The Scottish Government

« Previous | Contents | Next »

Listen

Comparison of Police "STATS 19" Road Casualty Statistics with some other figures for Scotland

Figure 10: Comparison of Police "Stats 19" and GROS figures for numbers of road deaths

NB: there are definitional differences between "Stats 19" and GROS figures. For example:

• GROS figures cover all deaths in accidents involving motor vehicles, wherever they occur, so include those occurring on private roads, in car parks, "off-road" in farmyards and at workplaces, etc - whereas "Stats 19" figures relate only to accidents on public roads.
• someone who was fatally injured in a road accident but died more than 30 days later would be counted as a serious injury (rather than a fatality) in the "Stats 19" returns, but would be counted by GROS as a death due to a road traffic accident.
• someone who was fatally injured in Scotland and died in England less than 30 days later would be counted in the Scottish "Stats 19" returns as a fatality, but would not be counted as such by GROS because he/she did not die in Scotland.

so one cannot expect the two sets of figures to agree exactly.

1. Introduction

The injury road accident statistics produced from police forces' "Stats 19" returns cannot be comprehensive. The police can only report details of the accidents of which they are aware. There could be many injury accidents, not reported by the public to the police, which do not appear in the "Stats 19" returns. This note compares the official road accident casualty statistics for Scotland, produced from "Stats 19" returns, with figures from some other sources. It refers to:

• General Register Office for Scotland road death figures (Section 2)
• numbers of emergency admissions to hospital as the result of road traffic accidents (Section 3);
• findings from two studies of casualties at a few individual hospitals (Section 4);
• results from the Scottish Household Survey (Section 5);
• some other research into these matters (Section 6).

The main conclusions are:

• "Stats 19" figures are a reliable measure of the level of, and trends in, the number of road deaths - they are very similar to GROS figures, but not the same due to definitional differences;
• "Stats 19" killed and seriously injured ( KSI) figures have fallen by 30% between 1996 and 2004, compared with a fall of only 16% in hospital admissions due to road traffic accidents;
• "Stats 19" child KSI figures have fallen by 52% between 1996 and 2004, compared with a fall of only 31% in child hospital admissions due to road traffic accidents;
• the comparison between "Stats 19" and hospital figures suggests that pedal cyclists are the type of casualty most "under-reported" in police figures: in 2004, their "Stats 19" figure was only 29% of their hospital admissions figure (and only 20% for child pedal cyclist casualties). Research found that many casualties who went to a hospital with an injury serious enough for medical attention did not appear in "Stats 19" (after excluding those who reported their accident as being "off-road", who would not be within the scope of the "Stats 19" returns);
• 35% of adults interviewed in the Scottish Household Survey who had been injured in a road accident in the past year said that it had not been reported to the police;
• the work done so far cannot answer the question of whether there has been a systematic change in the degree of "under-reporting" and the Department for Transport plans further work.

2. Road deaths: comparison with General Register Office for Scotland figures

GROS records the numbers of deaths registered in Scotland each year due to injuries sustained in motor vehicle (and other road vehicle) accidents. There are definitional differences between "Stats 19" and GROS figures, so one cannot expect the two sets of figures to agree exactly. However, the net effect of such differences is slight: Figure 10 shows that the "Stats 19" and GROS numbers of road deaths are similar in every year, that they tend to rise and fall together, and that, in 2005, they were at the lowest level that has been recorded for many years. So the "Stats 19" figures are a reliable measure of the level of, and trends in, the number of road deaths.

3. Road accident casualties who were killed or seriously injured ( KSI): comparison with hospital admission figures

This section compares "Stats 19" KSI with hospitals' numbers of emergency admissions as the result of road traffic accidents. It uses "Stats 19" KSI because, in road accident statistics:

• "serious" injuries include any for which a person is detained in hospital as an in-patient; and
• a "fatal" injury results in death less than 30 days after the accident, so some hospital admissions will later be counted as road deaths (but other road deaths occur before reaching hospital).

Figure 11: Comparison of Police "Stats 19" and Hospital "admissions as a result of a road traffic accident" figures

The hospital admissions figures for 1980 to 1995 are Scottish Hospital In Patient System ( SHIPS) figures for emergency hospital admissions as a result of a road traffic accident, as shown in a TRL research report (see Section 6); the figures for 1996 onwards were supplied by the Healthcare Information Group of the health service's Information Services Division, or obtained from ISD Scotland's Web site www.isdscotland.org - from the home page click on: Information and StatisticsthenHealth and CarethenUnintentional Injuries which takes you to a page where there is a link to a summary table on emergency hospital admissions and to an Information page which leads to more detailed figures.

Hospital admission figures are based on periods of care ("episodes") under a particular consultant, so patients can be counted more than once (e.g. if they transfer to another consultant). On admission to hospital, patients who had been involved in road traffic accidents are recorded specifically as being injured in a road traffic accident, to differentiate them from those who were involved in accidents that occurred "off-road" (so the numbers should be comparable with the "Stats 19" figures). The break in the "hospital admissions" series between 1995 and 1996 is due to changes to (a) the way in which patients' conditions were recorded and (b) the reporting period (from the calendar year to the financial year) - it should not affect comparisons of the trends since 1996 in the "Stats 19" and "hospital" figures.

Figure 11 shows that both sets of figures have been falling over the past 20-or-so years. However:

• up to the mid-1990's the "Stats 19" and "hospital" figures were broadly the same, and tending to fall at similar rates;
• since the mid-1990's the "Stats 19" figures have been noticeably lower than the "hospital" figures, and have fallen more rapidly. The changes between 1996 and 2004 are:
• All ages:
  • "Stats 19" KSI - 30% fall
  • Hospital admissions - 16% fall
• Children:
  • "Stats 19" KSI - 52% fall
  • Hospital admissions - 31% fall

As a result, the "Stats 19" figures represent a decreasing percentage of the hospital figures. Between 1980 and 1995, the overall average for "Stats 19" figures as a percentage of the hospital figures was 107% for "killed and serious" (and 99% for "serious" alone); between 1996 and 2004, it was only 79% for "killed and serious" (and 72% for "serious" alone). One can speculate about possible reasons for the different trends - there might have been:

• reduced reporting of road accidents by the public to the police (and hence increased under-reporting in "Stats 19"); and/or
• an increase in the proportion of road casualties going to a hospital; and/or
• changes in hospitals' practices (e.g. changes in the kinds of injuries for which they would normally admit a casualty, or in how they record their data - e.g. perhaps they are now using "road accident" codes more often, and certain other codes less often); and/or
• road safety improvements which reduced the number of "less serious" serious injuries (those counted as "serious" in "Stats 19" but which do not involve hospital treatment); and/or
• other factors which have different effects on the trends shown by the two sets of figures.

However, we do not know which (if any) of these reasons caused the different trends in the "Stats 19" and "hospitals" figures. Further research may help (see section 6).

Hospital admissions figures by type of road user are only readily available from the current system, and hence are only available back to 1996-97. Table X shows the "Stats 19" KSI figures as percentages of the corresponding "hospital admissions due to road traffic accidents" figures.

Part (a) of Table X covers casualties ofall ages. Pedestrians and car users (the most numerous types of casualty) have the smallest percentage difference between the two sets of figures, but the gaps are widening (e.g. the "Stats 19" number of car user casualties represented 99% of the number of hospital admissions in 1998, but only 83% in 2004). Pedal cyclists show the greatest percentage difference, with recent "Stats 19" figures representing only about 30% of the numbers of hospital admissions. While many pedal cyclist accidents occur "off-road" and are therefore not within the scope of "Stats 19" (see Section 4), the use of the relevant code should ensure that only the "on-road" casualties were included in these "hospitals" figures. Some "on road" pedal cyclist accidents may not involve other vehicles, so may be less likely to be reported to the police. However, the "under-reporting" of pedal cyclist casualties has not caused the difference in trends between the "Stats 19" and "hospitals" figures: the "Stats 19" figure for pedal cyclists has remained at roughly 30% of the "hospitals" figure since 1997, fluctuating only slightly (between 28% and 33%) from year to year. The main cause of the different trends is the fall from around 100% to about 83% in the corresponding percentage for car users, who account for about half of all "Stats 19" KSI casualties.

Part (b) of Table X covers child casualties. The percentages are, in general, markedly lower than in part (a): for example, the "child" percentage for 2004 is only 56% compared with the "all ages" figure of 72% for the same year. Again, the greatest percentage difference is for pedal cyclists: the "Stats 19" figures represent only around 20% of the number of child pedal cyclists who are admitted to hospital as the result of a road traffic accident. Again, this could be due to the accidents occurring off-road and/or involving no other vehicles. Again, the "under-reporting" of child pedal cyclist casualties has not caused the difference in trends between the "Stats 19" and "hospitals" figures: the "Stats 19" figure for child pedal cyclists has remained at roughly 20% of the "hospitals" figure since 1997, fluctuating only slightly (between 16% and 22%) from year to year. The main cause of the difference in trends is the fall from over 90% to about 75% in the corresponding percentage for child pedestrians, who account for about two-thirds of all "Stats 19" child KSI casualties. In addition, there has been a smaller fall (with considerable year-to-year fluctuation) in the percentage for child car users, who account for roughly a fifth of "Stats 19" child KSI casualties.

4. Scottish Executive reports of studies of casualties at a few individual hospitals

"Extent and Severity of Cycle Accident Casualties" (2005) - Cyclists who reported to one of five Accident and Emergency Departments in the Lothian and Borders areas were asked to complete a questionnaire relating to their accident. 806 forms were collected from those (aged 5+) who had been involved in a pedal cycle accident between September 2003 and August 2004. The research found that many of the casualties who reported to hospital with a cycling injury serious enough for medical attention did not appear in the official road accident statistics. A large proportion of the accidents (41%) occurred off-road and therefore were not within the scope of the "Stats 19" returns. However, comparing only those who reported their accident as being on the road (excluding pavements), the "Stats 19" data appeared to under-report the extent of on-road cycling accidents. The cyclists attending A&E gave a wide range of causes for the accidents, and no single cause stood out. By contrast, "Stats 19" data described a smaller range of causes, with the involvement of a motor vehicle being the predominant factor. The research also found that the official statistics on road accidents were much less likely to record pedal cycle accidents involving children than those involving adults.

"Alcohol and the Pedestrian Road Casualty" (1998) - This investigated the link between pedestrian accidents and the consumption of alcohol. Five hospitals were included in the study between October 1996 and April 1997. Casualties at Accident and Emergency who had been involved in a road traffic accident were asked to take part in the study. As part of the research, pedestrian casualties only were linked with the "Stats 19" data, and additional analysis carried out where a match was found. Of 145 pedestrian casualties in the sample, 98 (68%) resulted in a match with "Stats 19" records. Two possible reasons were given for this: (a) insufficient information available to make a match or (b) some accidents resulting in the presentation of a casualty were not reported to the police.

5. Scottish Household Survey ( SHS) Results

The Scottish Household Survey includes an interview with one randomly selected adult (aged 16+) per household in a sample spread across Scotland. The results are weighted to take account of differences in selection probabilities and response rates. Between February 1999 and March 2003, respondents were asked whether they had been injured in a road accident in the past twelve months, and if so, how they were involved (e.g. as a driver, passenger, etc). The questions were then dropped from the survey, and reinstated in 2005 with an addition: respondents were also asked whether the accident had been reported to the police.

Table Y compares percentages of adults who had been injured (any severity) in an accident in a year calculated from the SHS and "Stats 19" data (although they are not on the same basis, this should not matter given the extent of the difference between the two sets of figures - even if the SHS percentages were halved to take account of possible "over-estimation", they would still be significantly higher than those obtained from "Stats 19"):

• part (a) shows that, overall, "Stats 19" data suggest that about 0.4% of the adult population is injured in a road accident per year, whereas the SHS figure is 1.4%. Therefore, the "Stats 19" figure is only about 29% of the SHS one, a percentage which does not vary much with age;
• part (b) shows how this percentage varies with the type of road user - it is lowest for pedal cyclists (14%) and highest (39%) for pedestrians.

In 2005, 35% of SHS respondents who said they had been injured in a road accident in the past year said that the accident had not been reported to the police. As this figure is based on only 174 adults who said that they had been injured in a road accident in the past year, it may be subject to a large sampling error. However, whatever the true value is (say 25%, or 45%), it is clear that a large percentage of accidents involving personal injury are not reported to the Police.

6. Some other research

In 2004, the Department for Transport (DfT) commissioned research to review the previous evidence of "under-reporting", to identify the potential for using additional sources of healthcare data, and to assess the level of under-reporting and mis-classification of road traffic casualties in order to find out whether there have been any changes in reporting and/or recording practice over the period from 1996 to 2004. In June 2006, DfT published:

• the report of the results of this study: Under-reporting of road accidents: Phase 1 (Road Safety Research Report 69) by Heather Ward, Ronan Lyons and Roselle Thoreau; and also
• a paper which compared some "Stats 19" and "hospital" figures for road casualties in England: Road accident casualties: a comparison of STATS19 data with Hospital Episodes Statistics.

Both reports are available on the DfT web site. At that time, the main points that DfT made were:

• it is widely known that not all road traffic accidents are reported to the police - neither a new nor solely British phenomenon, and the subject of a major DfT study in 1996;
• from the limited data available, the serious group of casualties could be up to twice as large as indicated by the "Stats 19" serious category - entirely consistent with the earlier work, so the data for 1994-1998 (the baseline for the road casualty reduction targets) would also have been affected by under-reporting;
• the trends in the "Stats 19" and "hospitals" data are different;
• the work done so far cannot answer the question of whether there has been a systematic change in the degree of under-reporting. Any conclusions drawn from a simple comparison of "Stats 19" and "hospitals" figures would be misleading. As a result of the reports, DfT will shortly commission a more extensive project to address this question.

Subsequently, in the Preface to Road Casualties Great Britain 2005, DfT stated that:

• … an appreciable proportion of non-fatal accidents are not reported to the police. In addition some casualties reported to the police are not recorded and the severity of injury tends to be underestimated. …. The Department is undertaking further research to investigate whether the level of under-reporting has changed.

TRL Report 420 "Linkage of STATS 19 and Scottish hospital in-patient data - analyses for 1980-1995" (published in 1999) contains a comparison of the police "Stats 19" road accident statistics for serious injury (the definition of which includes any non-fatal-within-30-days-injury for which the casualty is detained in hospital as an in-patient) and Scottish Hospital In Patient System ( SHIPS) figures for emergency hospital admissions as a result of a road traffic accident up until 1995. These sets of figures show similar downward trends (that report's series of SHIPS figures was used to produce the "hospital" 1980-1995 line in Figure 11).

Table X - "Stats 19" numbers killed and seriously injured as a percentage of the corresponding "hospital admissions" figures

(a) casualties - all ages

(b) child*** casualties

* Calendar year for "Stats 19"; financial year for hospital figures - e.g. "Stats 19" figure for 1996 as a percentage of hospital figures for 1996-97.
** Other includes Bus/Coach, Goods and Other (Suicides and Natural Causes) - NB: neither suicides nor natural causes should be counted in the "Stats 19" figures.
*** : in the "Stats 19" figures, a "child" is someone aged 0-15; in the "hospital admissions" data, a child is someone aged 0-14.

Because the comparisons are based on overall numbers they do not represent the full extent of the differences between the two sources of data (a casualty counted in "Stats 19" but not in the hospital admissions figures will "off-set" one counted in the hospital figures but not in "Stats 19").

Table Y - adults who were injured (any severity) in a road accident in a year - as calculated from the SHS and the "Stats 19" data.

(a) by age group

(a) by type of road user

NB: the SHS and "Stats 19" figures are not on quite the same basis - for example:

• they relate to different periods - the "Stats 19" figures are calculated from the data for the years from 2000 to 2004, inclusive, whereas the SHS figures are based on the combined data from the samples for two periods (February 1999 to March 2003, and January 2005 to September 2005)
• the SHS respondent is asked whether he/she was injured in a road accident in the past year. An injury obtained (say) 13/14 months ago might be counted, if the respondent couldn't remember exactly when it was, which might have inflated the SHS figures slightly.
• the term "injury" is subjective - what an SHS respondent regards as an injury may differ from what the police would count as an injury, which could also affect the comparison.
• the SHS data relate only to members of Scottish households; the "Stats 19" data will include casualties from outwith Scotland who are injured in Scotland (and exclude Scots who are injured outwith Scotland).

« Previous | Contents | Next »