- Email: [email protected]
The effects of age and experience on accidents with injuries: Should the licensing age be raised?
Accid. Anal. & Prev. Vol Printed in Great Britain.
24. No. 2, pp. 107-116.
1992
IXXll-4575192 15.00+ .M 0 1992 Pergamon Press Ltd.
THE EFFECTS OF AGE AND EXPERIENCE ON ACCIDENTS WITH INJURIES: SHOULD THE LICENSING AGE BE RAISED? CLAIRE LABERGE-NADEAU,* URS MAAG~ AND *Department of Social and Preventive Medicine. Faculty of Computer Science and Operations Research, SDepartment of Transportation Safety, Center for Research on Transportation, 6128, Succ. A Montreal. Quebec, Canada, (Received 26 September
ROBERT BOURBEAU$ Medicine, tDepartment of Demography, Laboratory on Universitt de Montreal, C.P. H3C 337
1989; in revised form 27 January 1991)
Abstract-In Canadian Provinces and in several states of the United States the minimal legal age to drive a motor vehicle is 16 years old and in some. it is 15. The excess mortality and morbidity registered by 15 to 24-year-old drivers is well known. Several studies have reported that accident rates decrease with experience, but the effect of the age of new drivers has not been well documented. The objective is to study injury accident rates in terms of the age and experience factors. The data sources are computer files of the Government Insurance Corporation (Societe de I’assurance automobile du Quebec), which covers all Quebec drivers. For each driver, the file contains birth date, sex, year and month of first license, involvement in accidents, and other parameters. The yearly rates (1970-1984) of new permits per age last birthday and sex show an increase over time, particularly for 16-year-old men. For the period 1979-1984, injury accident involvement rates were computed for all Quebec drivers by age, sex, and driving experience. An experienced driver has been defined as a person who has been licensed for at least one year. The results show, for experienced as well as inexperienced young men (16-IS), a high injury accident rate that decreases with age. For women. the rates are much lower and decrease more gradually than for men. This study does not take into account the kilometers driven. Since young drivers (16-18) have the highest accident rates. the question of regulating access to first licensing for such drivers must be examined as a possible strategy for injury prevention.
INTRODUCTION
The excess road injuries registered by young drivers have been widely recognized (Coleman 1975; Insurance Institute for Highway Safety 1981; Waller 1988; Mayhew 198.5; Laberge-Nadeau 1981). However, the minimal age for a person to be licensed to drive a car is a debated question. Recently, in Quebec, the Coroner’s Office and the President of the Conseil des affaires sociales et de la famille have recommended raising the legal driving age from 16 to 18. In Europe, it is 18, except in Great Britain where it is 17. In Canada and in the United States it varies from 15 to 17, most frequent is 16. In New Zealand it is 15 years old (Toomath and White 1982).The literature on new and young drivers has documented the importance of experience but without much debate on the age effect (Shaoul 1975, 1976; Simmonet 1988; Groeger and Brown 1989). The purpose of this paper is: (i) To look at the age of obtaining the first driver’s license over the period from 1971 to 1984 for the group 16 to 24 years old, (ii) To assess the relation between accident involvement and age, as distinct from driving experience.
METHOD
Data source
The data were obtained from the SociCtC de l’assurance automobile du Quebec, the government licensing and insurance office, which also keeps accident statistics for all Quebec drivers. The analysis was carried out on all Quebec passenger car permit holders. To attain the first objective, the age of first permit acquisition was obtained and computed separately for each year from 1971 to 1984, for men and women at age 16, 17, 18, 19, and so forth up to 24. To attain the second objective, for all permit holders who were involved 107
108
C. LABERGE-NADEAU
etal.
in an accident between 1979 and 1984 in Quebec, time elapsed since acquiring the first license (experience) and age were taken into account. Over this period of six years, we obtained the total number of license holders on June 1 of each year and also the number of drivers who had an accident reported by the police. June 1 is nearly mid-year; considering that Canadian censuses are done at this date, this choice is preferable. For new drivers, the numbers of months since first license was obtained for each of the years studied. The linkage of the driver’s license to the accident record is done as a normal procedure in Quebec for the whole driving population. Age, sex, and driving experience are included in the driver’s license file. The variable “driving experience” was split into two categories: less than one year and one year or more. Group ages were as follows: per year from 16 to 24, then the groups 25-34, 3.5-44, and 45 or more. The nature of the crashes in which the drivers were involved were classified by the police, as follows: serious injuries (dead or hospitalized patient), mild injuries, and material damage only; serious and mild injuries were retained.
Quality of the data These data are not perfect: for example, some 16-year-old drivers were given more than one year of experience. As the legal age for licensing in Quebec is 16, twelve months’ driving experience can only be achieved at age 17 or almost 18. This error stems from other permit categories such as permits for motorcycles, skidoos, ATV vehicles; it is estimated at about 1% in any year and is the same in the two groups of license holders, those involved in an accident and those who were accident free. This error was not corrected except for the 16-year-olds. The data on drivers involved in a crash needed the following exclusions and corrections: 1. Drivers without a Quebec license were discarded because rates are calculated on all Quebec driver’s licences; this accounted for 1.9% of the drivers with an accident record. 2. Missing data on sex and more on age and date of first license accounted for 10.4% (see Table 1). In order to judge their distribution in comparison to the total driver population, drivers were cross-checked with their involvement in accident severity (see Table 2). Tabie 2 shows that the drivers of unknown sex are concentrated mostly in the category of accidents with material damage only (12%). In the other categories, they represent 3.2%, 3.5%, and 3.8%. This result goes along with the known underreporting of less serious accidents (LabergeNadeau 1976; Bourbeau 1983). The drivers involved in these accidents were omitted as were those involved in accidents with material damage only; this resulted in a loss of data of about 3.5%. 3. For another 9% of the data, driving experience was missing but other variables were known. Cross tabulations showed that age was not uniformly distributed, younger ages being more involved. To minimize bias, these data were kept and distributed according to demographic practice: for each sex separately, drivers with unknown age and experience were distributed among the drivers with known
Table 1. Proportion of drivers with accidents for which sex is unknown, Quebec, 1979-1984
Driving
experience
Lass than one year One year or more Not stated Total
Drivers sex unknown 518 10 122 192 535 203 175
(0.3%) (5.0%) (94.8%) (100%)
Ali drivers
76 1 517 363 1 958
921 380 930 231
(3.9%) (77.5%) (18.6%) (100%)
Proportion
0.7% 0.7% 52.9% 10.4%
(%)
Should the licensing age be raised?
109
Table 2. Proportion of drivers with accidents for which sex is unknown, according to accident severity, Quebec, 1979-1984
Driving experience
Drivers
sex
unknown
Fatal Serious injuries Mild injuries Damage only Total
346 682 434 713 175
1 12 188 203
All drivers
10 48 326 1 572 1 958
686 359 331 855 231
Proportion
(%)
3.2% 3.5% 3.8% 12.0% 10.4%
age but unknown experience in the same proportion as this age distribution. Then, for each age group, these drivers were distributed into the two groups defined by the experience variable in the proportion of the drivers of the same age and known experience. Method of analysis
Two types of analysis were carried out according to the two objectives. First, for each year from 1971 to 1984, the age and sex specific rates were computed as follows:
AR,.,
ND, \,
= 2
P X.!
x 1,000
= Acquisition rate of first permits at age x for the year y, calculated separately for men and women. ND,., = Number of first permit holders of age x in year y. p x.y ‘= Average population of age x in year y, estimated from the Canadian censuses taken on June 1st.
AR,,,
Second, the contribution of age and experience in injury crash involvement was established by calculating separately for men and women the following rates per 1,000 drivers:
; ID,”
=
?t9
ID,” DA;,, D!y
:
DA:,
DA!,
v=79
v=79
0:’
= Y$9 +
+ NDx;1.‘-’
= Rates of new drivers involvement
in injury crashes at age X, done separately for men and women. = Drivers involved in injury accidents with less than one year of experience at age x in year y. = Number of drivers licensed less than one year at age x and year y. For all ages, except the 16-year-olds, this denominator was calculated as follows: ND,,, 2
I W-L-I
2
= Half of of the new drivers of age x in year y plus half of the new drivers of age (X - 1) in year (Y - 1).
This computation assumes that the license acquisition is uniformly distributed through the year. The distribution per month showed that this hypothesis is verified for men, but women take their first permit more often during the second half of the year.
I10
C. LABERGE-NADEAUet al.
No corrections were done for women as no major effects were anticipated. The best estimation for 16year-olds was the number of license holders of age 16 on June 1. For the experienced drivers, we computed:
ID.; =
ID;
=
Rates of experienced drivers involvement in injury crashes at age X, done separately for men and women. DA&, = Drivers involved in injury accidents with one year or more of time elapsed since the first license (Experienced). D,& = Number of drivers licensed for at least one year at age x and in year y. This denominator was calculated in substracting the new drivers (D.:,.) from the total population of drivers (Dl-.v) of age x in year y.
RESULTS
Results for the acquisition rate of first permit are illustrated in Fig. 1. Young men drove at an earlier age in 1984 than in 1971: 42% of the population aged 16 drove in 1984 compared with only 6% in 1971. This trend is also shown for women, 27% to 2%, respectively. Table 3 gives the accidents as a function of age and experience and the corresponding accident rates. Mean rates for the period are presented here and illustrated in Fig. 2. Annual rates were also calculated and analyzed; they showed similar trends but are not presented here. Our first observation is that at any age or driving experience, rates registered by women drivers are much lower than men’s, less than a third. Despite this level difference, trends for both women and men have similarities. It is the age that seems to have an effect on drivers’ involvement in injury accidents. Drivers with less than one year of experience reach the highest rate at age 16, a lower one at 17, and a higher rate again at 18, declining gradually thereafter. Men drivers with one year or more of driving experience show a peak at age 18 and then a steady decline. For women drivers with one year or more of experience, the peak is at age 17, one year earlier than for men. followed by a gradual decline. The variation of the rates at ages 16 and 17 is difficult to explain and should be regarded with caution. This could be due to a data problem-more adjustments were made for these ages. It could also reflect a real situation. However, in this study, it is the trend analysis that is important. It confirms an age effect independently from the experience effect. Simple regression lines were fitted for the accidents rates (y) as a function of age (x) for the ages from 18 to 24. For men with less than one year of experience, y = 115.97 - 3.568~; with at least one year of experience y = 133.16 - 4.271~. These two straight lines are statistically distinct [F(2,10) = 4.9, p < ,051. The line for the experienced drivers is slightly above the line for the ones with less than one year of experience. For women the two lines are not statistically different [F(2,10) = 3.881, the common equation becomes y = 29.01 0.664~. The slope is clearly different from zero [F(1,12) = 132.7, p < .OOl]. The second factor studied is driving experience. For both men and women drivers, the group with less than one year of experience registered fewer injury accidents than the more experienced group, but the difference is much higher in men, at the ages of 17, 18, 19 and in women, at 17. Indeed, the relative risks (drivers with less than one year of experience versus those with at least one year) for these age groups are below one at the 5% level. After that, the levels of all other age groups are nearly the same. It must be said that these two groups are not homogeneous in terms of experience.
Should the licensing age be raised?
111
WOMEN Rate (W 400
-
16 17
-----*_--*_.
21 22
se... + _*_
18 19 26
-.-.-.-*-
23 24
300 I 200
I
1ooJ I 0
. ..I.... 70717273747576777879808t82838485
,r.,r,l
YEAR
MEN Rate fxd 400
0 76 71 72 73 74 75 76 77 78 79 86 81 82 83 84 86 YEAR Fig. 1. Acquisition rate per 1,000 first permits for 16- to 24”year-olds, QuCbec, 1971-1984.
In our set of data on license holders on June 1, we could consider driving experience in the following categories: less than 1 year, 1 year, 2 years, 3-5 years, 6-10 years, 11 years or more. We have estimated the mean driving experience of the group with at least one year experience for the ages 17, 18, and 19 under the hypothesis that one full year of experience corresponds to 1.5 exact years of experience (2.5 exact for the 19year-old). Results can be seen in Table 4. At each age, we compare a group that has the same driving experience (0.5 year) with another where experience increases with age. In the next section we shall relate this result to the interpretation of Fig. 2. DISCUSSION
These results shed a new light on the relative importance of driving experience and age-related factors in determining accident involvement in young driver populations. As driving experience is somewhat correlated with age, the results of previous evaluation studies of driver education (Shaoul 1976) suggested that the overrepresentation of male
WOMEN
MEN
Sex
35.2 19.7 12.1 13.5 12.5 10.7 10.3 9.2 9.1 8.3 ii:; 8.9 10.1
11 857 741 __" 28 650 77 402 127 589 173 166 210 478 229 512 248 922 252 876 2 539 530 1 849 321 1 995 951 7 733 663
429 375 32 031 55 805 55 254 54 439 39 740 30 391 23 201 18 823 15 362
287 968 ___ 499 087 694 080 202 337 317 293 83 083 29 328 22 423 459 880
1 1 2 2 2 2 2 20 278 14 188 13 078 63 176
792 264 199
4 655
ii: 23 24
25-34 35-44 45 +
Tot.&
16 17 18 19 20
632 673 747 682 427 312 214 171 127
Total
25.9 18.5 16.7
3 242 153 2 435 439 4 065 014
38 752 13 541 10 342
951 599 705 956 792 678
_a...
81 078 48 363 62 563
25-34 35-44 45 +
3 9 12 13 13 12
1 003 251 173
4 3 3 2 1
16 17 18 19 20 21 22 23 24
11 815 10 836
54.7 38.4 51.0 48.0 46.0 40.3 40.2 30.5 31.3
8.2
8.0 7.7 6.6
___ 17.4 14.0 13.3 12.0 10‘5 10.2 9.3 9.1
24.3
25,O 19.9 15.4
_m55.0 57.4 51.6 47.7 42.6 38.4 34.4 32.1
Less than 1 y. 1 year and +
Rate par 1000
73 506 97 995 75 842 50 824 25 475 16 079 11 118 8 847 7 054
_..m. 71 796 167 101 246 273 292 316 324 137 330 048 343 901 337 831
All driving licenses
021 761 870 437 172 648 447 270 221
Less than 1 y.11 year and +
Driverswith injuryaccident
Table 3.Rates of drivers with injury accidents according to sex, age, and driving experience per 1,000, Quebec, 1979-19M
Should the licensing age be raised?
113
EXPERIENCE
RATE %o
less than 12 months licensed
60
50 40 30 20 10 0
I
1
15
16
I
17
I
I
1
I
18
19
20
21
I
AGE(completed
I
22 23 year 1
I
I
24
25
5,‘,
1
1
35
Fig. 2. Rates of drivers involved in injury crashes according to driving experience, 1,000, Qukbec, 1979-1984.
45+
age and sex per
teenagers among accident-involved drivers would mainly be a matter of inexperience (see Harrington 1972). However, our data do not confirm this contention: injury accident rates still show a substantial decrease with age (18 to 24) for male drivers and a slight decrease for the female population, for both experienced and inexperienced driver subpopulations. Furthermore, young male drivers with at least one year of experience have higher accident rates than the ones with less than one year of experience. Table 5, taken from the RCgie de l’assurance automobile du Quebec (R.A.A.Q.) in 1985, shows the drivers involvement rate in injury accident for 10,000 driver permit holders (see Table 5). One observes that the involvement rate is high for the 20-year-old and younger ages, fluctuating over 250 per 10,000 drivers, the peak being observed at 19. The rate is also fairly stable across years of experience. This table does not separate men and women drivers and might register some fluctuations of the rates because of small numbers, since only the year 1985 was examined. However, it shows an age effect that supports the contention taken from our results. The fact that accident-involvement rates for experienced drivers look slightly higher than those of inexperienced drivers may be regarded as an artifact, since it can be reasonably assumed that inexperienced drivers accumulate lower mileage. In New Zealand, a survey of driver exposure to risk of accidents (Toomath and White 1982) found that the 15-19 age group accounted for about 14% of the over-14 population, 8% of the total annual mileage, and 27% of all the drivers involved in injury accidents. Looking at Figure 2, it seems a paradox that the more experienced drivers, particularly at the age of 18 and 19 show a higher accident rate than the less experienced drivers of the same age. If we offer the hypothesis that at young ages the age effect remains the same, it seems that (Table 4) for men a reduced driving experience (0.5 year) is safer than a longer one (1.52 years, 1.92, 2.44) up until 2.5 years when added Table 4. Estimations of the exact number of years of driving experience for each group according to age and sex, Quebec, 1979-1984 Groups
17 years 18 years 19 years
old old old
Men
Women
- 1 y.
1 y.
+
0.50 y. 0.50 y. 0.50 y.
1.52 y, 1.92 y. 2.44 y.
- 1 y.
1 y.
0.50 y. 0.50 y. 0.50 y.
1.51 y. 1.84 y. 2.24 y.
+
I14
C. LAU~KG~-NADEAU et al. Table
5. Drivers
involved
in injury accidents for 10,000 driver permit and driving experience. QuCbec, 1985
Experience Age (years
1
0
1
16
204
17
199
231
18
243
246
2
of driving 3
4
in years 5
6
holders
according
since
first
7
8
to their
age
license 9
10
145
139
Over 10
274
19
267
270
228
250
20
276
240
252
234
247
21
211
196
215
230
242
229
22
189
205
175
196
209
207
219
23
182
179
159
173
172
205
185
204
24
174
154
157
122
165
163
200
186
193
25-34
147
146
127
128
121
124
130
143
143
136
35-44
125
126
115
110
103
107.
107
101
107
113
111
118
45 or more
117
115
81
100
80
95
91
110
93
100
93
102
Source:
Simard R. Rapport de recherche. Synthese sur les accidents de la route impliquant des automobiles, 1982-1986. Regie de l’assurance automobile du Quebec, p. 17, Tableau 6, 1988.
driving experience
reduced the accident rate (see Table 4). For women, the safer level is attained in less than two years of driving experience. Two studies concluded that driving experience is acquired in a safe way at the beginning because it is imperative for the new driver to become familiar with the vehicle. The Simmonet (1985) study was based on a careful follow-up of new drivers using log books that specified their driving experience; the driver then goes through an overconfident phase before he acquires a better control of the vehicle. According to I’ONSER (1974) (now called INRETS in France, I’Institut national de recherche en transports et leur securite), it is necessary to drive 3,000 kilometers before experience becomes profitable. The examination of driving experience associated with injury accidents in our study has been based on a time dimension, measures of kilometers driven would have been more precise: this variable has been reported by Shaoul as being an index more associated with accidents. She stated that the age was not an important factor. However. her study population was quite homogenous in ages 17-18 years old. Our data shows an age factor. It thus appears that age-related factors, other than driving inexperience, are substantially associated with accident involvement, especially for young male drivers. Our results may have a bearing on practical applications: what is the appropriate minimal licensing age? A formal recommendation given by the National Committee for the age of licensure is an Injury Prevention and Control in 1989 states: “Increasing intervention that has not been sufficiently studied. It needs further research.” It should be noted that people in this age range undergo important changes with respect to personality development, socialization. and life cycle, which are likely to influence many aspects of behavior. including driving. According to research done by Groeger (Groeger and Brown 1985)) on 54 subjects, observational studies of driver involvements in dangerous situations suggest that young males drive differently from other road users. Another study (Trankle, Gelau. and Metker 1990) observed that young male drivers rated the traffic situations as less risky than did older male drivers and that no comparable age effect was present for females.
Should the licensing age be raised?
115
To the extent that driving behavior is regarded as a major determinant of accident involvement, policies aimed at behavior modification are warranted. The evidence gathered thus far (Shaoul 1975, 1976; Stock et a1. 1983; Strang et al. 1985) indicates that there is still a problem with drivers 16-25 years old and that driver education, as it is now extensively impIemented, can simply not be regarded as a satisfactory solution by itself. Other countermeasures are needed. Most curricula focus on knowledge acquisition and skill development to attain their goals. They undoubtedly intend to exert some social influence upon students, but the way this concern has been operationalized up to now remains less clear. In this respect, the power of formal education does not appear to be very strong. Such measures as probation period, night curfew, or prolonged accompanied driving might provide more vivid experiences for the desired social influence to take place. Raising the licensing age would reduce early licensing which, according to Robertson (1980), is associated with high accident involvement. A recent report to Congress (Rice et al. 1988) ranks second the strategy of raising licensing to 17, based on estimated cost and savings of interventions to reduce injury and severity. However, such a measure would meet tough political resistance in North America. From our results, it has to be recognized that age and experience in road safety does not act in the same way in men and women. Women registered much lower accident rates and appear less in~uenced by the factors age and experience than men. Given that the drivers of the youngest age group (16-18 years) have the highest injury accident rate and that the same age group shows a high acquisition rate of driver permits. we think that the licensing age should be raised to 18 as in most European countries. thank the Societe de l’assurance automobile du Quebec for provision of computerized data, particularly Andre Vie1 and Robert Simard. Our thanks are addressed to Denise Desjardins who revised the text and to Carole Bureau who typed the manuscript. This research was funded by Health and Welfare Canada. and partly by C.A.F.I.R., Universite de Montreal.
Acknowledgments-We
REFERENCES Bourbeau R. Les accidents de la route au Quebec, 1926-1978. Etude dtmographique et Cpidemiologique. In: Coil. Demographie Canadienne. Montreal; Les Presses de I’Universite de Montreal; 1983. Coleman. T. W. The driver education evaluation program (DEEP) study. Report to the Congress. DOT-HSX01-766. Washington, DC: U.S. Department of Transportation, National Highway Traffic Safety Administration: 197.5. Groeger, J. A.; Brown, I. D. Assessing one‘s own and others’ driving ability: Influences of sex, age and experience. Accid. Anal. Prev. 21: 155-168; 19X9. Harrington, D. M. The young driver followup study: An evaluation of the role of human factors in the first four years of driving. Accid. Anal. Prev., 4: 191-240, 1972. Insurance Institute for Highway Safety. Special issue teens and autos: A deadly combination. Status Report, vol. 16. no. 14. Washington, DC: Insurance Institute for Highway Safety; 1981. Laberge-Nadeau. C. Road mortality and morbidity in Canada and provinces, 1960-1975. Thesis, London. London School of Hygiene and Tropical Medicine, 1976. Laberge-Nadeau. C.; Bourbeau, R.; Maag. U. Universal driving instruction in high school fails to reduce morbidity and mortality of young Canadians. Twenty-fifth Proceedings American Association for Automotive Medicine. Morton Grove, IL: A.A.A.M.; 1981: 381-396. Mayhew. D. R.; Simpson, H. M.; Donelson. A. C. (ed). Young drivers accident: In search of solutions. Ottawa: ‘I’rPffic Injury Research Foundation of Canada; 1985. Ornanisme national de s&mite routiere. Un conducteur debutant a besoin de parcourir 3000 kms pour se sentir a I’aise au volant. Actualites, no. 44. Paris: ONSER; 1974. Rice. D. P.. et al. Cost of Iniurv in the United States. A Report to Coneress 1989. Ido VSW Red. Editor. 280 p., 1989, p. 113. d ’ Robertson, L. S. Crash involvement of teenaged drivers when driver education is eliminated from high school. Am. J. Public Health 70(6): 599-603; 1980. Shaoul. J. E. The use of accidents and traffic offences as criteria for evaluating courses in driver education. Report. Salford, England: Road Safety Research Unit, Department of Civil Engineering, University of Salford; 1975. Shaoul, J. E. The use of intermediate criteria for evaluating the effectiveness of accident countermeasures. Hum. Factors 1816): 575-586; 1976. Simard R. Rapport de recherche. Synthese stir les accidents de la route impliquant des automobiles, 19821986. RCgie de I’assurance automobile du Quebec; 1988; 17. Simmonet, M. Experimentation Yvelynes-Essonne. Bilan de l’expttrience, juin 198baoitt 1985. Personal communication and report. France, Ministtre de l’llrbanisme, du Logement et des Transports, 1985.
116
C. LABERGE-NADEAU
eta1
Stock, J. R.; Weaver, J. K.; Ray, H. W.; Brink, J. R.; Sadof, M. G. Evaluation of safety performance secondary school driver education curriculum demonstration project. Springfield, U.S. Final Report, Department of Transportation, National Highway Traffic Safety Administration, DOT-H.%806 568,1983. Strang, P. M.; Deutsch, K. B.; James, R. S.; Manders, S. M. A comparison of on-road and off-road driver training. Victoria (Australia): Report # l/82, Road Safety and Traffic Authority, 1982. The National Committee for Injury Prevention and Control, Injury Prevention: Meeting the Challenge, American Journal of Preventive Medicine. Oxford University Press, 1989, 303 p., p. 132. Toomath, J. B.; White, W. T. New Zealand survey of driver exposure to risk of accidents. Accid. Anal. Prev. 14: 407-411; 1982. Trlnkle, U.; Gelau, C.; Metker, T. Risk perception and age specific accidents of young drivers. Accid. Anal. Prev. 22: 119-125; 1990. Wailer. J. A. Injury control. A guide to the causes and prevention of trauma. Lexington, MA: Lexington Books; 1985.
24. No. 2, pp. 107-116.
1992
IXXll-4575192 15.00+ .M 0 1992 Pergamon Press Ltd.
THE EFFECTS OF AGE AND EXPERIENCE ON ACCIDENTS WITH INJURIES: SHOULD THE LICENSING AGE BE RAISED? CLAIRE LABERGE-NADEAU,* URS MAAG~ AND *Department of Social and Preventive Medicine. Faculty of Computer Science and Operations Research, SDepartment of Transportation Safety, Center for Research on Transportation, 6128, Succ. A Montreal. Quebec, Canada, (Received 26 September
ROBERT BOURBEAU$ Medicine, tDepartment of Demography, Laboratory on Universitt de Montreal, C.P. H3C 337
1989; in revised form 27 January 1991)
Abstract-In Canadian Provinces and in several states of the United States the minimal legal age to drive a motor vehicle is 16 years old and in some. it is 15. The excess mortality and morbidity registered by 15 to 24-year-old drivers is well known. Several studies have reported that accident rates decrease with experience, but the effect of the age of new drivers has not been well documented. The objective is to study injury accident rates in terms of the age and experience factors. The data sources are computer files of the Government Insurance Corporation (Societe de I’assurance automobile du Quebec), which covers all Quebec drivers. For each driver, the file contains birth date, sex, year and month of first license, involvement in accidents, and other parameters. The yearly rates (1970-1984) of new permits per age last birthday and sex show an increase over time, particularly for 16-year-old men. For the period 1979-1984, injury accident involvement rates were computed for all Quebec drivers by age, sex, and driving experience. An experienced driver has been defined as a person who has been licensed for at least one year. The results show, for experienced as well as inexperienced young men (16-IS), a high injury accident rate that decreases with age. For women. the rates are much lower and decrease more gradually than for men. This study does not take into account the kilometers driven. Since young drivers (16-18) have the highest accident rates. the question of regulating access to first licensing for such drivers must be examined as a possible strategy for injury prevention.
INTRODUCTION
The excess road injuries registered by young drivers have been widely recognized (Coleman 1975; Insurance Institute for Highway Safety 1981; Waller 1988; Mayhew 198.5; Laberge-Nadeau 1981). However, the minimal age for a person to be licensed to drive a car is a debated question. Recently, in Quebec, the Coroner’s Office and the President of the Conseil des affaires sociales et de la famille have recommended raising the legal driving age from 16 to 18. In Europe, it is 18, except in Great Britain where it is 17. In Canada and in the United States it varies from 15 to 17, most frequent is 16. In New Zealand it is 15 years old (Toomath and White 1982).The literature on new and young drivers has documented the importance of experience but without much debate on the age effect (Shaoul 1975, 1976; Simmonet 1988; Groeger and Brown 1989). The purpose of this paper is: (i) To look at the age of obtaining the first driver’s license over the period from 1971 to 1984 for the group 16 to 24 years old, (ii) To assess the relation between accident involvement and age, as distinct from driving experience.
METHOD
Data source
The data were obtained from the SociCtC de l’assurance automobile du Quebec, the government licensing and insurance office, which also keeps accident statistics for all Quebec drivers. The analysis was carried out on all Quebec passenger car permit holders. To attain the first objective, the age of first permit acquisition was obtained and computed separately for each year from 1971 to 1984, for men and women at age 16, 17, 18, 19, and so forth up to 24. To attain the second objective, for all permit holders who were involved 107
108
C. LABERGE-NADEAU
etal.
in an accident between 1979 and 1984 in Quebec, time elapsed since acquiring the first license (experience) and age were taken into account. Over this period of six years, we obtained the total number of license holders on June 1 of each year and also the number of drivers who had an accident reported by the police. June 1 is nearly mid-year; considering that Canadian censuses are done at this date, this choice is preferable. For new drivers, the numbers of months since first license was obtained for each of the years studied. The linkage of the driver’s license to the accident record is done as a normal procedure in Quebec for the whole driving population. Age, sex, and driving experience are included in the driver’s license file. The variable “driving experience” was split into two categories: less than one year and one year or more. Group ages were as follows: per year from 16 to 24, then the groups 25-34, 3.5-44, and 45 or more. The nature of the crashes in which the drivers were involved were classified by the police, as follows: serious injuries (dead or hospitalized patient), mild injuries, and material damage only; serious and mild injuries were retained.
Quality of the data These data are not perfect: for example, some 16-year-old drivers were given more than one year of experience. As the legal age for licensing in Quebec is 16, twelve months’ driving experience can only be achieved at age 17 or almost 18. This error stems from other permit categories such as permits for motorcycles, skidoos, ATV vehicles; it is estimated at about 1% in any year and is the same in the two groups of license holders, those involved in an accident and those who were accident free. This error was not corrected except for the 16-year-olds. The data on drivers involved in a crash needed the following exclusions and corrections: 1. Drivers without a Quebec license were discarded because rates are calculated on all Quebec driver’s licences; this accounted for 1.9% of the drivers with an accident record. 2. Missing data on sex and more on age and date of first license accounted for 10.4% (see Table 1). In order to judge their distribution in comparison to the total driver population, drivers were cross-checked with their involvement in accident severity (see Table 2). Tabie 2 shows that the drivers of unknown sex are concentrated mostly in the category of accidents with material damage only (12%). In the other categories, they represent 3.2%, 3.5%, and 3.8%. This result goes along with the known underreporting of less serious accidents (LabergeNadeau 1976; Bourbeau 1983). The drivers involved in these accidents were omitted as were those involved in accidents with material damage only; this resulted in a loss of data of about 3.5%. 3. For another 9% of the data, driving experience was missing but other variables were known. Cross tabulations showed that age was not uniformly distributed, younger ages being more involved. To minimize bias, these data were kept and distributed according to demographic practice: for each sex separately, drivers with unknown age and experience were distributed among the drivers with known
Table 1. Proportion of drivers with accidents for which sex is unknown, Quebec, 1979-1984
Driving
experience
Lass than one year One year or more Not stated Total
Drivers sex unknown 518 10 122 192 535 203 175
(0.3%) (5.0%) (94.8%) (100%)
Ali drivers
76 1 517 363 1 958
921 380 930 231
(3.9%) (77.5%) (18.6%) (100%)
Proportion
0.7% 0.7% 52.9% 10.4%
(%)
Should the licensing age be raised?
109
Table 2. Proportion of drivers with accidents for which sex is unknown, according to accident severity, Quebec, 1979-1984
Driving experience
Drivers
sex
unknown
Fatal Serious injuries Mild injuries Damage only Total
346 682 434 713 175
1 12 188 203
All drivers
10 48 326 1 572 1 958
686 359 331 855 231
Proportion
(%)
3.2% 3.5% 3.8% 12.0% 10.4%
age but unknown experience in the same proportion as this age distribution. Then, for each age group, these drivers were distributed into the two groups defined by the experience variable in the proportion of the drivers of the same age and known experience. Method of analysis
Two types of analysis were carried out according to the two objectives. First, for each year from 1971 to 1984, the age and sex specific rates were computed as follows:
AR,.,
ND, \,
= 2
P X.!
x 1,000
= Acquisition rate of first permits at age x for the year y, calculated separately for men and women. ND,., = Number of first permit holders of age x in year y. p x.y ‘= Average population of age x in year y, estimated from the Canadian censuses taken on June 1st.
AR,,,
Second, the contribution of age and experience in injury crash involvement was established by calculating separately for men and women the following rates per 1,000 drivers:
; ID,”
=
?t9
ID,” DA;,, D!y
:
DA:,
DA!,
v=79
v=79
0:’
= Y$9 +
+ NDx;1.‘-’
= Rates of new drivers involvement
in injury crashes at age X, done separately for men and women. = Drivers involved in injury accidents with less than one year of experience at age x in year y. = Number of drivers licensed less than one year at age x and year y. For all ages, except the 16-year-olds, this denominator was calculated as follows: ND,,, 2
I W-L-I
2
= Half of of the new drivers of age x in year y plus half of the new drivers of age (X - 1) in year (Y - 1).
This computation assumes that the license acquisition is uniformly distributed through the year. The distribution per month showed that this hypothesis is verified for men, but women take their first permit more often during the second half of the year.
I10
C. LABERGE-NADEAUet al.
No corrections were done for women as no major effects were anticipated. The best estimation for 16year-olds was the number of license holders of age 16 on June 1. For the experienced drivers, we computed:
ID.; =
ID;
=
Rates of experienced drivers involvement in injury crashes at age X, done separately for men and women. DA&, = Drivers involved in injury accidents with one year or more of time elapsed since the first license (Experienced). D,& = Number of drivers licensed for at least one year at age x and in year y. This denominator was calculated in substracting the new drivers (D.:,.) from the total population of drivers (Dl-.v) of age x in year y.
RESULTS
Results for the acquisition rate of first permit are illustrated in Fig. 1. Young men drove at an earlier age in 1984 than in 1971: 42% of the population aged 16 drove in 1984 compared with only 6% in 1971. This trend is also shown for women, 27% to 2%, respectively. Table 3 gives the accidents as a function of age and experience and the corresponding accident rates. Mean rates for the period are presented here and illustrated in Fig. 2. Annual rates were also calculated and analyzed; they showed similar trends but are not presented here. Our first observation is that at any age or driving experience, rates registered by women drivers are much lower than men’s, less than a third. Despite this level difference, trends for both women and men have similarities. It is the age that seems to have an effect on drivers’ involvement in injury accidents. Drivers with less than one year of experience reach the highest rate at age 16, a lower one at 17, and a higher rate again at 18, declining gradually thereafter. Men drivers with one year or more of driving experience show a peak at age 18 and then a steady decline. For women drivers with one year or more of experience, the peak is at age 17, one year earlier than for men. followed by a gradual decline. The variation of the rates at ages 16 and 17 is difficult to explain and should be regarded with caution. This could be due to a data problem-more adjustments were made for these ages. It could also reflect a real situation. However, in this study, it is the trend analysis that is important. It confirms an age effect independently from the experience effect. Simple regression lines were fitted for the accidents rates (y) as a function of age (x) for the ages from 18 to 24. For men with less than one year of experience, y = 115.97 - 3.568~; with at least one year of experience y = 133.16 - 4.271~. These two straight lines are statistically distinct [F(2,10) = 4.9, p < ,051. The line for the experienced drivers is slightly above the line for the ones with less than one year of experience. For women the two lines are not statistically different [F(2,10) = 3.881, the common equation becomes y = 29.01 0.664~. The slope is clearly different from zero [F(1,12) = 132.7, p < .OOl]. The second factor studied is driving experience. For both men and women drivers, the group with less than one year of experience registered fewer injury accidents than the more experienced group, but the difference is much higher in men, at the ages of 17, 18, 19 and in women, at 17. Indeed, the relative risks (drivers with less than one year of experience versus those with at least one year) for these age groups are below one at the 5% level. After that, the levels of all other age groups are nearly the same. It must be said that these two groups are not homogeneous in terms of experience.
Should the licensing age be raised?
111
WOMEN Rate (W 400
-
16 17
-----*_--*_.
21 22
se... + _*_
18 19 26
-.-.-.-*-
23 24
300 I 200
I
1ooJ I 0
. ..I.... 70717273747576777879808t82838485
,r.,r,l
YEAR
MEN Rate fxd 400
0 76 71 72 73 74 75 76 77 78 79 86 81 82 83 84 86 YEAR Fig. 1. Acquisition rate per 1,000 first permits for 16- to 24”year-olds, QuCbec, 1971-1984.
In our set of data on license holders on June 1, we could consider driving experience in the following categories: less than 1 year, 1 year, 2 years, 3-5 years, 6-10 years, 11 years or more. We have estimated the mean driving experience of the group with at least one year experience for the ages 17, 18, and 19 under the hypothesis that one full year of experience corresponds to 1.5 exact years of experience (2.5 exact for the 19year-old). Results can be seen in Table 4. At each age, we compare a group that has the same driving experience (0.5 year) with another where experience increases with age. In the next section we shall relate this result to the interpretation of Fig. 2. DISCUSSION
These results shed a new light on the relative importance of driving experience and age-related factors in determining accident involvement in young driver populations. As driving experience is somewhat correlated with age, the results of previous evaluation studies of driver education (Shaoul 1976) suggested that the overrepresentation of male
WOMEN
MEN
Sex
35.2 19.7 12.1 13.5 12.5 10.7 10.3 9.2 9.1 8.3 ii:; 8.9 10.1
11 857 741 __" 28 650 77 402 127 589 173 166 210 478 229 512 248 922 252 876 2 539 530 1 849 321 1 995 951 7 733 663
429 375 32 031 55 805 55 254 54 439 39 740 30 391 23 201 18 823 15 362
287 968 ___ 499 087 694 080 202 337 317 293 83 083 29 328 22 423 459 880
1 1 2 2 2 2 2 20 278 14 188 13 078 63 176
792 264 199
4 655
ii: 23 24
25-34 35-44 45 +
Tot.&
16 17 18 19 20
632 673 747 682 427 312 214 171 127
Total
25.9 18.5 16.7
3 242 153 2 435 439 4 065 014
38 752 13 541 10 342
951 599 705 956 792 678
_a...
81 078 48 363 62 563
25-34 35-44 45 +
3 9 12 13 13 12
1 003 251 173
4 3 3 2 1
16 17 18 19 20 21 22 23 24
11 815 10 836
54.7 38.4 51.0 48.0 46.0 40.3 40.2 30.5 31.3
8.2
8.0 7.7 6.6
___ 17.4 14.0 13.3 12.0 10‘5 10.2 9.3 9.1
24.3
25,O 19.9 15.4
_m55.0 57.4 51.6 47.7 42.6 38.4 34.4 32.1
Less than 1 y. 1 year and +
Rate par 1000
73 506 97 995 75 842 50 824 25 475 16 079 11 118 8 847 7 054
_..m. 71 796 167 101 246 273 292 316 324 137 330 048 343 901 337 831
All driving licenses
021 761 870 437 172 648 447 270 221
Less than 1 y.11 year and +
Driverswith injuryaccident
Table 3.Rates of drivers with injury accidents according to sex, age, and driving experience per 1,000, Quebec, 1979-19M
Should the licensing age be raised?
113
EXPERIENCE
RATE %o
less than 12 months licensed
60
50 40 30 20 10 0
I
1
15
16
I
17
I
I
1
I
18
19
20
21
I
AGE(completed
I
22 23 year 1
I
I
24
25
5,‘,
1
1
35
Fig. 2. Rates of drivers involved in injury crashes according to driving experience, 1,000, Qukbec, 1979-1984.
45+
age and sex per
teenagers among accident-involved drivers would mainly be a matter of inexperience (see Harrington 1972). However, our data do not confirm this contention: injury accident rates still show a substantial decrease with age (18 to 24) for male drivers and a slight decrease for the female population, for both experienced and inexperienced driver subpopulations. Furthermore, young male drivers with at least one year of experience have higher accident rates than the ones with less than one year of experience. Table 5, taken from the RCgie de l’assurance automobile du Quebec (R.A.A.Q.) in 1985, shows the drivers involvement rate in injury accident for 10,000 driver permit holders (see Table 5). One observes that the involvement rate is high for the 20-year-old and younger ages, fluctuating over 250 per 10,000 drivers, the peak being observed at 19. The rate is also fairly stable across years of experience. This table does not separate men and women drivers and might register some fluctuations of the rates because of small numbers, since only the year 1985 was examined. However, it shows an age effect that supports the contention taken from our results. The fact that accident-involvement rates for experienced drivers look slightly higher than those of inexperienced drivers may be regarded as an artifact, since it can be reasonably assumed that inexperienced drivers accumulate lower mileage. In New Zealand, a survey of driver exposure to risk of accidents (Toomath and White 1982) found that the 15-19 age group accounted for about 14% of the over-14 population, 8% of the total annual mileage, and 27% of all the drivers involved in injury accidents. Looking at Figure 2, it seems a paradox that the more experienced drivers, particularly at the age of 18 and 19 show a higher accident rate than the less experienced drivers of the same age. If we offer the hypothesis that at young ages the age effect remains the same, it seems that (Table 4) for men a reduced driving experience (0.5 year) is safer than a longer one (1.52 years, 1.92, 2.44) up until 2.5 years when added Table 4. Estimations of the exact number of years of driving experience for each group according to age and sex, Quebec, 1979-1984 Groups
17 years 18 years 19 years
old old old
Men
Women
- 1 y.
1 y.
+
0.50 y. 0.50 y. 0.50 y.
1.52 y, 1.92 y. 2.44 y.
- 1 y.
1 y.
0.50 y. 0.50 y. 0.50 y.
1.51 y. 1.84 y. 2.24 y.
+
I14
C. LAU~KG~-NADEAU et al. Table
5. Drivers
involved
in injury accidents for 10,000 driver permit and driving experience. QuCbec, 1985
Experience Age (years
1
0
1
16
204
17
199
231
18
243
246
2
of driving 3
4
in years 5
6
holders
according
since
first
7
8
to their
age
license 9
10
145
139
Over 10
274
19
267
270
228
250
20
276
240
252
234
247
21
211
196
215
230
242
229
22
189
205
175
196
209
207
219
23
182
179
159
173
172
205
185
204
24
174
154
157
122
165
163
200
186
193
25-34
147
146
127
128
121
124
130
143
143
136
35-44
125
126
115
110
103
107.
107
101
107
113
111
118
45 or more
117
115
81
100
80
95
91
110
93
100
93
102
Source:
Simard R. Rapport de recherche. Synthese sur les accidents de la route impliquant des automobiles, 1982-1986. Regie de l’assurance automobile du Quebec, p. 17, Tableau 6, 1988.
driving experience
reduced the accident rate (see Table 4). For women, the safer level is attained in less than two years of driving experience. Two studies concluded that driving experience is acquired in a safe way at the beginning because it is imperative for the new driver to become familiar with the vehicle. The Simmonet (1985) study was based on a careful follow-up of new drivers using log books that specified their driving experience; the driver then goes through an overconfident phase before he acquires a better control of the vehicle. According to I’ONSER (1974) (now called INRETS in France, I’Institut national de recherche en transports et leur securite), it is necessary to drive 3,000 kilometers before experience becomes profitable. The examination of driving experience associated with injury accidents in our study has been based on a time dimension, measures of kilometers driven would have been more precise: this variable has been reported by Shaoul as being an index more associated with accidents. She stated that the age was not an important factor. However. her study population was quite homogenous in ages 17-18 years old. Our data shows an age factor. It thus appears that age-related factors, other than driving inexperience, are substantially associated with accident involvement, especially for young male drivers. Our results may have a bearing on practical applications: what is the appropriate minimal licensing age? A formal recommendation given by the National Committee for the age of licensure is an Injury Prevention and Control in 1989 states: “Increasing intervention that has not been sufficiently studied. It needs further research.” It should be noted that people in this age range undergo important changes with respect to personality development, socialization. and life cycle, which are likely to influence many aspects of behavior. including driving. According to research done by Groeger (Groeger and Brown 1985)) on 54 subjects, observational studies of driver involvements in dangerous situations suggest that young males drive differently from other road users. Another study (Trankle, Gelau. and Metker 1990) observed that young male drivers rated the traffic situations as less risky than did older male drivers and that no comparable age effect was present for females.
Should the licensing age be raised?
115
To the extent that driving behavior is regarded as a major determinant of accident involvement, policies aimed at behavior modification are warranted. The evidence gathered thus far (Shaoul 1975, 1976; Stock et a1. 1983; Strang et al. 1985) indicates that there is still a problem with drivers 16-25 years old and that driver education, as it is now extensively impIemented, can simply not be regarded as a satisfactory solution by itself. Other countermeasures are needed. Most curricula focus on knowledge acquisition and skill development to attain their goals. They undoubtedly intend to exert some social influence upon students, but the way this concern has been operationalized up to now remains less clear. In this respect, the power of formal education does not appear to be very strong. Such measures as probation period, night curfew, or prolonged accompanied driving might provide more vivid experiences for the desired social influence to take place. Raising the licensing age would reduce early licensing which, according to Robertson (1980), is associated with high accident involvement. A recent report to Congress (Rice et al. 1988) ranks second the strategy of raising licensing to 17, based on estimated cost and savings of interventions to reduce injury and severity. However, such a measure would meet tough political resistance in North America. From our results, it has to be recognized that age and experience in road safety does not act in the same way in men and women. Women registered much lower accident rates and appear less in~uenced by the factors age and experience than men. Given that the drivers of the youngest age group (16-18 years) have the highest injury accident rate and that the same age group shows a high acquisition rate of driver permits. we think that the licensing age should be raised to 18 as in most European countries. thank the Societe de l’assurance automobile du Quebec for provision of computerized data, particularly Andre Vie1 and Robert Simard. Our thanks are addressed to Denise Desjardins who revised the text and to Carole Bureau who typed the manuscript. This research was funded by Health and Welfare Canada. and partly by C.A.F.I.R., Universite de Montreal.
Acknowledgments-We
REFERENCES Bourbeau R. Les accidents de la route au Quebec, 1926-1978. Etude dtmographique et Cpidemiologique. In: Coil. Demographie Canadienne. Montreal; Les Presses de I’Universite de Montreal; 1983. Coleman. T. W. The driver education evaluation program (DEEP) study. Report to the Congress. DOT-HSX01-766. Washington, DC: U.S. Department of Transportation, National Highway Traffic Safety Administration: 197.5. Groeger, J. A.; Brown, I. D. Assessing one‘s own and others’ driving ability: Influences of sex, age and experience. Accid. Anal. Prev. 21: 155-168; 19X9. Harrington, D. M. The young driver followup study: An evaluation of the role of human factors in the first four years of driving. Accid. Anal. Prev., 4: 191-240, 1972. Insurance Institute for Highway Safety. Special issue teens and autos: A deadly combination. Status Report, vol. 16. no. 14. Washington, DC: Insurance Institute for Highway Safety; 1981. Laberge-Nadeau. C. Road mortality and morbidity in Canada and provinces, 1960-1975. Thesis, London. London School of Hygiene and Tropical Medicine, 1976. Laberge-Nadeau. C.; Bourbeau, R.; Maag. U. Universal driving instruction in high school fails to reduce morbidity and mortality of young Canadians. Twenty-fifth Proceedings American Association for Automotive Medicine. Morton Grove, IL: A.A.A.M.; 1981: 381-396. Mayhew. D. R.; Simpson, H. M.; Donelson. A. C. (ed). Young drivers accident: In search of solutions. Ottawa: ‘I’rPffic Injury Research Foundation of Canada; 1985. Ornanisme national de s&mite routiere. Un conducteur debutant a besoin de parcourir 3000 kms pour se sentir a I’aise au volant. Actualites, no. 44. Paris: ONSER; 1974. Rice. D. P.. et al. Cost of Iniurv in the United States. A Report to Coneress 1989. Ido VSW Red. Editor. 280 p., 1989, p. 113. d ’ Robertson, L. S. Crash involvement of teenaged drivers when driver education is eliminated from high school. Am. J. Public Health 70(6): 599-603; 1980. Shaoul. J. E. The use of accidents and traffic offences as criteria for evaluating courses in driver education. Report. Salford, England: Road Safety Research Unit, Department of Civil Engineering, University of Salford; 1975. Shaoul, J. E. The use of intermediate criteria for evaluating the effectiveness of accident countermeasures. Hum. Factors 1816): 575-586; 1976. Simard R. Rapport de recherche. Synthese stir les accidents de la route impliquant des automobiles, 19821986. RCgie de I’assurance automobile du Quebec; 1988; 17. Simmonet, M. Experimentation Yvelynes-Essonne. Bilan de l’expttrience, juin 198baoitt 1985. Personal communication and report. France, Ministtre de l’llrbanisme, du Logement et des Transports, 1985.
116
C. LABERGE-NADEAU
eta1
Stock, J. R.; Weaver, J. K.; Ray, H. W.; Brink, J. R.; Sadof, M. G. Evaluation of safety performance secondary school driver education curriculum demonstration project. Springfield, U.S. Final Report, Department of Transportation, National Highway Traffic Safety Administration, DOT-H.%806 568,1983. Strang, P. M.; Deutsch, K. B.; James, R. S.; Manders, S. M. A comparison of on-road and off-road driver training. Victoria (Australia): Report # l/82, Road Safety and Traffic Authority, 1982. The National Committee for Injury Prevention and Control, Injury Prevention: Meeting the Challenge, American Journal of Preventive Medicine. Oxford University Press, 1989, 303 p., p. 132. Toomath, J. B.; White, W. T. New Zealand survey of driver exposure to risk of accidents. Accid. Anal. Prev. 14: 407-411; 1982. Trlnkle, U.; Gelau, C.; Metker, T. Risk perception and age specific accidents of young drivers. Accid. Anal. Prev. 22: 119-125; 1990. Wailer. J. A. Injury control. A guide to the causes and prevention of trauma. Lexington, MA: Lexington Books; 1985.