Accident Analysis and Prevention 34 (2002) 71 84

www.elsevier.com/locate/aap

Risk factors for fatal road traffic accidents in Udine, Italy

Francesca Valent a, Flavio Schiava b, Cecilia Savonitto c, Tolinda Gallo d,
Silvio Brusaferro a, Fabio Barbone a,e,*
a
Cattedra di Igiene ed Epidemiologia, DPMSC, Uni6ersity of Udine, Via Colugna 40, 33100 Udine, Italy
Dipartimento di Pre6enzione, Azienda per i Ser6izi Sanitari N.3 Alto Friuli, Piazzetta Portuzza 1, 33013 Gemona del Friuli, UD, Italy
c
Dipartimento di Pre6enzione, Azienda per i Ser6izi Sanitari N.4 Medio Friuli, Via Manzoni 5, 33100 Udine, Italy
d
Dipartimento di Pre6enzione, Azienda per i Ser6izi Sanitari N.5 Bassa Friulana, Via dei Boschi 17, 33057 Palmano6a, UD, Italy
e
Department of Epidemiology and International Health, Uni6ersity of Alabama at Birmingham, 220J RPHB, 1665 Uni6ersity Boule6ard,
Birmingham, AL 35294 -0022, USA

Received 28 February 2000; received in revised form 10 September 2000; accepted 12 September 2000

Abstract
In the Province of Udine, Northeast Italy, mortality from road accidents is 37% higher than in the country as a whole. To
identify the major risk factors for fatal crashes in this area, we analyzed the Police reports of 10 320 road traffic accidents that
occurred from 1991 to 1996. Logistic regression was used to evaluate the association of characteristics of drivers and accidents
with accident severity. The risk of involvement in fatal rather than non-fatal accidents was lower among females than among
males (odds ratio (OR) = 0.65; 95%confidence interval (95%CI), 0.53 0.80). Compared with subjects B30 years of age, subjects
aged ]65 had a significantly increased risk of fatal injury as pedestrians (OR = 10.87; 95%CI, 4.45 26.54), car drivers
(OR=1.85; 95%CI, 1.083.18), moped riders (OR = 3.53; 95%CI, 1.42 8.78), and bicycle riders (OR = 7.72; 95%CI, 2.5623.29).
In accidents that occurred from 1:00 to 5:00 h the risk of death was higher than from 6:00 to 11:00 h among pedestrians
(OR= 8.88; 95%CI, 2.5830.52), car drivers (OR = 4.95; 95%CI, 3.09 7.95), motorcycle riders (OR = 13.44; 95%CI, 2.5471.05)
and moped riders (OR = 8.76; 95%CI, 2.4231.69). Risk of death among pedestrians, car drivers, moped, and bicycle riders was
also significantly increased on roads outside the urban center. Drivers injury was strongly associated with lack of use of seat belts
(OR=13.27; 95%CI, 9.3918.74, for fatal injury; OR = 2.49; 95%CI, 2.17 2.86, for non-fatal injury). Simple interventions
focused on protecting the weakest road users and based on law enforcement, behavioral change and environmental modification
might result in reducing the significant excess of road traffic accident mortality found in the study area. 2001 Elsevier Science
Ltd. All rights reserved.
Keywords: Injury; Road traffic accident; Mortality; Seat belts; Helmets; Elderly

1. Introduction
Among the 15 European Union member countries,
Italy ranks ninth in mortality from traffic accidents
(11.7 deaths per 100 000 population in 1997) and sixth
in accident rate (330 accidents per 100 000 population
in 1997, International Road Traffic and Accident Database, 1999). In the last years, as in most developed
countries, road traffic has increased (from 413 309 million passenger-kilometers in 1980 to 746 262 in 1994)
* Corresponding author. Tel.: + 1-205-9347163; fax: +1-2059757058.
E-mail address: fbarbone@epi2.soph.uab.edu (F. Barbone).

with a consequent increase in the risk of vehicle collision (Istituto Nazionale di Statistica, 1997). In 1986 a
national law was promulgated requiring motorcycle
riders of any age and moped riders aged less than 18
years to use helmets. This law has been into force until
March 2000. In 1989 another law was enacted establishing mandatory seat belt use among car drivers and
front-seat passengers, with very few exceptions. Enforcement, compliance and efficacy of these laws are
little known. In a survey conducted among moped and
motorcycle riders attending six high schools in Rome
(central Italy), only 50% of teenagers reported helmet
use sometimes or always (Centers for Disease Control,
1996). A comparison between the year before and the

0001-4575/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 0 1 - 4 5 7 5 ( 0 0 ) 0 0 1 0 4 - 4

72

F. Valent et al. / Accident Analysis and Pre6ention 34 (2002) 7184

year after the enactment of the seat belt law in the city
of Verona (North Italy) showed a reduction in the
number of hospital admissions following car accidents,
a decrease in mild head injuries with relative increase of
limb injuries and a slight improvement in the death/accident ratio in the post-legem period (Campello et al.,
1996). These results are difficult to generalize because
of substantial cultural, economic and environmental
differences between the northern and the southern part
of the country.
The Province of Udine, with approximately 520 000
inhabitants, is located in the North East of Italy, near
the Slovenian and Austrian borders. In addition to cars
and motorcycles, in this area also bicycles and mopeds
are very common means of transportation, both within
and outside the towns; they are frequently used by the
elderly and by the youth (mopeds can be ridden by
anyone aged ]14 years without license). An analysis
of death certificates of the residents from 1989 to 1995
estimated a standardized mortality ratio of 1.37
(95%confidence interval (95%CI), 1.28 1.47) for road
traffic accidents in Udine compared with Italy as a
whole (Valent, 1998). Limitations associated with use of
residence-based death certificates include, (a) non-correspondence between population-based measures of risk
and local environmental hazards when substantial
traffic within the area is generated by tourism, commerce, and commuting and substantial traffic in adjacent areas is generated by incoming traffic from the
original area; (b) lack of information on non-fatal
injuries; and (c) lack of information on major risk
factors for the accident. Because of the strong excess
mortality associated with road traffic accidents among
residents of the Province of Udine and in the absence of
other data, we conducted an analysis of all accidents
recorded by the local Police to identify the major
factors associated with road traffic fatalities and to
suggest possible preventive measures.

2. Materials and methods

In Italy, the Istituto Nazionale di Statistica (ISTAT)
collects data on all traffic accidents occurred on roads
open to public traffic, in which at least one person was
killed or injured and in which at least one moving
vehicle was involved. The Police officers (Traffic Policemen, Carabineers or Municipal Policemen) who arrive
at the site of the accident are responsible for filling the
appropriate official, structured form (Rapporto Statistico di Incidente Stradale or ISTAT/CTT/INC) and for
sending it to ISTAT. Information on ISTAT/CTT/INC
includes time and place of the accident, characteristics
of the vehicles involved, sex and age of drivers, injured
passengers and pedestrians. The ISTAT definition of
trucks includes vehicles used for carrying objects only,

trailer trucks with tow, articulated vehicles, semitrailers,

vehicles equipped with special instruments, non-farm
tractors and vehicles used for towing only. On the
contrary, vans and pickups are considered as cars.
Information is collected about seat belt and helmet use
at the time of accident but neither on airbags nor on
use of child restraint systems. Complete information is
obtained for up to three drivers. If more than three
vehicles are involved in the same accident, the exceeding drivers can only contribute to the total number of
injured persons, with no distinction from other vehicle
occupants. However, from 1991 to 1996 only 1% of all
accidents involved more than three vehicles (0.7% involved four vehicles, 0.2% involved five vehicles and
0.1% involved more than five). Consequences of accidents are defined as non-fatal injury (regardless of its
severity) or fatal injury (if death occurs within 7 days
from the date of the accident).
Using ISTAT/CTT/INC as data source, we analyzed
information on all accidents occurred in the Province of
Udine from 1991 to 1996. These dates were selected
because since 1991 the definition of a road accident
excludes events only causing damage to property and
1996 is the most recent year for which complete data
were available. In 6 years, 10 320 accidents were
recorded involving 18 227 drivers; 14 525 people were
non-fatally injured and 658 were killed. A validation
conducted in 1993 by the Statistical Service of the city
of Brescia (North Italy) showed that in 1991 and 1992
ISTAT received data from the Police on approximately
80% of the injuries caused by road traffic accidents
attended by local health personnel and funeral homes.
Accidents appeared not to be severe at first sight and
accidents involving a single vehicle were those more
likely to be missed by the Police (Istituto Nazionale di
Statistica, 1997).
Since the goal of our analysis was to identify factors
influencing crash severity, we chose to use the driver
rather than any subject involved in the accident as the
unit of analysis. In fact, for drivers, but not for passengers, information is collected regardless of whether they
are injured or not. In addition, information on passengers use of the seat belt is collected only for car and
truck occupants of the front seat. Finally, the driver
and, generally, not the passenger may hold some responsibility of the accident and should be the main
target of prevention. We also conducted separate analyses on pedestrians. We chose to evaluate the association
of seat belts and helmets regardless of whether their use
was mandatory, because we were interested in their
overall efficacy. Although alcohol and drug consumption are important risk factors for mortality from traffic
accidents, we did not include them in the model because, according to ISTAT/CTT/INC, during the study
period only 1.5% of all drivers were found to be driving
under the influence of alcohol (DUI) and less than 0.1%

F. Valent et al. / Accident Analysis and Pre6ention 34 (2002) 7184

under the influence of drugs. According to a survey

that we conducted in this area in 1998, values obtained from the Police grossly underestimate the current prevalence of DUI in our region. Even among
drivers who were killed, the Police reported very low
prevalence of DUI (2% of fatally injured drivers). No
blood alcohol concentration (BAC) levels were available since such an exam was not performed by forensic pathologists on the accident victims.

2.1. Statistical analysis

For all drivers combined, we estimated the likelihood of being involved in a fatal versus non-fatal accident. For car and truck drivers, we estimated both the
likelihood of being personally killed versus not injured
and the likelihood of being personally non-fatally injured versus not injured. For accidents involving riders
of motorcycles, mopeds and bicycles, or pedestrians,
we compared fatal versus non-fatal injuries, because it
is unlikely that these subjects escape injury in an accident involving at least one injured person.
The odds ratio (OR) was used to estimate the likelihood of the more severe outcome as compared with
the lesser one, under the condition that an accident
involving at least one injury had occurred. Therefore,
the ORs we present are estimates of the relative risk of
people dying or being injured given that they have
been involved in an accident. We conducted unadjusted and adjusted analyses using unconditional logistic regression (Hosmer and Lemeshow, 1989).
However, only adjusted analyses are presented. The
goal of adjusted analyses was to allow for the effect of
potential confounders so that the effect of a given
exposure was not distorted. Since in our study each
variable held some interest both as an exposure and as
a confounder of other exposures, adjustment was simply obtained including several exposure terms in a
single multivariate model. Multivariate logistic models
included terms for sex and age of involved persons,
seat belt or helmet use, vehicle type (and, for cars,
engine size), road type, accident type, time of day,
weekday and month. We could not include weight of
vehicle in our analyses, since the Police reported this
information for less than 1% of vehicles. Two-tailed
95%CI were also computed. The models were tested
against the global null hypothesis using the log likelihood ratio test. Their goodness of fit was tested using
the Hosmer and Lemeshow (1989) test.

3. Results

3.1. All dri6ers combined

Table 1 describes accident fatality according to

73

characteristics of the driver, the accident, the vehicle,

the road and the time of occurrence. Women were less
frequently involved in fatal accidents than men. When
adjustment for other factors was allowed, the OR increased with age. There was a strong direct association
with certain types of vehicles (i.e. motorcycle, bicycle
and truck rather than car), with type of road (i.e.
municipal roads within the urban center were the
safest; the OR increased for other urban roads and
was even higher for accidents outside the urban center), and with type of accident (i.e. the most fatal
involved pedestrians). The accident was more likely to
be fatal from 18:00 h to midnight and most of all in
the early morning hours. Fatality was approximately
20% lower for accidents which occurred in the summer.

3.2. Car dri6ers

Table 2 displays the relative fatality of accidents
involving car drivers. Of the 13 844 car drivers, 73.6%
were using seat belts at the time of the accident, 8.7%
were not using them and we lack information on use
for the remaining 17.6%. No relevant differences in
seat belt use were seen with regard to sex and age
category (data not shown). The utilization percentage
was lower on provincial and state roads within the
urban center (69.1 and 67.1%, respectively) and on
municipal roads outside the urban center (64.9%),
while it was higher on highways (81.5%). Overall,
when death was considered as the outcome, the OR
for non-users of seat belts was 13.27 (95%CI, 9.39
18.74). For non-fatal injury, the OR for non-users was
2.49 (95%CI, 2.172.86). Women had significantly
higher odds of non-fatal injury and lower odds of
fatal injury than men. Older age was strongly associated with a higher risk of death, whereas drivers aged
B 30 had the highest risk of non-fatal injury. There
was a strong association both of death and of non-fatal injury with road type, municipal urban roads
having the lowest odds. When non-fatal injury was
the outcome of interest, municipal urban roads had
half the risk of all other roads. When death is
considered, the association with driving on roads
outside the urban center was much stronger (OR=
5.52 (95%CI, 2.3912.77) for municipal nonurban roads; OR= 13.83 (95%CI, 8.4322.70) for
highways). There was a significantly increased OR
both of death and of non-fatal injury after 18:00 h. In
particular, from 1:00 to 5:00 h the OR of injury and
death were approximately two and five, respectively.
Car accidents were more likely to be injurious and
fatal during the fall and winter as compared with the
summer months.

F. Valent et al. / Accident Analysis and Pre6ention 34 (2002) 7184

74

3.3. Truck dri6ers

Among the 1193 truck drivers involved in accidents,

injury (n =305) and death (n = 12) were more likely to

occur when they were driving on roads outside the
urban center and on Sunday and holidays (Table 3).

Table 1
OR and 95%CI of involvement in fatal rather than non-fatal road accidents among vehicle drivers in the Province of Udine, Italy, 19911996
Independent variable

Adjusted modela

Drivers involved in road accidents

Total

Fatal

OR

(95% CI)

Total

18 227

903

4.9

Sex of dri6er
Maleb
Female
Unknown

14 154
4028
45

789
113
1

5.6
2.8
2.2

1.00
0.65

(0.530.80)

Age of dri6er
B30 yearsb
3064 years
]65 years
Unknown

6678
9467
1565
517

318
468
85
32

4.8
4.9
5.4
6.2

1.00
1.13
1.38

(0.971.32)
(1.061.79)

13 845
2105
1193
813
271

625
121
100
42
15

4.5
5.7
8.4
5.2
5.5

1.00
1.66
1.79
2.16
1.33

(1.352.05)
(1.422.27)
(1.543.04)
(0.762.32)

Road type
Within the urban center
Municipal roadb
Provincial road
State road

8298
915
1838

186
55
87

2.2
6.0
4.7

1.00
2.92
2.30

(2.134.00)
(1.762.99)

Outside the urban center

Municipal road
Provincial road
State road
Highway
Unknown

482
1697
3530
1386
81

36
131
285
111
12

7.5
7.7
8.1
8.0
14.8

3.64
3.92
4.33
3.81

(2.495.31)
(3.084.99)
(3.545.28)
(2.934.97)

14 257
648
1299
2023

567
73
86
177

4.0
11.3
6.6
8.7

1.00
4.96
1.48
1.61

(3.776.52)
(1.161.89)
(1.331.95)

4328
6497
5644
1514
244

164
251
336
130
22

3.8
3.9
5.9
8.6
9.0

1.00
1.02
1.58
2.00

(0.831.25)
(1.301.92)
(1.542.58)

12 196
3171
2860

570
166
167

4.7
5.2
5.8

1.00
1.01
1.03

(0.841.22)
(0.851.24)

5302
4850
3490
4585

246
248
179
230

4.6
5.1
5.1
5.0

1.00
1.20
1.17
1.12

(0.991.44)
(0.961.43)
(0.931.35)

Vehicle type
Carb
Motorcycle (including moped)
Truck
Bicycle
Other vehicle

Accident type
Moving vehicles crashb
Pedestrian knocking down
Collision with a stationary object
Other type
Time of day
611b
1217
1824
15
Unknown
Day of week
Working MondayFridayb
Saturday
Sunday and holidays
Month
JulySeptemberb
OctoberDecember
JanuaryMarch
AprilJune

Fatal (%)

Adjusted for all the variables which are shown in the table. 2LogLikelihood= 6624.869; 2(28) = 562.553; P =0.0001. HosmerLemeshow
statistic=14.86 with 8 DF; P= 0.0620.
b
Referent category.
a

F. Valent et al. / Accident Analysis and Pre6ention 34 (2002) 7184

75

Table 2
OR and 95%CI of injury rather than no consequence and of death rather than no consequence among car drivers in the Province of Udine, Italy,
19911996
Independent variable Number of drivers

Adjusted modela

Adjusted modelb

OR

(95%CI)

OR

(95%CI)

Dead

Total

13 844

252

1.8

6529

47.2

7063

51.0

Seat belt use

Yesc
No
Unknown

10 192
1208
2444

110
85
57

1.1
7.0
2.3

4570
764
1195

44.8
63.2
48.9

5512
359
1192

54.1
29.7
48.8

1.00
2.49

(2.172.86)

1.00
13.27

(9.3918.74)

Sex
Malec
Female

10 476
3368

219
33

2.1
1.0

4665
1864

44.5
55.3

5592
1471

53.4
43.7

1.00
1.61

(1.481.76)

1.00
0.77

(0.511.16)

Age
B30 yearsc
3064 years
]65 years
Unknown

5034
7526
975
309

86
139
21
6

1.7
1.8
2.1
1.9

2576
3383
459
111

51.2
44.9
47.1
35.9

2372
4004
495
192

47.1
53.2
50.8
62.1

1.00
0.86
1.01

(0.800.93)
(0.871.16)

1.00
1.24
1.85

(0.911.69)
(1.083.18)

Engine size (cm 3)

B1200
12001799
]1800
Unknown

4595
4856
2843
1550

66
97
61
28

1.4
2.0
2.1
1.8

2433
2232
1110
754

52.9
46.0
39.0
48.6

2096
2527
1672
768

45.6
52.0
58.8
49.5

1.00
0.78
0.59

(0.710.85)
(0.530.65)

1.00
0.95
0.79

(0.671.35)
(0.531.18)

Road type
Within the urban center
Municipal roadc
5965
Provincial road
692
State road
1422

32
12
17

0.5
1.7
1.2

2314
360
685

38.8
52.0
48.2

3619
320
720

60.7
46.2
50.6

1.00
1.81
1.55

(1.542.13)
(1.371.75)

1.00
3.95
2.73

(1.977.91)
(1.495.01)

Outside the urban center

Municipal road
362
Provincial road
1344
State road
2876
Highway
1124
Unknown
59

8
40
90
48
5

2.2
3.0
3.1
4.3
8.5

213
798
1553
579
27

58.8
59.4
54.0
51.5
45.8

141
506
1233
497
27

38.9
37.6
42.9
44.2
45.8

2.27
2.52
2.12
2.43

(1.812.84)
(2.222.86)
(1.932.33)
(2.112.79)

5.52
7.24
8.62
13.83

(2.3912.77)
(4.4011.92)
(5.6313.20)
(8.4322.70)

Time of day
611c
1217
1824
15
Unknown

3152
4802
4397
1308
185

41
62
88
57
4

1.3
1.3
2.0
4.4
2.2

1412
2126
2089
807
95

44.8
44.3
47.5
61.7
51.3

1699
2614
2220
444
86

53.9
54.4
50.5
33.9
46.5

1.00
1.00
1.17
2.33

(0.911.10)
(1.061.29)
(2.012.69)

1.00
1.09
1.75
4.95

(0.721.66)
(1.172.62)
(3.097.95)

8903

156

1.7

4199

47.2

4548

51.1

1.00

2583
2358

55
41

2.1
1.7

1195
1135

46.3
48.1

1333
1182

51.6
50.1

0.90
0.88

(0.820.99)
(0.800.97)

0.87
0.60

(0.621.23)
(0.400.88)

3856
3812
2761
3415

74
73
46
59

1.9
1.9
1.7
1.7

1654
1897
1399
1579

42.9
49.8
50.7
46.2

2128
1842
1316
1777

55.2
48.3
47.7
52.0

1.00
1.41
1.47
1.18

(1.281.55)
(1.321.63)
(1.071.31)

1.00
1.40
1.34
1.13

(0.982.02)
(0.892.01)
(0.781.65)

Month
JulySeptemberc
OctoberDecember
JanuaryMarch
AprilJune

Unhurt

Death vs. no
consequence

Total

Day of week
Working Monday
Fridayc
Saturday
Sunday and holidays

Non-fatally injured

Injury vs. no
consequence

1.00

a
Adjusted for all the variables which are shown in the table. 2LogLikelihood= 17674.172; 2(25) = 1147.356; P =0.0001. HosmerLemeshow
statistic= 7.73 with 8 DF; P= 0.4605.
b
Adjusted for all the variables which are shown in the table. 2LogLikelihood=1707.739; 2(25) = 485.078; P =0.0001. HosmerLemeshow
statistic=7.30 with 8 DF; P= 0.5045.
c
Referent category.

F. Valent et al. / Accident Analysis and Pre6ention 34 (2002) 7184

76

Table 3
OR and 95% CI of injury rather than no consequence and of death rather than no consequence among truck drivers in the Province of Udine,
Italy, 19911996
Independent
variable

Number of drivers

Total

Dead

1193

12

1.0

305

25.6

876

73.4

265
390
538

3
3
6

1.1
0.8
1.1

77
114
114

29.1
29.2
21.2

185
273
418

1161
32

12
0

1.0
0

288
17

24.8
53.1

334
806
21
32

5
7
0
0

1.5
0.9
0
0

103
186
9
7

Road type
Within the urban center
Municipal roadc
347
Provincial road
48
State road
137

2
1
2

0.6
2.1
1.5

Outside the urban center

Municipal road
30
Provincial road
108
State road
301
Highway
217
Unknown
5

0
3
2
2
0

Time of day
611c
1217
1824
15
Unknown

Total
Seat belt use
Yesc
No
Unknown
Sex
Malec
Female
Age
B30 yearsc
3064 years
]65 years
Unknown

391
493
216
78
15

Day of week
Working Monday 1030
Fridayc
Saturday
106
Sunday and
57
holidays

Non-fatally
injured

Unhurt

Injury vs. no consequence

Death vs. no consequence

Adjusted modela

Adjusted modelb

OR

(95%CI)

OR

(95%CI)

69.8
70.0
77.7

1.00
0.92

(0.931.33)

1.00
0.66

(0.123.60)

861
15

74.2
46.9

1.00
3.78

(1.798.02)

1.00

30.8
23.1
42.9
21.9

226
613
12
25

67.7
76.0
57.1
78.1

1.00
0.67
2.02

(0.500.91)
(0.775.30)

0.46

49
8
25

14.1
16.7
18.2

296
39
110

85.3
81.2
80.3

1.00
1.33
1.53

(0.573.08)
(0.892.63)

1.00
5.05
2.77

(0.4261.03)
(0.3720.65)

0
2.8
0.7
0.9
0

13
35
92
83
0

43.3
32.4
30.6
38.2
0

17
70
207
132
5

56.7
64.8
68.8
60.8
100.0

5.48
3.13
2.95
3.78

(2.4512.28)
(1.855.30)
(1.974.41)
(2.445.86)

8.77
1.63
1.97

(1.2661.04)
(0.2212.25)
(0.2416.40)

6
3
1
2
0

1.5
0.6
0.5
2.6
0

94
118
57
30
6

24.0
23.9
26.4
38.5
40.0

291
372
158
46
9

74.4
75.5
73.1
59.0
50.0

1.00
0.92
1.05
1.36

(0.661.28)
(0.701.58)
(0.772.38)

1.00
0.31
0.22
1.23

(0.071.36)
(0.022.03)
(0.198.02)

0.8

255

24.8

767

74.5

1.00

2
2

1.9
3.5

30
20

28.3
35.1

74
35

69.8
61.4

1.07
1.79

(0.671.72)
(0.973.32)

2.38
5.72

(0.4512.53)
(0.9933.09)

(0.131.58)

1.00

Month
JulySeptemberc
October

316
317

3
2

0.9
0.6

77
83

24.4
26.2

236
232

74.7
73.2

1.00
1.05

(0.721.54)

1.00
0.57

(0.093.59)

December
JanuaryMarch
AprilJune

257
303

4
3

1.6
1.0

58
87

22.6
28.7

195
213

75.9
70.3

1.00
1.24

(0.661.51)
(0.851.82)

1.34
1.10

(0.276.57)
(0.205.92)

a
Adjusted for all the variables which are shown in the table. 2LogLikelihood= 1243.580; 2(22) = 105.653; P =0.0001. HosmerLemeshow
statistic=6.42 with 8 DF; P= 0.6002.
b
Adjusted for all the variables which are shown in the table. 2LogLikelihood=108.732; 2(22) = 18.403; P= 0.6819. HosmerLemeshow
statistic=5.00 with 8 DF; P= 0.7572.
c
Referent category.

F. Valent et al. / Accident Analysis and Pre6ention 34 (2002) 7184

77

Table 4
OR and 95%CI of death rather than non-fatal injury among motorcycle riders in the Province of Udine, Italy, 19911996
Independent variable

Number of drivers

Death vs. non-fatal injury

Non-fatally injured

Adjusted modela

Total

Dead

Total

745

45

6.0

646

86.7

54

7.2

Helmet use
Yesb
No
Unknown

606
34
105

37
3
5

6.1
8.8
4.8

530
30
86

87.5
88.2
81.9

39
1
14

6.4
2.9
13.3

1.00
1.73

(0.456.70)

Sex
Maleb
Female

721
24

44
1

6.1
4.2

623
23

86.4
95.8

54
0

7.5
0

1.00
0.63

(0.075.23)

Age
B30 yearsb
3064 years
]65 years
Unknown

452
244
16
33

26
17
0
2

5.7
7.0
0
6.1

396
208
16
26

87.6
85.2
100.0
78.8

30
19
0
5

6.6
7.8
0
15.1

1.00
1.91

Road type
Within the urban center
Municipal roadb
Provincial road
State road

393
44
68

11
5
4

2.8
11.4
5.9

344
37
62

87.5
84.1
91.2

38
2
2

9.7
4.5
2.9

1.00
3.21
1.53

(1.0010.36)
(0.455.19)

Outside the urban center

Municipal road
Provincial road
State road
Highway
Unknown

17
64
125
29
5

2
8
10
2
3

11.8
12.5
8.0
6.9
60.0

14
55
109
23
2

82.3
85.9
87.2
79.3
40.0

1
1
6
4
0

5.9
1.6
4.8
13.8
0

3.58
3.51
2.30
1.69

(0.6220.64)
(1.269.75)
(0.915.80)
(0.328.97)

Time of day
611b
1217
1824
15
Unknown

147
255
276
53
14

2
9
22
9
3

1.4
3.5
8.0
17.0
21.4

131
228
234
42
11

89.1
89.4
84.8
79.2
78.6

14
18
20
2
0

9.5
7.1
7.2
3.8
0

1.00
2.83
6.67
13.44

(0.5913.58)
(1.4929.95)
(2.5471.05)

Day of week
Working MondayFridayb
Saturday
Sunday and holidays

428
137
180

23
9
13

5.4
6.6
7.2

379
118
149

88.5
86.1
82.8

26
10
18

6.1
7.3
10.0

1.00
0.95
1.20

(0.402.27)
(0.552.61)

Month
JulySeptemberb
OctoberDecember
JanuaryMarch
AprilJune

313
105
75
252

16
9
1
19

5.1
8.6
1.3
7.5

277
91
67
211

88.5
86.7
89.3
83.7

20
5
7
22

6.4
4.8
9.3
8.7

1.00
2.58
0.37
2.04

(1.016.61)
(0.052.99)
(0.954.36)

Unhurt
N

OR

(95%CI)

(0.933.92)

a
Adjusted for all the variables which are shown in the table. 2LogLikelihood= 279.102; 2(22) = 53.734; P =0.0002. HosmerLemeshow
statistic=12.43 with 8 DF; P= 0.1329.
b
Referent category.

3.4. Motorcycle riders

Female motorcycle drivers and motorcyclists aged
] 65 were 3 and 2%, respectively (Tables 2 and 4).
Among all riders, 81.3% were wearing helmets at the
moment of the accident; only 4.6% were not wearing

them and information on helmet use was missing for

14.1% of all motorcyclists. The OR for not wearing a
helmet was 1.73 (95%CI, 0.456.70). Fatal accidents
occurred relatively more frequently on provincial roads,
both within and outside the urban center. Other risk
factors for fatal accidents among motorcyclists were

F. Valent et al. / Accident Analysis and Pre6ention 34 (2002) 7184

78

evening (OR = 6.67; 95%CI, 1.4929.95) and early

morning hours (OR=13.44; 95%CI, 2.54 71.05) and
season (fall and spring).

3.5. Moped riders

Fifty moped riders died during the study period and
1253 were injured (Table 5). Only 15.6% of all moped

riders were using helmets when the accident occurred

while 42.9% were not and information on helmet use
was lacking for 41.5% of all riders. Wearing helmets at
the moment of the accident was almost exclusively
restricted to subjects under 30 years of age (30.7 vs.
3.0% among subjects 30 64 and 1.3% among elderly
riders, data not shown). The adjusted OR for not
wearing a helmet was 1.02 (95%CI, 0.362.89). When

Table 5
OR and 95%CI of death rather than non-fatal injury among moped riders in the Province of Udine, Italy, 19911996
Independent variable

Number of drivers

Death vs. non-fatal injury

Non-fatally injured

Total

Dead

1360

50

3.7

1253

212
583
565

6
25
19

2.8
4.3
3.4

1074
286

46
4

Age
B30 yearsb
3064 years
]65 years
Unknown

561
505
233
61

Road type
Within the urban center
Municipal roadb
Provincial road
State road

92.1

57

4.2

190
544
519

89.6
93.3
91.9

16
14
27

4.3
1.4

982
271

91.4
94.8

13
17
15
5

2.3
3.4
6.4
8.2

519
471
213
50

893
78
114

19
3
1

2.1
3.8
0.9

Outside the urban center

Municipal road
Provincial road
State road
Highway
Unknown

50
101
120
1
3

4
13
10
0
0

Time of day
611b
1217
1824
15
Unknown

309
526
461
49
15

Total
Helmet use
Yesb
No
Unknown
Sex
Maleb
Female

Day of week
Working MondayFridayb
Saturday
Sunday and holidays
Month
JulySeptemberb
OctoberDecember
JanuaryMarch
AprilJune

OR

(95%CI)

7.5
2.4
4.8

1.00
1.02

(0.362.89)

46
11

4.3
3.8

1.00
0.44

(0.151.29)

92.5
93.3
91.4
82.0

29
17
5
6

5.2
3.4
2.1
9.8

1.00
1.44
3.53

(0.633.29)
(1.428.78)

832
71
112

93.2
91.0
98.2

42
4
1

4.7
5.1
0.9

1.00
2.09
0.34

(0.587.49)
(0.042.68)

8.0
12.9
8.3
0
0

45
85
104
1
3

90.0
84.2
86.7
100.0
100.0

1
3
6
0
0

2.0
3.0
5.0
0
0

4.21
6.35
3.85

8
15
21
6
0

2.6
2.8
4.6
12.2
0

292
489
417
40
15

94.5
93.0
90.5
81.6
100.0

9
22
23
3
0

2.9
4.2
5.0
6.1
0

1.00
0.94
1.91
8.76

(0.382.32)
(0.794.64)
(2.4231.69)

1018
189
153

36
8
6

3.5
4.2
3.9

940
173
140

92.3
91.5
91.5

42
8
7

4.1
4.2
4.6

1.00
0.81
0.56

(0.341.92)
(0.201.51)

459
342
223
336

11
12
17
10

2.4
3.5
7.6
3.0

423
319
198
313

92.2
93.3
88.8
93.1

25
11
8
13

5.4
3.2
3.6
3.9

1.00
1.57
3.49
1.40

(0.653.80)
(1.518.08)
(0.563.49)

Adjusted modela

Unhurt

(1.2813.91)
(2.8813.98)
(1.658.96)

a
Adjusted for all the variables which are shown in the table. 2LogLikelihood= 354.261; 2(22) = 69.835; P =0.0001. HosmerLemeshow
statistic=2.02 with 8 DF; P= 0.9803.
b
Referent category.

F. Valent et al. / Accident Analysis and Pre6ention 34 (2002) 7184

79

Table 6
OR and 95%CI of death rather than non-fatal injury among cyclists in the Province of Udine, Italy, 19911996
Independent variable

Number of cyclists
Total

Dead

Total

813

41

Sex
Maleb
Female

500
313

Age
B30 yearsb
3064 years
]65 years
Unknown

Death vs. non-fatal injury

Non-fatally injured
%

Adjusted modela

Unhurt

OR

(95%CI)

5.0

733

90.2

39

4.8

29
12

5.8
3.8

437
296

87.4
94.6

34
5

6.8
1.6

1.00
0.86

(0.401.82)

262
252
279
20

5
13
23
0

1.9
5.2
8.2
0

234
231
253
15

89.3
91.7
90.7
75.0

23
8
3
5

8.8
3.2
1.1
25.0

1.00
3.28
7.72

(1.089.99)
(2.5623.29)

Road type
Within the urban center
Municipal roadb
Provincial road
State road

586
33
64

16
3
5

2.7
9.1
7.8

545
29
55

93.0
87.9
85.9

25
1
4

4.3
3.0
6.2

1.00
3.19
2.86

(0.8012.70)
(0.928.91)

Outside the urban center

Municipal road
Provincial road
State road
Highway
Unknown

18
55
54
0
3

2
6
9
0
0

11.1
10.9
16.7
0
0

16
43
43
0
2

88.9
78.2
79.6
0
66.7

0
6
2
0
1

0
10.9
3.7
0
33.3

4.93
4.85
10.23

Time of day
611b
1217
1824
15
Unknown

253
321
215
12
12

9
10
20
1
1

3.6
3.1
9.3
8.3
8.3

233
291
187
11
11

92.1
90.6
87.0
91.7
91.7

11
20
8
0
0

4.3
6.2
3.7
0
0

1.00
1.00
5.02
3.22

(0.382.66)
(2.0012.61)
(0.2836.69)

Day of week
Working MondayFridayb
Saturday
Sunday and holidays

616
105
92

32
7
2

5.2
6.7
2.2

553
94
86

89.8
89.5
93.5

31
4
4

5.0
3.8
4.3

1.00
0.79
0.22

(0.312.03)
(0.041.09)

Month
JulySeptemberb
OctoberDecember
JanuaryMarch
AprilJune

279
202
109
223

11
10
6
14

3.9
4.9
5.5
6.3

256
184
98
195

91.8
91.1
89.9
87.4

12
8
5
14

4.3
4.0
4.6
6.3

1.00
1.38
1.27
1.72

(0.523.64)
(0.413.95)
(0.714.15)

(0.9425.85)
(1.6814.02)
(3.8527.19)

a
Adjusted for all the variables which are shown in the table. 2LogLikelihood= 259.537; 2(19) = 61.168; P =0.0001. HosmerLemeshow
statistic=11.91 with 8 DF; P= 0.1554.
b
Referent category.

we repeated the analysis stratifying by age, road type

and type of accident, we found that any excess was
restricted to riders under 18 years of age (OR= 4.11
(95%CI, 0.8918.98)), to municipal roads (OR= 3.13
(95%CI, 0.3924.95) and falls from mopeds, mopeds
crashing into stationary objects, leaving the road or
knocking down pedestrians (OR = infinity (95%CI,
0.31)). Women represent 21% of moped riders and
among them the odds of death were lower than among
men. There is evidence of an increase in odds ratios as
age increased, on roads outside the urban center, by

night (OR= 8.76 from 1:00 to 5:00 h; 95%CI, 2.42

31.69) and in winter (OR= 3.49; 1.518.08).

3.6. Cyclists
Forty-one cyclists died in the 6-year period and 733
got injured on the road (Table 6). Age was strongly
associated with death (OR=3.28 (95%CI, 1.08 9.99)
among middle-aged cyclists and OR= 7.72 (95%CI,
2.5623.29) among the elderly). Fatality was significantly increased among cyclists who were involved in

F. Valent et al. / Accident Analysis and Pre6ention 34 (2002) 7184

80

accidents on non-municipal urban roads; the highest

OR refers to non-urban state roads (OR= 10.23;
95%CI, 3.8527.19) and at night.

(between 18:00 and 5:00 h) and season (winter) were

also associated with fatality among pedestrians.

3.7. Pedestrians

4. Discussion

Table 7 shows the distribution of events and OR

regarding pedestrians. Males have significantly higher
odds of fatal injuries than women. Among the elderly
there was a 10-fold increased fatality as compared with
subjects B30 years of age. Non-urban roads, time

4.1. Car seat belts

Many studies have reported the effectiveness of car
seat belts in preventing injury or death from traffic
accidents (Newman, 1986; Latimer and Lave, 1987;

Table 7
OR and 95%CI of death rather than non-fatal injury among pedestrians in the Province of Udine, Italy, 19911996
Independent variable

Number of pedestrians

Death vs. non-fatal injury

Adjusted modela

Total

Dead

Non-fatally injured

Total

715

76

10.6

639

89.4

Sex
Maleb
Female

376
339

53
23

14.1
6.8

323
316

85.9
93.2

1.00
0.48

(0.270.85)

Age
B30 yearsb
3064 years
]65 years
Unknown

215
252
220
28

7
20
43
6

3.3
7.9
19.5
21.4

208
232
177
22

96.7
92.1
80.4
78.6

1.00
2.20
10.87

(0.865.59)
(4.4526.54)

Road type
Within the urban center
Municipal roadb
Provincial road
State road

509
29
73

37
4
9

7.3
13.8
12.3

472
25
64

92.7
86.2
87.7

1.00
2.43
1.66

(0.718.25)
(0.713.88)

Outside the urban center

Municipal road
Provincial road
State road
Highway
Unknown

8
20
61
12
3

3
6
17
0
0

37.5
30.0
27.9
0
0

5
14
44
12
3

62.5
70.0
72.1
100.0
100.0

7.97
4.34
5.55

Time of day
611b
1217
1824
15
Unknown

183
224
271
33
4

13
19
35
8
1

7.1
8.5
12.9
24.2
25.0

170
205
236
25
3

92.9
91.5
87.1
75.8
75.0

1.00
1.54
2.14
8.88

(0.683.46)
(0.994.60)
(2.5830.52)

Day of week
Working MondayFridayb
Saturday
Sunday and holidays

523
108
84

53
9
14

10.1
8.3
16.7

470
99
70

89.9
91.7
83.3

1.00
0.62
1.17

(0.271.40)
(0.552.52)

Month
JulySeptemberb
OctoberDecember
JanuaryMarch
AprilJune

170
223
157
165

14
27
25
10

8.2
12.1
15.9
6.1

156
196
132
155

91.8
87.9
84.1
93.9

1.00
1.53
2.64
1.00

(0.703.36)
(1.195.87)
(0.392.59)

OR

(95%CI)

(1.4444.02)
(1.3813.62)
(2.6611.57)

Adjusted for all the variables which are shown in the table. 2LogLikelihood=380.018; 2(20) = 104.316; P =0.0001. HosmerLemeshow
statistic= 7.62 with 8 DF; P= 0.4716.
b
Referent category.
a

F. Valent et al. / Accident Analysis and Pre6ention 34 (2002) 7184

Cooper and Salzberg, 1993; Evans, 1996; Petridou et

al., 1998). Our results support these findings, providing
evidence that restriction prevents both fatal and non-fatal injuries, although risk reduction is larger in association with fatalities.
A number of methodological issues deserve a comment. Evans (1996) reported clear evidence that, when
severity of the crash was measured by the amount of
vehicle deformation, unbelted drivers were involved in
more severe crashes because travelling at higher speed
and because of other risky habits. This observation
suggests that the association between lack of use of seat
belts and injury severity might be confounded by other
risk taking behaviors. For this reason, we adjusted our
OR for variables related to characteristics of the driver,
the road, and the time of the accident. The association
did not disappear. Nevertheless, since we lack information on amount of vehicle deformation (a proxy of
change in speed of the car as a consequence of crashing), we cannot exclude residual confounding or effect
modification by risk taking behaviors other than lack of
seat belt use.
More than 70% of all car drivers were using seat belts
according to the Police inspection of the accident scene,
and little more than 10% were certainly not. A differential misclassification of seat belt use could have been
possible if some uninjured or not seriously injured
drivers were erroneously classified as restrained. If this
were true, the real protective effect provided by safety
belts would be lower, but this seems to be unlikely for
two reasons. First, according to a survey we conducted
in this area in 1998, 63% of people having a car driving
license were regular seat belt users, 26% were occasional
users and 11% were non-users. These values include
subjects both when driving and when sitting as car
passengers, and may be somehow different from percentages among drivers only. In spite of that, they
strengthen our confidence in the reliability of the Police
classification, although this value is much higher than
from other national (Campello et al., 1996) and international (Centers for Disease Control, 1993, 1995; Petridou et al., 1998) estimates. Second, the Police did not
record the information on restraint for a relevant fraction of drivers (more than 20%), indicating that they
did not report seat belt use unless they were absolutely
sure. We also repeated the analyses considering all
drivers with unknown safety belt use as restrained and
as unrestrained, in the first case the results did not
change, in the second we obtained slightly lower ORs,
but still showing a clear protective effect both for injury
and for death. The ORs for the seat belt use unknown
category were 1.30 (95%CI, 1.18 1.43) for non-fatal
injuries and 2.79 (95%CI, 1.98 3.94) when death was
considered.
We did not take into account accidents only causing
damage to property, and this probably led to an under-

81

estimation of the real seat belt protective effect because

it is likely that many drivers involved in accidents had
escaped injury just because they were restrained. Because the Police classified subjects as dead if death had
occurred within 7 days from the date of the accident,
fatalities were certainly more than ISTAT reported. It
has been estimated that about 92% of deaths occur
within 7 days and about 97% within 30 days (Istituto
Nazionale di Statistica, 1997). In addition, in Italy
deaths occurring from the second to the seventh day are
also underestimated because it is difficult for the Police
to follow an injured subject through his hospital movements after 24 h (Istituto Nazionale di Statistica, 1997).
However, this underreporting may affect both restrained and unrestrained drivers; therefore, it should
not have affected our findings.
The study had several other limitations. First, we did
not perform separate analyses for each accident type,
therefore, it is possible that the real seat belt protection
is lower for certain mechanisms and higher for others,
such as frontal crashes, as shown by Newman (1986).
Second, we only estimated the protection for restrained
drivers, therefore, generalization of our results to car
passengers deserves caution. Third, we did not know if
cars were equipped with airbags, but as they work
better if used together with correctly fastened safety
belts, higher ORs among non-users of seat belts may
reflect poorer airbag performances as well. Finally, we
considered non-fatally injured subjects as a homogeneous category because ISTAT/CTT/INC does not distinguish between light and severe traumas or among
anatomic sites of injury. Different risk values would
have probably been obtained if we had considered them
as separate outcomes.

4.2. Helmets
In contrast with others (Sarkar et al., 1995; Rowland
et al., 1996; Petridou et al., 1998), our study found only
a weak, imprecise association between wearing helmets
and fatality among motorcycle riders, and no effect
with regard to moped riders. Several factors, depending
on the nature of the available data, could have prevented us from obtaining clear evidence of helmet
protective effect, (a) chance; (b) our study did not
include just those riders who sustained head or neck
injuries, which are the only ones that could be prevented by helmets; (c) different helmet types could be
worn by moped operators, providing various degrees of
protection; (d) we considered as non-fatally injured
riders who died more than 7 days after the accident.
This outcome misclassification could have been differential if non-helmeted riders had sustained more head
injuries leading them to coma and to delayed death
than helmeted subjects had in this case, helmet
efficacy would have been underestimated; (e) helmets
may not actually reduce the risk of death.

82

F. Valent et al. / Accident Analysis and Pre6ention 34 (2002) 7184

The analysis stratified by age suggested a protective

effect of helmet at least among moped riders B 18
years of age. The difference in protective efficacy between the youngest riders and the others could be
accounted for by various factors, (a) they could have
been involved in accident of different severity because
of a different riding pattern; (b) helmets could have
caused impairment of seeing and hearing among older
moped operators increasing rather than reducing the
risk of death; (c) helmets could have damaged older
riders directly, although there are studies contrasting
with this possibility (Wagle et al., 1993); (d) the extremely low number of events among older helmet users
made the estimates very imprecise; (e) chance alone,
since the analyses are based on small numbers. Helmets
seem to be more effective when riding on municipal
than on provincial and state roads. Since road type may
be considered a good proxy for vehicle speed, accidents
on municipal streets involve on average slower vehicles
and accidents are generally less severe; in such a situation helmets appear to be the most useful. This is in
accordance with Sarkar et al. (1995) who found that
helmets did not offer much protection in crashes that
led to very severe body damage.

4.3. Sex
Women were involved in less serious accidents than
men. In addition, female drivers were less likely to die
in a car accident than male drivers. This is consistent
with findings by Li et al. (1998). However, non-fatal
injuries as opposed to no consequence occurred more
often among women than men. A similar tendency was
also emphasized by Massie et al. (1995). A concurrent
higher risk of accident with lower risk of being involved
in fatal accidents and of being personally killed could
be explained by behavioral differences between sexes,
women being possibly more inattentive or less experienced or skilled than men, but perhaps less risk-taker
(lower speeds, less driving by night, less DUI). In fact,
although we did not have information on vehicle speed,
our data showed that in this area nighttime driving and
DUI are habits almost completely limited to men.
Risky driving habits and lack of vehicle-handling skills
were also pointed out by Laapotti and Keskinen (1998)
as different causes of fatal loss-of-control accidents
between male and female drivers.

dangerous means of transportation for them, not only a

much higher proportion of old subjects was involved in
accidents while riding two-wheeled vehicles than while
driving cars, but, after an accident had occurred, the
risk of death in those cases was much higher than
among subjects B 65. A very high proportion of elderly
pedestrians was also involved in accidents and these
subjects seem to be at extremely high risk of being
fatally injured.

4.5. Trucks
Our study showed that Sundays and holidays are
particularly dangerous for truck drivers. This finding
supports the law forbidding most types of trucks to
move during these days.

4.6. Season
In our area, car drivers and motorcycle riders should
be particularly careful during the fall, because the risk
of death from road accidents is higher than in the rest
of the year. A previous study (Valent, 1998) also
demonstrated a high number of deaths from traffic
accidents in the 3 months OctoberDecember among
the inhabitants of the province of Udine. This mortality
excess might be accounted for by a number of characteristics typical of this area, (a) in October daylight
duration becomes shorter and sunset coincides with
heavy traffic; moreover, in November there is the
switch from daylight saving time to solar standard time
and, therefore, at rush hour people must drive in the
dark without having got used to that; (b) the province
of Udine is a rather foggy area and in fall this phenomenon becomes considerable; (c) fall is the most
rainy season; (d) in the province of Udine there are
many small mountain villages and the first snowfalls
and frosts often occur in November or in December; (e)
leaves fall from the trees and make the ground slippery;
(f) the consumption of alcohol, that is a well-known
risk factor for severe motor vehicle crashes (Decker et
al., 1988; Baker et al., 1992), may be particularly high
during the fall. Despite the fact that we lack information on alcohol consumption by season, in our wine
producing area the new wine starts to be sold and
drunk after Septembers grape harvest. Moreover, alcohol intake may be increased by the cold because of its
warming effect on drinkers.

4.4. Age
4.7. Two-wheeled 6ehicles
The elderly are the weakest road users. In fact, they
have the highest risk of death after an accident both
when driving vehicles, as shown also by Massie et al.
(1995), and when walking, probably because of comorbility, which reduces the possibility of recovery from
traumas. Mopeds and, most of all, bicycles are the most

Some factors may explain the high fatality among

moped riders from January to March, (a) these months
are darker and mopeds could be seen by car drivers
with more difficulty, especially if they have weak or no
light; (b) riding a moped on iced surface is particularly

F. Valent et al. / Accident Analysis and Pre6ention 34 (2002) 7184

challenging; (c) in the coldest months less people move

on foot or ride two-wheeled vehicles and more drive
cars, thus increasing car density on the roads and the
probability for mopeds to collide with larger vehicles.
Cyclists are the drivers having the highest risk of involvement in fatal accidents, and, unfortunately, in
most cases they suffer the worst injuries. Helmets are
virtually unknown to Italians, except for few young
persons who cycle as a sport. The habit of wearing a
helmet would reduce that risk (Wasserman et al., 1988;
Thompson et al., 1989; Sacks et al., 1991; Rivara et al.,
1994). Since under-reporting by the Police is greater for
bicycle accidents than for car accidents and for non-fatal rather than fatal injuries (Istituto Nazionale di
Statistica, 1997), it is possible that part of the increased
risk of involvement in fatal accidents could be explained by a greater overestimate of the proportion of
fatalities among cyclists than among car drivers. However, in Italy there is no evidence of such a differential
effect of under-reporting and, according to the validation study in Brescia (Istituto Nazionale di Statistica,
1997), the proportion of injured bicyclists reported by
the Police is high (85% of those reported by local health
personnel and funeral homes). It is apparent from the
results that riding mopeds and bicycles in the evening
and at night entails an increased fatality.

4.8. Pedestrians
Road type is a relevant risk factor for accidents
involving pedestrians, especially outside the urban center where separate paths for pedestrians, sidewalks, and
pedestrian crossing with traffic lights are rare. Lighting
conditions are factors accounting for the very high risk
of death in the evening, at night and in winter, when
natural light is poor and visibility is worsened by rain
and fog. Consistently with findings by Fontaine and
Gourley (1997), male gender is associated with a particularly high risk of death among pedestrians in our
study.

5. Conclusions
Although we were unable to address personal risk
factors for fatal accidents such as risk-taking behaviors
or DUI because of lack of information, we identified
groups of subjects and environmental conditions that
were associated with higher road accident fatality. All
things considered, a program tailored addressing the
needs of the elderly, the pedestrians, riders of twowheeled vehicles as well as any driver of a vehicle at
night is likely to be effective in reducing the fatality of
road traffic accidents. A stricter enforcement of seat
belt and helmet laws would be effective to increase
compliance with active protection devices. Stricter law

83

enforcement also should be carried out to prevent

alcohol-related accidents. Accurate assessment and
monitoring of the prevalence of DUI are needed to
estimate the actual relevance of this phenomenon.

Acknowledgements
We thank the reviewers and Stephanie Hall for their
useful comments.

References
Baker, S.P., ONeill, B., Ginsburg, M.J., Li, G., 1992. The Injury
Fact Book, second ed. Oxford University Press, New York.
Campello, C., Preite, G., Poli, A., Zuppichini, F., Marigo, M., 1996.
The effect of the seat belt law on the injuries of the road traffic
accident. Epidemiologia e Prevenzione 20, 313 317.
Centers for Disease Control, 1993. Driver safety-belt use Budapest, Hungary, 1993. MMWR-Morb and Mortal-Wkly-Rep.
42, pp. 939 941.
Centers for Disease Control, 1995. Use of safety belts Madrid,
Italy, 1994. MMWR-Morb and Mortal-Wkly-Rep. 44, pp. 150
153.
Centers for Disease Control, 1996. Helmet use among adolescent
motorcycle and moped riders Rome, Italy, 1994. MMWRMorb and Mortal-Wkly-Rep. 45, pp. 311 314.
Cooper, W.E., Salzberg, P., 1993. Safety restraint usage in fatal
motor vehicle crashes. Accident Analysis and Prevention 25,
67 75.
Decker, M.D., Graitcer, P.L., Schaffner, W., 1988. Reduction in
motor vehicle fatalities associated with an increase in the minimum drinking age. Journal of the American Medical Association
260, 3604 3610.
Evans, L., 1996. Safety-belt effectiveness: the influence of crash
severity and selective recruitment. Accident Analysis and Prevention 28, 423 433.
Fontaine, H., Gourley, Y., 1997. Fatal pedestrian accidents in
France: a typological analysis. Accident Analysis and Prevention
29, 303 312.
Hosmer, D.W., Lemeshow, S., 1989. Applied Logistic Regression.
Wiley, New York.
International Road Traffic and Accident Database, 1999. Selected
risk values for the year 1997. Issue January 1999. IRTAD
(OECD).
Istituto Nazionale di Statistica, 1997. Gli incidenti stradali negli anni
90. Rischio e sicurezza sulle strade italiane. Argomenti n.7.
ISTAT, Roma.
Laapotti, S., Keskinen, E., 1998. Differences in fatal loss-of-control
accidents between young male and female drivers. Accident Analysis and Prevention 30, 435 442.
Latimer, E.A., Lave, L.B., 1987. Initial effects of the New York State
auto safety belt law. American Journal of Public Health 77,
183 186.
Li, G., Baker, S.P., Langlois, J.A., Kelen, G.D., 1998. Are female
drivers safer? An application of the decomposition method. Epidemiology 9, 379 384.
Massie, D.L., Campbell, K.L., Williams, A.F., 1995. Traffic accident
involvement rates by driver age and gender. Accident Analysis
and Prevention 27, 73 87.
Newman, R.J., 1986. A prospective evaluation of the protective effect
of car seat belts. The Journal of Trauma 26, 561 564.

84

F. Valent et al. / Accident Analysis and Pre6ention 34 (2002) 7184

Petridou, E., Skalkidou, A., Ioannou, N., Trichopoulos, D., 1998.

The Ellenic Road Traffic Police. Fatalities from non-use of seat
belts and helmets in Greece: a nationwide appraisal. Accident
Analysis and Prevention 30, 87 91.
Rivara, F.P., Thompson, D.C., Thompson, R.S., Rogers, L.W.,
Alexander, B., Felix, D., Bergman, A.B., 1994. The Seattle childrens bicycle helmet campaign: changes in helmet use and head
injury admissions. Pediatrics 93, 567 569.
Rowland, J., Rivara, F., Salzberg, P., Soderberg, R., Maier, R.,
Koepsell, T., 1996. Motorcycle helmet use and injury outcome
and hospitalization costs from crashes in Washington State. Journal of Public Health 86, 41 45.
Sacks, J.J., Holmgreen, P., Smith, S.M., Sosin, D.M., 1991. Bicycleassociated head injuries and deaths in the United States from 1984
through 1988. How many are preventable. Journal of the American Medical Association 266, 3016 3018.

Sarkar, S., Peek, C., Kraus, J.F., 1995. Fatal injuries in motorcycle
riders according to helmet use. The Journal of Trauma 38,
242 245.
Thompson, R.S., Rivara, F.P., Thompson, D.C., 1989. A case control study of the effectiveness of bicycle safety helmets. New
England Journal of Medicine 320, 1361 1367.
Valent, F., 1998. Epidemiologia delle cause esterne dei traumatismi e
avvelenamenti in provincia di Udine: mortalita`. Doctoral thesis,
University of Udine, Italy.
Wagle, V.G., Perkins, C., Vallera, A., 1993. Is helmet use beneficial to motorcyclists. The Journal of Trauma 34, 120
122.
Wasserman, R.C., Waller, J.S., Monty, M.J., Emery, A.B., Robinson,
D.R., 1988. Bicyclists, helmets and head injuries: a rider-based
study of helmet use and effectiveness. American Journal of Public
Health 78, 1220 1221.