Using collision data in a transport assessment
Collision analysis is the part of a transport assessment read most closely by objectors. Classify by mechanism rather than by location, and state the limitations.
Collision analysis is the part of a transport assessment most likely to be read closely by someone hostile to the scheme, and the part most often written as a paragraph saying there were no clusters. That paragraph invites a parish council to produce its own map. If the data supports the scheme, show the working. If it does not, say what the pattern is and what the design does about it.
The underlying material is recorded personal injury collision data, collected by police at the scene under the national reporting system and held by the local highway authority.
What the data records, and what it leaves out
Only collisions involving personal injury that were reported to the police appear. Damage-only collisions are absent entirely, which removes the majority of events at any junction. Under-reporting is well documented and is worst for collisions involving cyclists and for single-vehicle incidents.
Each record carries a location, a date and time, the severity, the vehicles and casualties involved, and contributory factors recorded by the attending officer. Two cautions apply to that last field. Contributory factors are an officer's opinion formed at the scene rather than a finding after investigation, and they are recorded inconsistently between forces.
Location coordinates deserve their own check. A collision recorded at a junction node may have occurred on an approach 40 metres away, and a cluster that appears to sit on a proposed access can dissolve when the records are read individually. Read the records, not just the plot.
Request the data from the highway authority with a defined study area and period, and record the extraction date. Some authorities charge, some provide it freely, and the national road safety statistics collection offers an alternative route where an authority is slow.
Choosing the study area and the period
The study area should be the junctions and links where development traffic makes a material difference to flows, agreed with the highway authority at scoping. A commonly used convention is a percentage increase in flow above which a junction is included, with a lower figure applied in sensitive locations, and the practical point is to agree it in writing rather than to defend it later.
Five years is the usual period. Three is arguably defensible on a small scheme, and anything shorter reads as selection. Where a junction has been altered during the period, split the analysis at the date of the change, because collisions before a scheme was built say nothing about the layout in place now.
Include the site access location even where flows barely change, because that is the point the highway authority cares about most. Scale the wider study area to the scheme: the threshold logic that decides whether a Transport Assessment or a Transport Statement is required also indicates how far the collision study should reach, and trip generation and the assessment threshold is the right place to start that conversation.
Severity, and the reporting change that breaks trends
Collisions are classified as fatal, serious or slight. The classification is not stable over time.
From the mid 2010s, police forces progressively moved to injury-based severity reporting systems, which increased the proportion of collisions recorded as serious without any change on the ground. Any five year analysis spanning that transition will show an apparent worsening in severity that is an artefact of recording.
Say so in the report. An analysis that presents a rising serious collision count across the transition period, without the caveat, will be corrected by the highway authority and the rest of the analysis will be read with less trust.
Telling a pattern from noise
Small numbers are the recurring statistical problem. Three collisions at a junction in five years might be a pattern or might be three unrelated events, and the difference is not decided by the total.
Look for repetition in the mechanism rather than in the location. Three right-turn-out collisions at the same arm, at similar times of day, involving the same manoeuvre, is a pattern. Three collisions at the same junction involving a rear shunt, a pedestrian crossing at night and a motorcyclist losing control in the wet is a location with traffic on it.
Classify each record by manoeuvre, by casualty mode, by time of day, by lighting and surface conditions, and by direction of travel. Then say plainly whether a pattern exists. Where there is no pattern, the honest sentence is that the records do not show a recurring mechanism, not that the junction is safe.
Comparison against national average rates per vehicle kilometre can be useful on links, and is much weaker at individual junctions where the exposure denominator is hard to establish. Use it carefully or not at all.
Linking collisions to the proposed access
This is where the analysis earns its place, and it comes down to one question: does the development add exposure to a mechanism that is already producing collisions?
A scheme generating right turns out of a new access onto a road with an existing pattern of right-turn collisions has a problem, and the assessment should say so and address it with a design response: a ghost island, a left-in left-out arrangement, a relocated access, or a reduction in the movements concerned.
A scheme adding traffic to a junction where the recorded collisions involve pedestrians crossing at night has a different problem, and lighting, crossing provision and desire lines are the answer rather than junction capacity.
Where the access is onto the strategic network, the position is more formal still, and access to the strategic road network brings the national operator in as a consultee with its own requirements.
Presenting it fairly
A defensible presentation contains six elements.
- The source of the data, the study area, the period covered and the extraction date.
- A plan showing every record, with severity distinguished, over a current base map.
- A table of records with date, severity, casualty mode, manoeuvre and conditions.
- The classification analysis, showing what was grouped and how.
- A plain statement of whether a recurring mechanism exists, and where.
- The limitations, including under-reporting, the severity recording change, and coordinate accuracy.
Leave out the record count and you look selective. Leave out the limitations and you look unaware. Neither costs anything to include.
When a road safety audit is the right answer instead
Collision analysis is retrospective. It tells you what has happened on the existing layout, and it cannot assess a layout that does not yet exist.
A road safety audit does the opposite. An independent team reviews the proposed design against safety criteria at defined stages, from preliminary design through to completion and post-opening monitoring. Highway authorities routinely require a stage one audit for a new access, and it is a design assurance process rather than an evidence base.
Confusing the two produces two familiar failures. An applicant who submits an audit in place of collision analysis has not addressed whether there is an existing problem. An applicant who submits collision analysis in place of an audit has not shown that the new geometry works. Larger schemes need both, and the audit should follow the collision analysis so that the audit team knows what the existing record shows.
Browse all 66 PlanScope report types and start a draft to see which transport documents a scheme of a given scale normally needs.
This is general guidance, not legal advice; check the position with your local planning authority or a qualified professional before relying on it. PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval.
Frequently asked questions
Where does collision data for a transport assessment come from?
Recorded personal injury collision data collected by police at the scene under the national reporting system, held by the local highway authority. Request it with a defined study area and period and record the extraction date. The national road safety statistics collection is an alternative route where an authority is slow to respond or charges heavily.
What period should a collision analysis cover?
Five years is the usual period, and three is arguably defensible on a small scheme. Anything shorter reads as selection. Where a junction has been altered during the period, split the analysis at the date of the change, because collisions recorded under a previous layout say nothing about the arrangement in place today.
Why do serious collision numbers appear to rise over time?
From the mid 2010s, police forces progressively moved to injury-based severity reporting systems, which increased the proportion of collisions recorded as serious without any change on the ground. Any five year analysis spanning that transition shows an apparent worsening that is an artefact of recording, and the report should say so.
How do I tell a collision pattern from random events?
Look for repetition in the mechanism rather than the location. Three right-turn-out collisions at the same arm, at similar times, involving the same manoeuvre, is a pattern. Three unrelated events at one junction is a location with traffic on it. Classify every record by manoeuvre, casualty mode, time of day and conditions before concluding.
Is a road safety audit the same as collision analysis?
No. Collision analysis is retrospective and describes what has happened on the existing layout. A road safety audit is an independent review of a proposed design against safety criteria at defined stages. Larger schemes need both, and the audit should follow the collision analysis so the audit team knows what the existing record shows.
Sources and further reading
This article is general guidance about the planning system in England and, where stated, the rest of the UK. It is not legal or professional advice. Check your local planning authority’s validation requirements and take professional advice for your specific site.