STATS19 DATA LAB · GREAT BRITAIN

STATS19 Data Lab

Police-reported injury collisions in Great Britain · geographic history 2016–2025 · record analysis 2021–2025

Descriptive result available

What changes when we account for miles?

Derived rates describe reported injury outcomes, not insurance claim frequency.

Charts displayed; exact values remain available with each chart.

FREQUENCY · EXPOSURE BASIS

What changes when we account for miles?

Compare road exposure, vehicle involvement and serious-injury burden on the same scale. Every rate uses the counts and mileage in this release.

Selected evidence: Road category · 2025 · injury collisions per 100,000 motor vehicle miles. Analytical release lens-2026-10-09-cbec43fe76c4.

Road category · 2025 · injury collisions per 100,000 motor vehicle milesAnalytical release: lens-2026-10-09-cbec43fe76c4

Export, share & cite

Use this result

Readable results, definitions and provenance in separate CSV files, with the exact selection and citation.

Exact selected records and established audit fields. Full audit · all selections

The evidence link is shareable. It keeps the analytical release and selected filters. Browser review decisions and notes stay local; use Download decisions separately to share a proposed specification.

Strydom, J. STATS19 Data Lab — frequency per mile. Analytical release lens-2026-10-09-cbec43fe76c4. /data-lab/stats19?release=lens-2026-10-09-cbec43fe76c4&view=frequency&frequencyDimension=road&frequencyYear=2025&frequencyGroup=all&frequencyMetric=collisions&frequencyTrend=a&resultView=chart#explore. The link contains the evidence selection; browser review decisions are excluded.

Download recovery: Open analyst ZIP directly · Open audit CSV directly

Formats, units and reproducibility

Sources used by this result

Loading the selected result’s source record…

Full source-monitoring register

Selected comparison: Road category · 2025

Change comparison 2 applied filtersRoad category · 2025

ComparePeriodOutcome

Derived descriptive rate · 20250.0296injury collisions per 100,000 motor vehicle miles · pooled eligible groups

Official aggregate mileage342.55bnVehicle miles across eligible groups

Count coverage101,519 / 101,525Records with a usable rate · 6 unavailable

The denominator uses official annual mileage estimates. The calculated rate is a derived descriptive measure. The collision numerator includes non-motor-only collisions; the denominator is motor vehicle mileage.

Road category

injury collisions per 100,000 motor vehicle miles

Bar: rate. Whisker: conditional 95% interval where available. Vertical marker: pooled eligible baseline. Select a category to follow its annual trend.

Relative rates use the pooled rate of displayed eligible groups. Grey / unavailable cells retain their counts. Low count means fewer than 20 observed records; pooling years can improve credibility.

Exact values · also in Data (4 rows)

injury collisions per 100,000 motor vehicle miles. Rates use unrounded inputs; intervals are conditional and exclude mileage uncertainty.
Category	Numerator	Matched miles	Per 100,000 miles	95% interval	Relative rate	Exposure method	Credibility
A roads	43,827	151,591,800,000	0.0289	0.0286–0.0292	0.98×	Official aggregate mileage	Available
Minor roads	54,440	119,023,200,000	0.0457	0.0454–0.0461	1.54×	Official aggregate mileage	Available
Motorways / A(M)	3,252	71,935,900,000	0.0045	0.0044–0.0047	0.15×	Official aggregate mileage	Available
Unknown road class	6	Unavailable	Unavailable	Unavailable	Unavailable	Exposure unavailable	Low count

Intervals use an approximate conditional 95% Poisson model; the zero-count upper limit is exact.

Derived descriptive rate

Numerator / population: Reported injury collisions · All police-reported injury collisions, including non-motor-only events
Denominator / unit: Motor vehicle miles in the selected road, region or time group · Reported injury collisions per 100,000 miles
Scope: Great Britain · 2025
Exposure basis: Official aggregate mileage
Coverage: 101519 of 101525 records have usable exposure; 6 excluded. Unknown exposure is not treated as zero.

Measurement details · outcome and adjustment

Outcome: Reported injury collisions

Adjustment: Descriptive; no risk-factor adjustment

Interpretation limit: Includes non-motor-only injury collisions over motor mileage: a contextual road-exposure rate, not matched motor-involved collision frequency.

Evidence details · Annual trends, compatibility and reconciliation

Does the rate persist over time?

Category to follow

injury collisions per 100,000 motor vehicle miles

A roads

Select a point, or use the Data tab for exact values. Missing observations remain gaps.

Exact values · also in Data (5 rows)

injury collisions per 100,000 motor vehicle miles
Year	A roads	A roads · conditional 95% interval
2021	0.03473	0.03441–0.03505
2022	0.03306	0.03276–0.03336
2023	0.03151	0.03123–0.0318
2024	0.02968	0.0294–0.02996
2025	0.02891	0.02864–0.02918

2021 was affected by pandemic travel restrictions. Monthly trends also reflect changing trailing five-year profile windows.

Exposure compatibility register

Only compatible numerators and denominators produce a rate. A modelled entry states its allocation assumption.

Exact values · also in Data (10 rows)

Dataset and exposure compatibility
Analysis	Status	Exposure	Limits
Annual road category	Publishable	GB annual motor miles, matching road category	Includes non-motor-only injury collisions.
Annual vehicle class	Publishable	Same-class annual GB vehicle miles	Cars/taxis, motorcycles, known-weight LGV/HGV only; involvement is not fault.
Annual region and road × region	Publishable	Published joint regional road-category motor miles	Extension must be in the approved release; unknown highway/region mappings retained.
Annual vehicle × road / vehicle × region	Publishable	TRA0104 joint road/vehicle estimates; regional vehicle CSV	Published joint strata only. TRA0104 rounding retained; no synthetic three-way cross.
Month	Modelled	TRA0305 trailing five-year profile × calendar days × annual GB miles	Smooths yearly variation, especially the pandemic. All roads and all motor vehicles only.
Day of week	Modelled	Same-year TRA0306 flow index × weekday counts	All roads, GB. Does not measure local or monthly weekday mileage.
Hour / time band / hour × weekday	Modelled	Joint TRA0307 hour/weekday index × local calendar hours	All roads, GB. No road, region or month filter.
Time × vehicle class	Awaiting exposure	TRA0308 reviewed for a later vehicle/time extension	Selected classes only; not included in this release’s validated selectors.
Road × hour / region × time / month × hour	Awaiting exposure	Requires joint observed traffic or an explicitly validated exposure model	Separate marginal profiles are never multiplied.
Age / sex / fuel / ADAS / weather	Awaiting exposure	Matched driver, fleet or condition-specific vehicle miles	Counts, licensed stock and total GB mileage do not supply these denominators.

Download compatibility register

Method, sources and reconciliation

Rate = pooled numerator ÷ pooled matched vehicle miles × 100,000. Annual rates are not averaged. Missing or non-positive exposure is unavailable, never zero.
Injury collisions count each collision identifier once, including collisions involving only non-motor road users. All-motor mileage is a road-exposure index, not an individual driver probability. Damage-only incidents, unreported injuries, insured claims and fault are outside this measure.
Vehicle involvements count each collision–vehicle pair once and use that vehicle class’s mileage. A collision may contribute several involvements. LGV/HGV results include known-weight vehicles only; unknown-weight goods vehicles are retained in Other. Bus/minibus classifications and other unsupported classes have no rate.
Adjusted KSI counts each casualty once: fatalities plus DfT adjusted serious-injury weights. Casualties are joined to the collision, never copied to every involved vehicle. Fractional adjusted KSI is not an integer count.
Derived descriptive rate identifies the calculated result. Exposure provenance is separate: official aggregate mileage or profile-allocated mileage. The legacy Official estimate and Modelled estimate codes in archived data describe the denominator, not DfT endorsement of the derived rate. Traffic estimation and reporting uncertainty remain.
Monthly exposure allocates the annual GB total using the trailing five-year TRA0305 all-road, all-motor daily-flow index × actual days in the month, then normalises to the annual total. These modelled profiles smooth real annual changes, particularly during the pandemic; they are not observed monthly miles.
Day-of-week exposure uses same-year TRA0306 indices × occurrences of each weekday. The hour × weekday grid uses the joint same-year TRA0307 indices × actual Europe/London labelled hours, then normalises. Hour and broad-band views sum that grid. TRA0306 and TRA0307 are separate official profile estimates and need not have identical day margins.
Leap years and calendar weekday counts are explicit. STATS19 clock time is treated as local civil time; the autumn repeated hour is pooled and the spring missing hour contributes no elapsed exposure. Full-year clock changes cancel in the Sunday 01:00 exposure total. There is no inference from 12-hour manual counts to overnight traffic.
Month, day and hour profiles are never multiplied together to manufacture a joint denominator. Road × region and road × vehicle views use published joint annual mileage. Hour × road, month × hour, region × time and road × region × vehicle rates are unavailable.
Collision intervals are approximate conditional 95% Poisson intervals (Byar approximation; exact zero-count upper limit). Vehicle intervals, when collision-level sums of squared involvement counts are available, use a collision-cluster variance approximation. KSI intervals are unavailable: severity weights and within-collision dependence require separate uncertainty modelling. No interval includes uncertainty in estimated mileage or time allocation.
Cells below 20 observed events are labelled Low count; their counts and rates remain visible with caution. Pool years to improve credibility. A higher descriptive rate is not a causal effect or an insurer pricing factor.
Unknown road, region, vehicle and time categories remain in reconciliation and exact data. Unsupported categories retain their numerators but have unavailable mileage and rates. Source retrieval, dataset approval and data periods are separate dates. Historical releases never receive unreleased input files.

Method frequency-v1. Monthly profile windows: 2021: 2017 to 2021; 2022: 2018 to 2022; 2023: 2019 to 2023; 2024: 2020 to 2024; 2025: 2021 to 2025.

Exact values · also in Data (15 rows)

Allocated exposure returns to the corresponding GB annual mileage total.
Year	Allocation	Annual miles	Allocated miles
2021	month	297,558,300,000	297,558,300,000
2021	day	297,558,300,000	297,558,300,000
2021	dayHour	297,558,300,000	297,558,300,000
2022	month	323,677,300,000	323,677,300,000
2022	day	323,677,300,000	323,677,300,000
2022	dayHour	323,677,300,000	323,677,300,000
2023	month	330,796,500,000	330,796,500,000
2023	day	330,796,500,000	330,796,500,000
2023	dayHour	330,796,500,000	330,796,500,000
2024	month	336,176,800,000	336,176,800,000
2024	day	336,176,800,000	336,176,800,000
2024	dayHour	336,176,800,000	336,176,800,000
2025	month	342,550,900,000	342,550,900,000
2025	day	342,550,900,000	342,550,900,000
2025	dayHour	342,550,900,000	342,550,900,000

Download frequency data and methodology Download source audit

Full data & reproducibility

Geographic inputs & release manifest Detailed STATS19 data & manifest Read the actuarial interpretation →

Sources & release

Loading source monitoring. Selected results are already available; live checks do not change them.

Source	Evidence period	Source published	Checked in release	Latest check	Monitoring state
RAS0403: collisions and casualties by local authority	2016–2025	Date unavailable	23 Sept 2026	21 Oct 2026	No source issue recorded
Local authority motor vehicle mileage	2016–2025	Date unavailable	23 Sept 2026	21 Oct 2026	No source issue recorded
Rainfall — England	2016–2025	Date unavailable	23 Sept 2026	23 Oct 2026	Source content changed; review pending
Rainfall — Wales	2016–2025	Date unavailable	23 Sept 2026	23 Oct 2026	Source content changed; review pending
Rainfall — Scotland	2016–2025	Date unavailable	23 Sept 2026	23 Oct 2026	Source content changed; review pending
Air frost — England

Loading approved evidence

STATS19 Data Lab

Methods and limitations

All investigations

What changes when we account for miles?

What changes when we account for miles?

Use this result

Road category

Does the rate persist over time?

Sources & release

Sources & release