Requesting and reading collision data for an access

When a highway authority objects to a new access on safety grounds, the evidence that answers it is recorded personal injury collision data for a defined study area over a defined period. Ask the authority for it in writing, specify the area and the period, and expect to pay a small handling charge. The published national road safety datasets will give you a cross check, but the authority's own extract is what its officers will accept.

Five full calendar years is the normal period. Anything shorter invites the argument that the sample is too small; anything longer brings in network conditions that no longer exist.

Where the data comes from

Collision records originate with the police, recorded at the scene or shortly after on the national reporting form, and are passed to the local highway authority and to the Department for Transport. Only collisions involving personal injury are recorded. Damage-only collisions are not in the data at all.

Three routes to a copy:

1. A written request to the local highway authority's road safety or traffic team, specifying the study area by plan and the period by calendar years. This produces an extract with location, date, severity, number of casualties, vehicle types, manoeuvres and contributory factors.
2. The published national road safety statistics, which give the same underlying records at a coarser level and are useful for a sanity check or for a quick early assessment before the formal request lands.
3. The authority's own published road safety plan or local transport plan, which may identify cluster sites and priority routes, and read it, because it tells you what the authority is already worried about.

Make the request early. Some authorities turn an extract around in a week, others take a month, and a transport assessment written without the data is an assessment that will be rewritten.

Defining the study area and period

The study area is an analytical choice and it will be challenged, so make it on principle rather than on convenience.

For a single new or altered access, the usual approach covers the site frontage plus the nearest junctions in each direction, together with any junction where the development's traffic would form a meaningful proportion of a turning movement. A radius of 100 to 200 metres from the access, extended along the frontage, covers most minor schemes. For a larger scheme the area follows the assessed network in the Transport Assessment, so that the safety analysis and the capacity analysis cover the same junctions.

Two mistakes to avoid. Drawing the area so tightly that a known cluster 150 metres away falls outside it looks like concealment and will be treated as such. Drawing it so widely that the analysis dilutes a genuine local problem in a mass of unrelated records looks like the same thing from the other direction.

State the area, the period and the source in one sentence at the start of the analysis, with a plan. Then stick to it.

Reading severity and contributory factors

Severity is recorded as fatal, serious or slight, and there is a data caveat that every practitioner should know. Several police forces changed the way they record severity when injury-based reporting systems were introduced during the middle of the last decade, and that change increased the proportion of collisions classified as serious without any change on the road. Comparing a five year period that spans the change, or comparing one authority with another across it, produces misleading trends. Say in the report which years are affected if the authority's data covers the transition.

Contributory factors are the police officer's judgement at the scene, recorded as a set of coded factors with a confidence level. They are indicative rather than evidential. The ones that matter for an access argument are those pointing at the road layout, such as poor or obscured visibility, junction overshoot, slippery road surface or inadequate signing, as opposed to those pointing at behaviour, such as impairment, excessive speed or failure to look properly.

That distinction does most of the work in the written argument. A cluster of collisions where the recorded factors are driver impairment and failure to look tells you little about whether a new access is safe. A cluster where visibility is repeatedly recorded tells you a great deal.

Clusters against random distribution

Collisions are rare events and they arrive unevenly. Three collisions at a junction in five years may be a pattern or may be chance, and the honest analyst says which.

The test is whether the collisions share a mechanism. Look for common direction of travel, common manoeuvre, common time of day, common weather or light condition, and common casualty type. Three right-turn collisions by vehicles emerging from the same side road in daylight is a pattern. Three collisions involving a pedestrian at night, a rear-end shunt in queuing traffic and a motorcyclist losing control in the wet is not.

Rate matters alongside count. A junction carrying 20,000 vehicles a day with four collisions in five years is performing differently from a junction carrying 2,000 with the same four. Where traffic counts exist, express the figures as a rate and say what the denominator is.

Where a pattern does exist, the question becomes whether the development makes it worse, better or neither. A scheme that adds turning movements to a junction with a turning-movement problem has work to do. A scheme that funds a crossing, improves visibility or removes a conflicting access can improve the position, and that improvement is normally delivered through the mechanism described in section 278 agreements and how they follow a permission.

What the data cannot show

Being straight about the limits strengthens rather than weakens the argument.

• Damage-only collisions are absent, and they are the majority, so an access with frequent minor shunts can show a clean record
• Near misses are absent entirely
• Collisions involving cyclists and pedestrians are under-reported relative to those involving motor vehicles, particularly where no motor vehicle was involved
• Location coordinates are recorded to the nearest point and can be a junction or two out on long links
• The data describes the network as it was, so a layout changed three years ago makes the earlier records less relevant
• Absence of recorded collisions is not evidence of safety, only of an absence of recorded injury

The last point cuts both ways and is worth stating plainly. A clean five year record at a field access with two vehicle movements a day proves nothing about the same access serving forty houses.

Using it to answer a safety objection

National policy sets a high bar for refusal on highway grounds: development should only be prevented or refused on highways grounds where there would be an unacceptable impact on highway safety, or the residual cumulative impacts on the road network would be severe. That wording is what the collision analysis is being measured against, and it is a higher threshold than a highway officer's preference for a different arrangement.

A written response that works has four moves.

First, establish the baseline: the recorded record over five years, with no pattern attributable to the existing access arrangement, or with a pattern identified and described. Second, establish the geometry: visibility splays achieved in both directions against the appropriate stopping sight distance for the measured speed, with speed survey data rather than the posted limit. Third, establish the change: the additional movements the development generates, by turning movement and by time period. Fourth, offer the mitigation, if any is needed, with a delivery mechanism.

Where a national road is involved the objector is a different body with a different test, and the arguments that move it are described in when National Highways objects to an application.

Presenting it in an assessment

Make the data readable in two minutes.

Put the full extract in an appendix. Officers want the table and the plan in the body of the report; a reviewer or an objector will want the raw records, and withholding them invites a request that delays determination.

Where the scheme is small enough that only a Transport Statement is required, the same structure applies at a shorter length. Two pages with a plan, a table and a conclusion is a complete answer for a single house on an existing access.

Enter your postcode on PlanScope to see which transport documents a new access at your site is likely to need.

PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval. Councils usually expect highway safety analysis to be prepared by a suitably qualified transport planner, and professional review is strongly advised. This article is general guidance rather than legal advice.