Estimation of risk factors for fatal single vehicle crashes

Monash University Accident Research Centre – Report #121 - 1997

Authors: N. Haworth, P. Vulcan, L. Bowland & N. Pronk

Full report in .pdf format [2.5MB]

Abstract:

This report presents the findings of the Case-control study of fatal single-vehicle crashes. The cases are single-vehicle crashes (or crash trips) which occurred during the period from 1 December 1995 to 30 November 1996, in which at least one vehicle occupant was killed. The cases have location, driver/rider and vehicle characteristics. The controls are (non-crash) trips which also have location, driver/rider and vehicle characteristics.

A number of driver and vehicle factors were identified which increased the relative risk of being involved in a fatal single vehicle crash.

Executive Summary

This report presents the findings of the Case-control study of fatal single-vehicle crashes. The cases are single-vehicle crashes (or crash trips) which occurred during the period from 1 December 1995 to 30 November 1996, in which at least one vehicle occupant was killed. The cases have location, driver/rider and vehicle characteristics. The controls are (non-crash) trips which also have location, driver/rider and vehicle characteristics.

The cases are described in detail in the companion volume entitled Characteristics of fatal single vehicle crashes. A summary of the overall study has also been produced (Single vehicle crash study: Summary report).

The aims of the Case-control study of fatal single-vehicle crashes study were to:

  1. investigate single vehicle crashes to determine the circumstances and factors contributing to them
  2. estimate the over-involvement (relative risk) of these factors
  3. identify improvements in procedures for the investigation of road deaths and life threatening injuries
  4. provide information from which countermeasures can be developed to the agencies responsible for road safety in Victoria

This report addresses the second aim of the study.

COLLECTION OF CONTROLS

During the 12 months of the study, 865 drivers or riders were stopped at 100 control sites. The majority of vehicles stopped were cars (90%), with 19 motorcycles (2%) and 72 light commercial vehicles (8%).

Four sets of information were collected for control motorists: licence data, roadside observations, roadside interviews and follow-up telephone interviews. Follow-ups were completed for 70% of motorists stopped.

COMPARISONS OF CASES AND CONTROLS

Case-control comparisons were conducted for cars and light commercial vehicles. In this study, the controls were not involved in a crash, so the relative risks relate to both crash involvement and crash severity, rather than to crash involvement alone.

Preliminary analyses showed that the effects of age of the driver and BAC level on risk of crashing were very strong. These effects masked or accentuated the true effects of factors which were correlated with age or BAC level. Therefore logistic regression was used to calculate the odds ratios adjusted for the effects of age and BAC level, and other variables where appropriate.

Driver age and experience

Drivers aged under 25 and those aged 60 and over were at higher risk of being involved in a fatal single vehicle crash than drivers aged 25 to 59. Among these groups the risks were greatest for drivers aged under 21 and 70 and over.

Compared with full licence holders, learner permit and probationary licence holders were at higher risk. Having driven for less than three years and having driven the current vehicle for less than 10,000 km were also associated with increased risk of being involved in a fatal single vehicle crash.

Among drivers with at least five years experience, those who had been involved in a previous crash in the last five years were at greater risk of crashing.

Alcohol and cannabis

Alcohol was present in many crashes and was associated with significantly increased crash risk. The lack of control drivers with high BAC levels prevented estimation of relative risk for BAC>.15.

Alcohol or cannabis was present in some cases and some controls. Among the cases (but not the controls), cannabis was most commonly present at high levels of alcohol and very rarely present where there was no alcohol. This meant that statistical tests of the effects of the interaction of the two drugs could not be performed.

Differences in data collection methods for cases and controls may have inflated the risk estimate for cannabis. The analysis showed that the presence of cannabis increased the risk of crashing when BAC<=.05. When all BAC data was combined, the presence of cannabis was associated with an increased risk of crashing but it is unclear the extent to which this resulted from an effect of cannabis alone or from an interaction with alcohol.

Passengers

Among the drivers who had passengers, having male passengers rather than female or male and female passengers was associated with higher crash risk. Having passengers aged 15-24 was also identified as a risk factor.

Vehicle-related factors

Not wearing a seatbelt was associated with significantly increased risk, as was driving a vehicle manufactured before 1978. Driving someone else’s car (not the employer’s) was also associated with increased crash risk.

Trip factors and activities in the previous 24 hours

Unfamiliarity with the road and reason for the trip did not affect crash risk after adjusting for potentially confounding variables.

Assessment of the crash risk related to excessive speed was restricted by the unavailability of speed data for more than half of the crashes. It is recommended that a future project involve estimation of pre-collision speeds from the scale plans available for most of the crashes followed by a reanalysis of the speed data.

The available data shows higher median speeds for crash than control vehicles. Factors associated with higher speed in crashes were drivers aged under 25 and BAC>.050.

Drivers who had slept for more than ten hours in the previous 24 hours were found to be at higher risk of being involved in a fatal single vehicle crash.

Site factors

The analyses of the site variables were complicated by a number of spurious relationships and the interrelationships of the site variables. In addition, the data set included only 127 crash sites and 100 control sites which meant that an effect would have to be large before it could be detected. Taking into consideration the problems of insufficient statistical power, the following conclusions can be drawn.

The risk of occurrence of a single vehicle crash at a site was found to increase by 3% with every extra 1000 vehicles per day. Relative risk was also higher at sites on curves.

The reductions in risk observed with the presence of side drains and traffic controls are likely to have resulted from biases in data collection procedures.

Trees and poles were present in about three-quarters of the crashes. However, the difficulties in identification of control objects prevented the relative risks for trees and poles from being calculated.

While the inability of this study to identify road factors which contribute strongly to the risk of occurrence of fatal single vehicle crashes is somewhat disappointing, it is a similar finding to that of earlier studies.

A number of other factors which were found to increase crash risk are identified in the body of the report. Interpretation of these findings may be clouded by extensive missing data for cases or other complications.

SUMMARY OF FINDINGS

The adjusted odds ratios for the significant risk factors are summarised in the tables which follow.

Table 1. Matrix of risk factors and their magnitude.
Within each category, risk factors are ranked in order of the size of the odds ratio.

Risk factor

% crashes

% controls

Odds ratio

Driver age, licensing, experience      
Learner permit

5

0

10.9

Passengers aged 15 to 24

25

9

8.7

Previous crash in last 5 years 1

25

4

6.1

Driver aged under 25

41

15

3.9

Driven vehicle less than 10,000 km

53

21

3.7

Driven less than 3 years 1

31

6

3.5

Probationary licence

22

5

3.4

Driver aged 60 and over 2

17

12

2.3

Unlicensed driver

13

0

undefined

Alcohol and cannabis      
Using both alcohol and cannabis

16

0

118.4

BAC>.05 3

26

0

95.8

Cannabis only

3

1

9.3

Other driver characteristics      
Depressed

14

4

2.7

Vehicle-related characteristics      
Not wearing seat belt 2

13

2

8.4

Windows closed 1

78

57

2.4

Pre-1978 vehicle

21

9

2.3

Heater on 1

53

32

2.2

1 Considerable missing data for cases
2 Percent in drivers with BAC<=.05
3 Percent in drivers without cannabis

Table 2. Matrix of risk factors for which countermeasures are less likely.
Within each category, risk factors are ranked in order of the size of the odds ratio.
Risk factor

% crashes

% controls

Odds ratio

Other driver characteristics      
Receiving a benefit

21

10

5.0

Sleeping more than 10 hours in previous 24

19

7

4.4

Having male passengers only

20

11

3.0

Never married

53

23

2.2

Vehicle-related characteristics      
Driving someone else’s car

36

7

4.5

Site factors      
Curves

30

16

2.4

Sponsoring Organisation: Baseline Research Program - Department of Justice, Transport Accident Commission, Royal Automobile Club of Victoria (RACV) Ltd, VicRoads