Official Statistics

Contributory factors in fatal collisions: comparing STATS19 with post-investigation recording, 2021 data

Published 29 November 2023

This report compares contributory factors (CFs) in fatal road collisions from STATS19, based on the initial assessment of the police attending the scene, with those subsequently recorded following more detailed investigation. This provides some quantification of the extent to which the factors in STATS19 might differ if based on information available within a longer time of the collision. We are grateful to the National Police Chiefs Council (NPCC) and all police forces that provided data to enable this work to be completed.

Main findings

This work presents an initial attempt to show how the contributory factors assigned in fatal road collisions might differ when they are assigned after more information is available to reporting police officers than is typically the case in the STATS19 system.

This report suggests:

• over a third of fatal collisions had a speed-related factor (exceeding the speed limit or travelling too fast for conditions) assigned after further investigation, compared with around a quarter in STATS19, making speed, when considered in this way, the most prevalent factor in fatal collisions

• some factors, particularly those related to speed and impairment by drink or drugs appear more frequently when based on information available after further investigation, meaning that STATS19 data potentially understates the impact of these factors

• overall, at national level, the main contributory factors in fatal collisions are broadly similar whether based on those captured in STATS19 or updated with further information available a longer time after the collision

• the overall change between CFs recorded initially (STATS19) and following further information masks larger changes at record level, where over a third of factors originally coded are removed or changed

• there appears to be some variation between police forces in the recording of contributory factors, which could be worthy of further exploration with forces where resources allow

These findings are broadly in line with related previous work, for example studies using in-depth collision investigation data. As noted, this work is an initial study based on one year of data (2021), which was partly impacted by lockdowns associated with the COVID-19 pandemic, and further work could be done to build on this depending on user interest.

Background

Contributory factors in STATS19

Annual statistics on road collisions in Great Britain are based on the STATS19 reporting system. Police forces make regular data returns to the Department for Transport (DfT), with final statistics published in September following the year to which the data relates after receipt and validation of data from all forces.

STATS19 does not record causes of collisions, but does include information on contributory factors (CFs) with published statistics based on these. Reporting police officers are able to select up to 6 factors from a list which in their opinion may have contributed to the collision.

These factors are typically based on officer judgement usually at the roadside or within a short time of the collision, rather than detailed forensic investigation, which means that although CFs provide useful information they have notable limitations as some factors may be under-reported by officers based on the information available at the scene or at the time of reporting. Further information on the strengths and weaknesses of CFs are outlined in our guide to contributory factors. Despite the known limitations, the most recent STATS19 review concluded that the current approach to recording should be maintained, with some modifications (for example, reducing the number of factors to make it easier for police to record the most relevant ones).

However, to better understand factors involved in fatal collisions, this report compares CFs as originally recorded in STATS19 with those reported post-investigation. This illustrates which factors might be understated in STATS19, and to what extent.

Approach

This analysis is based on comparing the CFs captured for fatal collisions in STATS19 (extracted from the STATS19 database) with equivalent information provided by police forces via a bespoke spreadsheet

To capture factors which would have been recorded post-investigation, a spreadsheet was sent to all police forces in Great Britain by the NPCC asking for them to complete up to 6 factors that were involved, from the same list as used in STATS19. To minimise the burden of this additional request, only the names of the factors were captured for the collision level - that is, there was no attempt to link to participants on the collision as in STATS19.

This information was requested between August and November 2023, for collisions occurring in 2021, meaning that at least 18 months had passed from the time of the collision. Therefore there was more information on which to base the choice of factors available than is typically the case than when they are coded in STATS19, as all fatal collisions receive an investigation, considering all forms of evidence including forensic results. In this report the information gathered this way will be referred to as ‘post-investigation’ factors.

By the nature of what is being requested, this post-investigation recording will inevitably be less timely than what is initially recorded in STATS19, but it is likely to more accurately capture the actual factors contributing to collisions as it is based on facts that are established during the investigation (and, for example, some factors will only become clear after further information is available).

However, it should be kept in mind that the way in which this information is captured (via a central spreadsheet at the collision level, rather than assigning factors to individual participants as in STATS19) may also have an effect on the results.

For this analysis, the STATS19 factors and ‘post investigation’ factors were linked at the collision level using the STATS19 collision reference. Any collisions where a police officer did not attend the scene and record at least one CF in STATS19 have been disregarded. Because responses were not received for all STATS19 collisions, the STATS19 figures presented below will not exactly match those in the published road casualty statistics on contributory factors (though should be similar).

Overall results

Responses received

Post-investigation factors were received from 38 police forces, covering 1053 collisions (out of a total of 1,271 fatal collisions in 2021 within scope - that is, where an officer attended and recorded at least one factor). This means that the following analysis is based on the majority of data available, though not all forces responded.

Overall, the number of contributory factors assigned per collision was similar in both the STATS19 and post-investigation datasets - 2.3 compared with 2.6.

Main contributory factors recorded

Table 1 shows the top 10 factors most frequently reported following further investigation (as a proportion of collisions), compared to STATS19. This shows that, in broad terms, the main factors are similar in STATS19 as those coded following a longer time after the collisions; of the top 10 factors post-investigation, only one (driver or rider impaired by drugs) does not appear in the top 10 when based on STATS19.

However, as noted below, the prevalence of some factors does change notably, and if grouped together, the two speed related factors (‘exceeding the speed limit’ and ‘travelling too fast for conditions’) would be recorded in 35% of fatal collisions, more than any other single factor when based on the post-investigation data.

Table 1: Top 10 most commonly recorded contributory factors based on post-investigation recording, compared with STATS19, 2021 fatal collisions

Contributory factor	Post-investigation percentage	STATS19 percentage
Loss of control	24.7	24.7
Exceeding speed limit	24.4	18.7
Driver or rider failed to look properly	21.7	26.6
Driver or rider careless, reckless or in a hurry	17.4	17.5
Travelling too fast for conditions	14.1	9.8
Driver or rider impaired by alcohol	14.0	11.2
Poor turn or manoeuvre	12.3	10.3
Driver or rider failed to judge other person`s path or speed	11.2	12.7
Driver or rider impaired by drugs (illicit or medicinal)	10.1	5.8
Driver or rider illness or disability, mental or physical	8.4	6.7

In general, the correlation in the ranking of factors recorded in STATS19 and post-investigation is strong, with a correlation coefficient of over 0.9.

Biggest changes in factors recorded

Although the broad ranking of factors is similar, the extent to which some are captured is notably different. Tables 2 and 3 show respectively the 5 factors with the biggest increase and biggest decrease in recording post-investigation, compared with STATS19.

This shows that overall, based on this data, speed and impairment related factors are those which are most likely to be coded once further information is available, compared to what is known within a short time of the collision when the STATS19 record is completed. This is not surprising; for example, the presence of drugs may not become apparent until test results are known.

These charts focus on those factors with the biggest percentage point changes; there are other less commonly reported factors which are also code more frequently in the post-investigation dataset (as shown in the accompanying data tables). These include factors related to vehicle defects, which again may not be easily identified at the time STATS19 is completed.

Table 2: Top 5 most factors showing the biggest increase in reporting post-investigation compared with STATS19, 2021 fatal collisions

Contributory factor	Post-investigation percentage	STATS19 percentage
Exceeding speed limit	24.4	18.7
Driver or rider impaired by drugs (illicit or medicinal)	10.1	5.8
Travelling too fast for conditions	14.1	9.8
Driver or rider impaired by alcohol	14.0	11.2
Poor turn or manoeuvre	12.3	10.3

Conversely, the factors which are more commonly reported in the original STATS19 data tend to be those which are related to driver or rider error or reaction, including ‘failed to look properly’ and ‘failed to judge path or speed’. These factors may be easier for the initial reporting officers to assign without a detailed investigation.

Table 3: Top 5 most factors showing the biggest decrease in reporting post-investigation compared with STATS19, 2021 fatal collisions

Contributory factor	Post-investigation percentage	STATS19 percentage
Driver or rider failed to look properly	21.7	26.6
Driver or rider failed to judge other person`s path or speed	11.2	12.7
Distraction in vehicle	3.9	4.9
Slippery road (due to weather)	4.0	5.0
Other	5.2	6.1

Impact on the ‘fatal 4’ factors

The ‘fatal 4’ factors in road collisions are those which police identify as the main factors associated with road deaths. They are typically considered to comprise speed, impairment by drink or drugs, distraction (including mobile phone use) and seatbelt non-wearing.

While seatbelt wearing is captured in STATS19, it is not recorded as a contributory factor. However, the other elements of the fatal 4 are coded as CFs. The proportion of collisions with these factors will depend on which factors are included, but typically the following are considered:

• Speed - exceeding the speed limit (CF 306) or travelling too fast for conditions (307)
• Drink or drugs - driver or rider impaired by alcohol (501), and driver or rider impaired by drugs (502)
• Distraction - driver using mobile phone (508)

Based on what is captured in STATS19, speed is reported as a contributory factor in 26% of fatal collisions, compared to 35% in the data based on information available after further investigation. As noted above, this would make speed considered in this way more prevalent than any other single factor in fatal collisions.

Similarly, the proportion of collisions with a driver or rider impaired by either drink or drugs increases from 15% to 20%.

Driver mobile phone use is much less commonly assigned as a factor in fatal collisions but again shows an increase in recording when based on post-investigation recording, appearing in 1% of collisions in STATS19 and 2% when based on further investigation.

Overall, for all the factors listed above, the proportion of fatal collisions with at least one of these factors increases from 35% to 45%. These proportions would be higher if other factors were considered within scope - for example, if ‘careless, reckless or in a hurry’ or ‘aggressive driving’ are included as speed-related. There is some evidence to suggest that these factors may sometimes be used as a substitute for speed factors.

Collision level results

The above section considers the change in the factors assigned across all fatal collisions in the dataset. This masks greater changes at the individual collision level. We can look at the proportion of factors assigned within STATS19 that were changed, and the proportion of factors assigned post-investigation that were added (that is, not originally coded in STATS19) as measures of the degree of churn in the factors allocated for individual collisions.

Variations by factor

Full details covering all factors are available in the accompanying data tables, with a summary for selected factors in table 4 below.

As an illustration, the number of collisions with ‘exceeding the speed limit’ increases from 199 to 257, there are only 160 collisions where this factor is recorded in both datasets, with 39 collisions where it was removed, and 97 where added. Overall, therefore, in 20% of the collisions with exceeding the speed limit originally recorded in was changed. Of the times where this factor was recorded post-investigation, in 38% of these it was not originally assigned.

Across all the factors together, in 36% of cases the originally assigned factor was not retained, with 41% of those finally recorded having been added to what was initially assigned in STATS19. This illustrates the extent to which the factors captured in STATS19 may be subject to change as further information becomes available.

Factors with a relatively high proportion of collisions added after not being originally assigned in STATS19 include ‘travelling too fast for conditions’ (54.1% of post-investigation collisions), ‘driver or rider impaired by drugs’ (60.4%) and ‘mobile phone use’ (70.8%, though based on a small number of collisions with this factor).

Conversely, we can look at factors with the biggest proportion of those recorded in STATS19 subsequently removed. For some factors, around half of those originally coded in STATS19 were removed in the post-investigation data, including ‘careless, reckless or in a hurry’ (48.4%) or ‘aggressive driving’ (54.9%) as many as half of the collisions with these factors no longer had it assigned following review based on further information.

Table 4: Selected factors recorded in STATS19 and post-investigation only, with extent of change in recording: 2021 fatal collisions

Contributory factor	In STATS19 and post-investigation	In STATS19 only	Post-investigation only	% in STATS19 that are removed	% in post-investigation that are added
All factors	1,600	880	1,105	35.5	40.9
Exceeding speed limit	160	39	97	19.6	37.7
Travelling too fast for conditions	68	37	80	35.2	54.1
Driver or rider impaired by alcohol	102	16	45	13.6	30.6
Driver or rider impaired by drugs (illicit or medicinal)	42	19	64	31.1	60.4
Driver using mobile phone	7	7	17	50.0	70.8
Aggressive driving	32	39	34	54.9	51.5
Driver or rider careless, reckless or in a hurry	95	89	88	48.4	48.1

Variations by police force

In the same way, it is possible to explore the extent to which the coding of factors in STATS19 and based on further information varies by reporting police force, which may reflect differences in operational practices and the systems they use for collision recording.

For instance, for some forces, most notably the Metropolitan police, the data captured based on further investigation shows very little variation from what is recorded in STATS19. It has become apparent as a result of this work that this is because the data for the Met recorded in STATS19 may already be updated with information available after the collision.

Conversely, for some forces the extent to which the factors have changed is large, with more than half of factors coded in STATS19 being removed or changed.

This work therefore suggests that there may be inconsistencies in the way that contributory factors are recorded across different forces, which may be worthy of further investigation.

Implications and next steps

This work provides an attempt to quantify the extent to which the contributory factors recorded in STATS19 - which are by design intended to be based on information available within a short time of the collision - might change if based on information available following a more thorough collision investigation.

The results illustrate the extent to which the proportion of collisions with speed or impairment factors in particular are likely to be understated based on the way the information is captured in STATS19, both absolutely and relative to other factors. This means that conclusions drawn from STATS19 data for these factors in particular should be interpreted with care, keeping in mind the way in which the data is collected.

While STATS19 remains a valuable source of data on road collisions, other data sources, for example detailed collision investigation data, help to provide further insight for fatal collisions in particular. Possible follow up to this initial study could include more in-depth analysis of the data presented here alongside analysis of a wider range of available data, to provide a more complete picture of the causes of fatal road collisions.

While the most recent STATS19 review recommended no change to the fundamental way in which contributory factors are captured within the STATS19 system, the CFs themselves will be replaced with new ‘road safety factors’. This work suggests some variation in how different forces code contributory factors within STATS19. Subject to resources, there could therefore be scope for action to attempt to improve the consistency of reporting, as the new road safety factors are introduced.

Subject to continued co-operation from reporting forces, this work could form phase 1 of a longer programme of work led by NPCC to capture CFs post-investigation, which could involve:

• workshops with forces to better understand how contributory factors are assigned in collisions, and explore how consistent this is done across forces
• revisiting a collection of this form for 2022 data, to see whether the same patterns are seen
• considering whether there would be sufficient support for an ongoing annual return of this data, on a voluntary basis, to allow post-investigation factors to be routinely presented alongside those from STATS19 in the department’s statistics

However, while this work has illustrated some potential limitations of STATS19 contributory factors, their strengths and value should also be borne in mind, in particular their availability for all severities of collisions, in a relatively timely way, in sufficient volumes to enable both national and more detailed analyses.

Any feedback on this work or suggestions for further analysis are welcome.

Background notes

Supporting data

This report provides a high level summary of the information collected and resulting analysis. Further details, including data for each of the 77 contributory factors recorded in STATS19, is available in the accompanying data tables.

Responses received

This report is based on responses from the following 38 police forces: Avon and Somerset, Cambridgeshire, Bedfordshire, Hertfordshire, Metropolitan Police, City of London, Derbyshire, Devon and Cornwall, Dyfed-Powys, Lincolnshire, Northamptonshire, Nottinghamshire, North Yorkshire, Wiltshire, West Mercia, Norfolk, Suffolk, Cumbria, Durham, Essex, Humberside, Kent, Lancashire, West Midlands, North Wales, Surrey, Leicestershire, Gloucestershire, West Yorkshire, Dorset, Sussex, Greater Manchester, Hampshire, Merseyside, South Wales, Cheshire, Staffordshire, Thames Valley. The Department is grateful to all those supplying this additional information in addition to the regular STATS19 dataset.

Related information

Previous work has illustrated the limitations of contributory factor recording in STATS19. In particular, in-depth collision investigation studies such as RAIDS collect very detailed information on road collisions based on in-depth investigations and follow up, and allow comparisons to be made with what is recorded in STATS19

Instructions for printing and saving

Depending on which browser you use and the type of device you use (such as a mobile or laptop) these instructions may vary.

Tablets and mobile devices normally have the option to “find in text” and “print or save” in their sharing or quick options menu of their browser, but this will vary by device model.

How to search

Select Ctrl and F on a Windows laptop or Command and F on a Mac

This will open a search box in the top right-hand corner of the page. Type the word you are looking for in the search bar and press enter.

Your browser will highlight the word, usually in yellow, wherever it appears on the page. Press enter to move to the next place it appears.

Contact details

Road safety statistics

Email roadacc.stats@dft.gov.uk

Public enquiries 020 7944 6595

Media enquiries 0300 7777 878