple system for collating the data at the national level.
Countries should also consider participating in regional systems such as the European International Road Traffic and Accident Database (IRTAD) (http://www.bast.de/htdocs/fachthemen/irtad/) or the Asia - Pacific Road Accident Database (APRAD) run by the UN-ESCAP (http://www.unescap.org/tctd/pubs/files/aprad_usermanual.pdf). This will encourage international harmonisation of the databases and ideally countries which are revising their systems should take into account the data required at the international level.
Sources
AERON-THOMAS A (2000). Under-reporting of road traffic casualties in low income countries. TRL Report PR/INT/199/00. TRL Ltd, Crowthorne.
Asian Development Bank (1999). Road Safety Guidelines. ADB, Manila, Philippines.
Moon J (2003). APRAD and road safety in ASEAN. Proceedings of the 3rd GRSP ASEAN Seminar, UNESCAP, Bangkok, Thailand.