Sunday, 24 July 2016

STATS19 collision data - Overview

The database of police collision reports comes from a standard form called STATS19. This collects a range of data on the collision, vehicles involved, casualties and injuries. A selection of this information is published annually as a database (details such as registration numbers are recorded but not published).

The data is published as three separate lists of information (in .csv format):
  • 'Accidents' (collisions) lists out the individual collisions with a reference number and data relating to the incident.
  • Vehicles lists out the vehicles involved, the accident reference, vehicle reference, plus data relating to the vehicle.
  • Casualties lists out the casualties, the accident reference and data relating to the casualty.
Each collision will have one or more vehicles and may have several casualties.

The STATS20 form gives details of what information is meant to be recorded and how the various categories work (for instance: what constitutes a minor injury, or how the vehicle codes should be used).

In this post I've summarised what is contained in the publicly available data, with a focus on information relevant to pedestrian and cyclist safety.


The list of collisions includes the reference number 'Accident Index', the location (easting & northing) and date & time in addition to:
  • The Police Force and Local Authority
  • Number of vehicles involved
  • Number of casualties 
  • Speed limit
  • Roundabout / One way street / Dual carriageway / Single carriageway / Slip Road / Unknown 
  • If the collision was near a junction/pedestrian crossing and if so the type of junction/crossing.
  • Lighting, weather and road conditions/hazards
  • Urban/Rural area


For each vehicle we are given the Accident Index and vehicle reference (001, 002, etc). If there is only one vehicle linked to a collision that would be 001. Details of the vehicle are:
  • The vehicle type (more info) and whether it was towing, left hand drive.
  • Vehicle manoeuvre, for instance: reversing, turning left, changing lane to right, overtaking on nearside etc.
  • Vehicle location: Main carriageway, Bus lane, marked Cycle Lane (advisory or mandatory), separate Cycle Way, Footway. Note that there is no code for shared space area so this information is not recorded.
  • Information on whether the vehicle is entering a junction, at a junction, leaving a junction, entering a roundabout etc. 
  • Whether the vehicle skidded/overturned, whether it left the carriageway and whether it hit an object.
  • Point of impact: 0. Did not impact, 1. Front, 2. Back, 3. Offside, or 4. Nearside 
  • Age and sex of driver.
  • Journey purpose of driver: 1. Journey as part of work 2. Commuting to/from work 3. Taking pupil to/from school 4. Pupil riding to/from school 5+. Other/Not known 


Similarly to vehicles, the records start with the Accident Index and then a casualty reference. The first casualty associated with the collision is 001, then 002, 003, etc. Details of the casualty are given as follows:
  • The 'casualty type' classification (more info)
  • Casualty class: 1. Driver or rider 2. Vehicle or pillion passenger 3. Pedestrian. Another field records whether they were a Bus/Coach passenger.
  • The severity of the injury. Injuries are recorded as serious/slight by the police usually within a short time of the accident and are often not based on the results of medical examinations.
    • 1, Fatal
    • 2, Serious: Serious injuries are generally those that would send the casualty to A&E or where they're likely to require further medical treatment. This includes: Broken neck or back, Severe head injury, Loss of arm or leg (or part), Fracture, Concussion, Deep cuts and Injuries to casualties who die 30 or more days after the accident from injuries sustained in that accident. 
    • 3, Slight: Slight injuries are generally those which require first aid or require no medical attention and include Whiplash, Shallow cuts, Sprains and Bruising.
  • Vehicle reference number - this is the number of the vehicle they occupied, or for pedestrians the number of the vehicle which first hit them. 
  • Age and sex of casualty.
  • Pedestrian location: footway/verge, refuge/central reservation or carriageway.


The data can be filtered using any of these categories and then mapped. One or two of the categories can be used to determine the map marker shape and colour. Other data can be displayed when the user clicks on a data-point.

For instance the data could be filtered to show only cyclists and pedestrians injured in collisions involving cars, with round markers for pedestrians and square markers for cyclists with the following colours: (fatal, serious, minor) = (red, yellow, orange).

Please let me know in the comments below if you have any suggestions for maps which might provide insights or illustrate particular problems.