Traumatic Brain Injury (TBI) in RAIDS collisions

Learn how to investigate TBI in UK in-depth road traffic collision database, RAIDS. Further details about how to get free access to RAIDS can be provided by contacting RAIDS@dft.gov.uk.

What is RAIDS data?

The RAIDS database is Great Britain’s free-to-access in-depth road traffic collision (RTC) database. RAIDS stands for Road Accident In-Depth Studies. It captures detailed information right from the moment a collision occurs through to the diagnosed injuries for all injured subjects, facilitating a range of research topics such as exploring the links between demographics, biomechanics and traumatic brain injury. It is a severe subset of collisions that happen across Great Britain, collected with the aim of mitigating the most severe collisions on our roads. The dataset is therefore not nationally representative, because it includes a higher proportion of fatally and seriously injured individuals.

The interactive PowerBI dashboard can be used to engage directly with the RAIDS data to see how different road user and collision characteristics including demographics such as age influence the injuries sustained. For more details on TBI in RAIDS, see Example of Research page.

What data fields are included?

RAIDS includes detailed information about the collision site (including highway features and environmental factors), vehicle damage and injuries sustained in the collision. For each case, 2,516 fields of data are collected. Data collection is ongoing, as of February 2025:

Why is RAIDS collected and how does this influence the data available?

The data collected for the Department for Transport is designed to identify the collision scenarios, including contributory factors relating to the vehicle, road and road users, which lead to collisions of varying severities. 2,500+ fields are collected, with the following key purposes:

Identify how people are injured in road traffic collisions, the injuries they sustain, and how these correlate to vehicle characteristics and highway design features.
Establish the extent to which a range of safety related measures have reduced the risk of injury to road users involved in collisions.
Identify measures to reduce further the risk of collisions and injuries (in terms of vehicle design and safety, the road environment and traffic management and human factors).

How is injury and TBI captured in RAIDS?

RAIDS uniquely captures the injuries sustained using a range of sources:

Figure 2 The range of data sources used by RAIDS to capture injury details, including ambulance, hospital, radiology and post-mortem notes which enables the dataset to capture highly detailed traumatic brain injury information to varying levels of detail.

Injuries are coded via two key methods: firstly by free-text taken from ambulance, hospital, ICU, radiology and post-mortem reports, and secondly using Abbreviated Injury Scale (AIS) codes. In some cases, longer-term outcomes are available in questionnaires. From a biomechanical perspective, the causation of each distinct injury is captured, making this an extremely rich dataset for injury biomechanics studies. While there is no distinct TBI field, the combination of available free-text and AIS coded injuries makes it possible to identify TBI in RAIDS and explore its prevalence and causation. A high-level dashboard summary of RAIDS is available publicly.

Introducing the RAIDS portal

The following video shows an intro to the RAIDS portal and the tabs that are relevant for this section.” to “The following video introduces the 2025 RAIDS portal, providing a walkthrough of tabs that are relevant for navigating the RAIDS database online.

Further details about how to get free access to RAIDS can be provided by contacting RAIDS@dft.gov.uk.

Important considerations for ways to identify TBI in RAIDS

Because RAIDS has a wealth of clinical data including from ambulance, hospital, post-mortem and radiology findings, captured directly in fields and free-text, it is a valuable source for investigating TBI. Despite this, it is not primarily designed for TBI research. The following sections walk through how to identify TBI cases in RAIDS using several approaches.

Introducing the Abbreviated Injury Scale (AIS)

The Abbreviated Injury Scale (AIS) codes injuries according to anatomy and severity. It was created by the Association for the Advancement of Automotive Medicine with road traffic collision injuries in mind. It comprises a 6-figure number with one decimal place (format ABCDEF.G) capturing the type, location and severity. Each number positionally signifies: A – body region, B – type of anatomical structure, C,D – specific anatomical structure, E,F – level and G – severity of score (ordinal from 1-6, plus 9 – unknown).

Identifying specific types of intracranial injury with Abbreviated Injury Scale coding

For investigating specific TBI injuries of any severity, the AIS summary fields can be used. These include three fields which appear in 2005 and 2015 AIS coding (6 total):

Occupant summary – MAIS in body region head 05 / 15
Occupant summary – MAIS in body region head, body part cranium & brain 05 / 15
Occupant summary – MAIS in ISS region – head and neck 05 / 15

The maximum injury severity is useful for identifying a head-injured cohort. In order to investigate AIS coded head injuries, a filter can be applied when extracting data from the RAIDS platform to any of the fields capturing AIS head injury.

For example, a filter on the maximum AIS (MAIS) in the body region head can be applied, as is shown in the video. This example is also presented below.

Selecting head injuries to the body part cranium and brain with severity ≥1 is discussed in the video and shown below:

Using this approach, it is possible to select all subjects with AIS-coded head/brain injuries.

Using AIS-based RAIDS fields to identify specific TBI

It is possible to identify specific types of TBI from their distinct ABCDEF.G AIS code, or from groups of AIS codes. The following video shows an example of a single AIS injury code for a type of diffuse axonal injury described and practically how the code can be used as a filter within a query.

If specific injury types are of interest, a filter can be applied to select e.g. only a certain type of diffuse axonal injury (see video) or only subdural haematoma (see written description below). In this case, we can first identify the relevant AIS codes using copies of the AIS 2005 and AIS 2015 manuals (2005 is provided freely by SWOV). For AIS 2005, these are:

The number after the decimal point indicates the severity which is separately coded in RAIDS. We are therefore interested in 140650, 140651, 140652, 140654, 140656 and 140655. Coders will capture injuries to the maximum level of detail possible, where no further details are used they will code 140650 meaning “subdural NFS (nothing further said)”. We can therefore create a query in the Case Finder.

In this example, we create a filter on the AIS 2005 injury codes to inclusively capture all codes from 140650 to 140655 by identifying the relevant field and threshold for the filter condition:

On selecting save, you reach the finalised query page in the Case Finder:

When selecting “Send full results to case finder”, a list of all cases which are compatible with your filter criteria will be provided. For more information on creating queries, please contact the RAIDS team on raids_support@trl.co.uk.

Important considerations for

Note that AIS 2005 is predominantly used for early RAIDS cases and AIS 2015 for later RAIDS cases. Note that as RAIDS has been collected since 2013, some injuries are coded using AIS 2005 and others are coded using AIS 2015. If you wish to utilise all subjects, both AIS listed fields must be combined. For AIS coding this must be done by 2 separate searches, i.e. one that has the AIS 2005 filter and a second that has AIS 2015 filter. If you have requested download access, it can be easier to merge AIS 2005 and 2015 injury lists using your preferred data processing software or coding language for easier processing. In addition to the six-digit numerical AIS-codes, RAIDS provides free-text as part of the listed AIS injury. This can be used either to explore free-text detail about an injury once a cohort is identified, or directly as outlined in the next section.

When using AIS, it is important to note that there were some changes to AIS coding made during this time including the numbers of codes available for “concussion”, which must be considered in the context of your research question (Loftis et al., 2018). Baker et al. (2022) and Baker (2023) discusses in detail the benefits and drawbacks of using AIS coding.

Available research tools for investigating TBI in RAIDS

If you wish to take a more tailored or advanced approach to analysing RAIDS, it is possible to use the extensive free-text about injuries sustained available in RAIDS without being limited to specific RAIDS fields. This type of analysis is possible when working with downloaded RAIDS data, for which researchers need special permission to access from the Department for Transport when placing the data access request. Note also that special permission must also be requested for variables including sex and exact age if it is relevant for the research being conducted.

The graphic below demonstrates how the free-text search tool utilised by Baker et al. (2022) is applied to RAIDS. This automates a large proportion of the process of identifying TBI according to Mayo classification. This tool is currently being made open source as part of the TBI-REPORTER grant and will be available open access via the TBI-REPORTER platform soon.

An example of TBI Research in RAIDS

National estimates were calculated for TBI from RAIDS using a scaling developed with the police-reported STATS19 dataset which is not otherwise able to investigate TBI as it only captures overall injury outcome (Slight/Serious/Fatal), as detailed in Baker et al. (2022).

In addition to detailed clinical information, data about the collision dynamics such as change in speed are available. Using this data, it was possible to show the relationship between the change-in-speed (“delta-V”) and risk of TBI of different severities and pathologies. The example relationships are shown below. This is of value as the “delta-V” can be recorded by vehicle sensors in real-time, demonstrating the feasibility of using vehicle sensors for TBI prediction and supporting the post-collision pre-hospital response in future.

Baker et al. (2022) classified TBI in RAIDS using the Mayo severity scale with a designated free-text search tool, which they manually validated and checked against AIS-coded injury information demonstrating high accuracy and capture. A range of key pathologies were identified for different road users. More details can be found in the associated PhD thesis: AutoTriage: A feasibility assessment of using in-vehicle sensors to predict traumatic brain injury (Baker, 2023).

Figure 3 A graphical example of the method employed by Baker et al. (2022) to identify TBI pathologies and classify Mayo TBI severity in the RAIDS data (left) leading to the following road user TBI distributions (right).

Figure 4 The level of collision detail together with the clinical detail enables RAIDS to be used to understand how the biomechanics of the collision influences the TBI outcomes. Baker et al. (2022) were able to demonstrate a relationship between the change in speed (“delta-V”) during the crash and the risk of TBI.

RAIDS has a wealth of clinical data including from ambulance, hospital, post-mortem and radiology findings that often includes detailed TBI information, both directly available through fields and captured in the free-text. Baker et al. (2022)’s detailed RAIDS TBI analysis investigates how TBI relates to collision dynamics (change-in-speed known as delta-V) for different road users. The first stage of this work was to identify all TBI cases in RAIDS. This was done by investigating AIS-coded injuries captured directly by RAIDS fields, head injury flag fields in addition to utilising the extra information available via free-text.

Examples of other work in RAIDS related to TBI include investigations into pedestrian/car collision reconstructions, the interaction between TBI and head impact locations in motorcyclists and how demographics interacts with seating position.

Publications

TBI data included in RAIDS has been used in several studies. Below is a list of open-access publications from these studies. This list will be updated as new studies are published.

The relationship between road traffic collision dynamics and traumatic brain injury pathology, CE Baker, P Martin, MH Wilson, M Ghajari, DJ Sharp, Brain Communications 4 (2), fcac033
Inherent uncertainty in pedestrian collision reconstruction: How evidence variability affects head kinematics and injury prediction, CE Baker, P Martin, A Montemeglio, R Li, M Wilson, DJ Sharp, M Ghajari, Accident Analysis & Prevention 208, 107726
How do demographic factors, non-standard and out-of-position seating affect vehicle occupant injury outcomes in road traffic collisions?, CE Baker, M Ghajari, Safety Science 187, 106834