Revisions to road casualty statistics
Updated 28 November 2024
Overview
Reported road casualty statistics are based on data reported by the police via the STATS19 data collection. Figures based on STATS19 are not routinely revised, with the exception of provisional data, or any figures relating to serious collisions or casualties based on severity adjustments.
Provisional figures are revised as they are finalised. For severity adjustments, separate guidance on the severity adjustments is available, but users should note that they will change with every publication including new data a as result of updates to the model used for the adjustment. Except where a police force has recently adopted injury-based reported, it is not expected that these changes will be large or that there will be notably change in trends for individual police force areas.
However, the road casualty statistics incorporate data from other sources which can be revised, resulting in updates to the previously published figures, in particular for casualty rates (which rely on traffic or population data)
This note provides details of any non-routine revisions to previously published statistics, by the year in which they were made. Further information is available from the road safety statistics team via the contact details below.
STATS19 data revision policy
The department’s long-established policy for STATS19 data is that once a year of data has been finalised (typically alongside the publication of Reported Road Casualties Great Britain at the end of September each year) it is not routinely revised. The data within STATS19 represents the best available data at the time the database was closed. In the case of fatalities, it can be the case that there are outstanding coroner verdicts, which, once received may change whether a record should or should not be within scope of STATS19.
This approach has strengths and weaknesses. It means that the historic time series of data does not change as more data is added (aside from the severity adjustments) avoiding the need to revise baselines for monitoring progress, for example. It also provides a focus for reporting police forces to provide final data to the department. It is likely that any changes following closure of the dataset will be small in number (for fatal or more serious collisions) and not affecting national level trends or conclusions drawn.
However, this approach does mean that some cases which subsequently prove to be within scope of STATS19 are excluded, and vice versa. This can result in small discrepancies between the department’s database and those held by local authorities, or individual countries within Great Britain, where later revisions may be made.
Changes to the department’s processing systems mean that it is now possible to consider making more routine revisions to earlier years of data. Consultation with users during 2023 revealed no strong preference, with some in favour of revisions and others preferring the current approach. Some users suggested that revisions for records involving fatalities may be worthwhile, but not for the less serious collisions.
Consequently, it remains the department’s policy not to make revisions to STATS19 data after the final annual dataset has been published each year. However, to help users to understand the scale of the issue, we have added a revisions log (available from the data quality page)where we intend to list cases which may have been included or excluded from the dataset were it possible to take account of information received after closing the database.
We welcome further views from users of the STATS19 data on this approach, which we will keep under review.
2020: Revisions to historic road casualty data
Alongside the publication of the 2020 reported road casualty figures in September 2021, historic figures for slight and serious injuries were revised from 2012. This section provides further details of this change.
Prior to 2020, whether a casualty was reported as admitted to hospital was factored into casualty severity calculations in the Collision Reporting and Sharing (CRASH) system. From 2020 this is no longer the case. This means that some casualties that would have been reported as ‘serious’ prior to 2020, will be reported as ‘slight’ in 2020 and in subsequent years.
To allow for a consistent comparison of casualty severities between years prior to 2020 and years following this, historic cases since the introduction of CRASH have been updated and corrected to reflect this change.
As a result, some reported casualty severities have been downgraded (from serious to slight) between 2012 and 2019 compared to numbers published in publications up to and including the 2019 annual report. These severity changes affect only casualties reported by police forces that have adopted CRASH. The correction to historic reported casualty severities is only required to be applied once.
Overall, the total number of police-reported ‘serious’ casualties was reduced, and the number of ‘slight’ casualties increased, with respect to previously published figures between 2012 and 2019 because of the corrections applied. However, the overall total number of injured casualties remains unchanged.
The total number of casualties affected is small, with the severity updated for less than 1,500 casualties over all years. The largest impact is for figures in 2019, which includes 572 affected casualty severities. Overall collision severities have also been updated, where appropriate and for the same years, to reflect the changes to casualty severities.
The revisions have been applied to the statistical publications published as part of the 2020 final annual statistics, and to the open data which is made available via data.gov.uk. It is anticipated that from this point on these figures will not be further revised.
2021: Minor historic revisions to casualty road user type
As a result of development of new data tables for the 2021 final annual statistics, a small number of inconsistencies came to light regarding the classification of casualty type in 2015 (where casualty type was not consistent with vehicle type, for vehicle occupant casualties). These inconsistences were corrected, affecting around 100 casualties, with no change in casualty totals.
2021: Minor revisions to contributory factor data
As part of the publication of the 2021 statistics, a small revision was made to data for contributory factors in collisions. This was to identify cases where data for a collision was updated by the reporting police force, but an incorrect version of the contributory factors was taken into the final database. Aside from a handful of records in years 2012 to 2014, this change affected only data for 2019.
Statistics published for contributory factors for 2019 have been revised as a result. While the factors assigned for some collisions have changed, at the aggregate level, the impact on the statistics published (for example in table RAS0701) is small.
2021: Revisions to traffic data used to calculate casualty rates
Following a revision in traffic estimates published by the department’s road traffic statistics team, casualty rates per billion vehicle miles have been revised for the period 2010 to 2020, as part of the publication of the 2021 final annual statistics in September 2022.
As traffic figures have been revised downwards, the impact on casualty rates is a small upwards revision, with the size of the revision greater in 2020 than other years. However, broad trends in casualty rates over time are not affected as a result of this revision.
The revised traffic figures are for minor roads only, so that any casualty rates by road type will show the greatest impact of the revision.
2022: Minor historic revisions for Greater Manchester police force
Following a change of reporting system used by Greater Manchester Police, a small number of records were submitted with incorrect date and time allocated to them when originally published in 2021. These records were updated alongside publication of the 2022 statistics. This change affected less than 35 records in the open dataset and data download tool, but has not impacted on the published statistical reports or tables. Full details of changes can be found in the revision table on the data.gov.uk page.
2022: Minor historic revisions to geographic administration areas and coordinates in Isles of Scilly
Following the continued development of the open data tables, a small number of geographic inconsistencies came to light regarding collisions in the Isles of Scilly for years between 1994 and 2009 which have been revised with the publication of the 2022 statistics. This change affected less than 50 records, and affects only the open dataset. Full details of changes can be found in the revision table on the data.gov.uk page.
2022: Historic population of generic make and model
In the 2002 update of the open data files, the historic back series has been populated with generic make and model for vehicles which have been matched with DVLA data. The generic_make_model field is populated where there are 10 or more instances of vehicle combinations of type, make and model. Vehicle combinations with less than 10 instances are treated as sensitive data, that can be requested from the road safety statistics team.
2023: Minor revisions to population and road traffic estimates used to calculate casualty rates
In producing 2023 casualty rates, figures for the denominators used in some historic years have been updated to reflect revisions to the traffic or population data used. For example, 2022 traffic figures were very slightly revised. In all cases, the impact of these revisions is considered very small and has no material impact on the published figures.