The Scottish Government

« Previous | Contents | Next »

Listen

CHAPTER NINE: STATS 19

One of the key issues which this study aimed to explore was the extent to which the police recorded STATS19 database provides an accurate picture of cycle accidents in Scotland.

9.1 DEPICTING THE EXTENT OF CYCLE ACCIDENT CASUALTIES

The hospital based data collected in this study was compared with the police recorded data and a match was sought using common fields such as age, gender, date, time and place of accident.

STATS19 only records on-road accidents where the definition of "on-road" includes the footway or pavement. It could not therefore be expected that STATS19 would record the off-road accidents captured by the hospital data.

Correspondingly if the objective is to have a complete picture of cycling accidents in Scotland, then STATS19 is not designed to capture the major component of these i.e. the off-road accidents.

There were 221 cycling casualties recorded by police over the period September 2003 - August 2004 compared to 806 hospital based records collected in this study. Of these only 36 were found to be present in both data sets. Of these 36, 32 were described in the hospital data as being on the road itself, 2 as being on the pavement, 1 as off-road and 1 was not specified.

Removing off-road leaves 452 hospital records for on-road accidents ( STATS19 definition) with only 34 matches i.e. 7.5% of the volume recorded by the hospitals.

Whilst the footway or pavement is by definition part of the road for STATS19 purposes, there is some recognition that accidents occurring there are less likely to be recorded. Removing these from consideration reduces the hospital volumes to 275, but still only 32 i.e. 11.6% matching.

Figure 12: Extent of match between hospital casualties and STATS19 recorded casualties

This breakdown is shown in more detail in Appendix 4.

The number of matches is too small a sample to provide any meaningful analysis of its profile.

The low number of matched records suggests that many of those casualties identified in STATS19 did not report to a hospital. Equally that many cycling casualties who did report to a hospital did not report to the police.

Of the STATS19 dataset 197 (89%) were identified as slight casualties and it is conceivable that many of these did not report to a hospital. There was also one fatality recorded in STATS19 and this did not appear in the hospital based study. However there were 23 casualties classed as serious by STATS19 and of those 7 (30%) appeared in the hospital based data. The match rate was therefore higher for serious casualties but still not at the levels which one might expect.

Of the 7 serious casualties, information on outcome is available for only 3 and they were all discharged with a need for follow up.

There therefore exists the likelihood that the hospital based data is incomplete as there is no check as to whether the hospitals were able to obtain data for every cycling casualty. Indeed for a one-off study such as this and in the absence of a formalised system for data capture it would seem extremely unlikely that 100% data capture by the hospitals could have been possible.

Of greater concern to the direct aim of this project however is the large number of casualties who report to the hospital with an injury serious enough for medical attention who do not appear in STATS19. It may be concluded that the STATS19 database under reports the full extent of cycling casualties in Scotland.

9.2 DEPICTING THE NATURE OF CYCLING ACCIDENT CASUALTIES

This section examines some key features of the two data sets with a view to identifying whether or not the police statistics are representative of cycling accidents overall.

9.2.1 Demographic representativeness

Table 25: Gender of casualty by source of data

Both datasets showed a consistent bias towards males.

Table 26: Age of casualty by source of data

* Defined as excluding pavements and footpaths

There is a notable difference in terms of the age profiles. STATS 19 is much less likely to record accidents sustained by children (even allowing for a narrow definition of on-road) and more likely to record accidents sustained by adults in particular the 25 - 44 age group which accounted for 51% of the STATS19 dataset but only 28% of the hospital on road sample.

9.2.2 Representing the circumstances of the accident

The vast majority of STATS19 cycling records include details of another vehicle being involved. Where other vehicles were involved, they were predominantly motorised. In only one instance did STATS19 record simply a collision between two cyclists. In every other case a motorised vehicle was involved in one respect or another. Legal and insurance obligations on motorised vehicle users make it much more likely that they will involve the police at the time of an accident and therefore be recorded in the STATS19 dataset.

By contrast, the hospital data records far greater numbers of other cyclists as the "other vehicle".

Table 27: Other vehicles involved by source of data

Buses and, to a lesser extent, goods vehicles, account for proportionately more than their licensed numbers would lead one to expect. This is conceivably due simply to the greater number of hours per day these vehicles spend on the road but it may also be partly attributable to other factors such as reporting requirements of employers or the greater likelihood of them being stationary at a place on the road where they constitute an obstruction for the cyclist. Together they account for 17% of the casualties reported in STATS19 compared to only 5% of the on road casualties in the hospital dataset.

9.2.3 Representing the underlying cause

When the police report on a cycling accident they prepare a free text description of the accident based on what all the parties report to the officer. Whilst this data was not gathered for the purposes of identifying cause, it was one possible source of identifying further factors which might have contributed to the accident and a possible source of comparison with the hospital based data. The hospital based data expressly asked the cyclist what the factors were that contributed to the accident and this source is only the view of one party and may offer a biased view of the accident.

Bearing these factors in mind an exercise was undertaken to analyse the STATS19 free text description using the higher level coding frame applied to the hospital based data.

The key finding is that whilst the hospital based data exhibits a wide range of factors the police recorded text focuses very much on the motorised vehicle, mentioned as a possible factor in 78% of cases. In the police data, where the cyclist has been at fault, the classification derived from STATS19's descriptions falls into two main categories, Inattention (20%) and Behavioural (16%), with a third similar category, Lack of Skill, coming a poor fourth (3%). On the face of it this is different from the broad picture arising from the hospital data.

« Previous | Contents | Next »