Safety Science 62 (2014) 114–120

Contents lists available at ScienceDirect

Safety Science
journal homepage: www.elsevier.com/locate/ssci

The relationship between urban street networks and the number of

transport fatalities at the city level
Mehdi Moeinaddini ⇑, Zohreh Asadi-Shekari, Muhammad Zaly Shah
Department of Urban and Regional Planning, Faculty of Built Environment, Universiti Teknologi Malaysia, 81310 Skudai, Johor, Malaysia

a r t i c l e i n f o a b s t r a c t

Article history: There are factors that impact traffic safety and the number of accident-related fatalities, such as street
Received 25 December 2012 users, environment, road design and vehicle characteristics, but there have been limited studies that
Received in revised form 20 July 2013 examine the relationship between street network factors and traffic-related crashes and fatalities at
Accepted 24 August 2013
the city level. Therefore, this paper focused on this relationship by introducing urban street network vari-
Available online 16 September 2013
ables, such as blocks per area, nodes per selected areas and length of roads and motorways, as indepen-
dent variables and the number of fatalities as the dependent variable. This study used Open Street Maps
Keywords:
(OSM) and International Association of Public Transport (UITP) data from 20 cities around the world. The
Urban transportation safety
Street network
number of blocks per area and nodes per selected areas resulted from modifying and analyzing OSM
Crash fatality maps in ArcGIS software. The strength of the relationship in this study was found using generalized linear
Block density modeling (GLM). The findings of this research indicated that increases in fatalities are correlated with an
Nodes per selected areas increasing number of blocks per area, number of nodes per selected areas and length of the motorways.
City level Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction Dee, 1999; Wilkinson, 1987), on traffic fatalities based on within-

country data. Alcohol consumption, location of crash, speed, street
Annually, road traffic crashes cause over 1.2 million deaths and conditions (lighting), time of day and day of the week, hit-and-run
more than 50 million severe injuries worldwide (WHO, 2009). The crashes, pedestrians and driver age are also some important factors
number of traffic crashes has increased in recent decades, espe- that influence pedestrian fatalities in the US (Shankar, 2003).
cially in developing countries, and statistics show that approxi- Permpoonwiwat and Kotrajaras (2012) concluded that policy
mately 70 percent of traffic-related deaths occur in these (government budget for road safety), personal and socio-economic
countries (Augustus, 2012). For example, between 1975 and (health and safety, economic growth, income and education),
1998, the number of road traffic deaths per capita increased over demographic (vehicle kilometers traveled, age and gender), and
200% in Colombia and Botswana and by 44% in Malaysia (Kopits, geographic (population density) factors influenced the fatality
2004). This issue is different in high-income countries. Over the rates for motor vehicle accidents in Thailand in 2006–2009. In In-
same period, traffic fatalities per person decreased by amounts dia, increased enforcement of road traffic rules can decrease road
ranging from 25% to 50% in most European countries and by 60% traffic crash fatalities (Grimm and Treibich, 2012).
in Canada and Hong Kong (Kopits, 2004). The Fatality Risk Generally, accidents involve damage to vehicles and potential
(Deaths/100,000 Persons) is estimated to be 7.8 in 2020 in high- loss of human life, in addition to causing travel costs as a result
income countries, and this rate may be between 14.9 and 31 in of delays in traffic (Hadayeghi, 2002). The causes of motor vehicle
other regions (Kopits, 2004). Therefore, concerns about traffic crashes are multi-faceted and involve the interaction of a number
safety are very serious in developing countries and still exist in of pre-crash variables that include humans, vehicles and road con-
developed countries. ditions (Augustus, 2012). Because there is no single variable that
Ample studies have been devoted to evaluating the effects of will reduce road crashes and fatalities, employing engineering,
specific safety factors, such as speed limits (Eluru et al., 2008; enforcement and education with a focus on both motorized and
Abdel-Aty et al., 2007; McCarthy, 2001; Snyder, 1989), helmet laws non-motorized users is a comprehensive approach.
(Lapparent, 2005; Sass and Zimmerman, 2000), seat belt laws The purpose of this paper was to:
(Evans, 1986; Wagenaar et al., 1988) and alcohol control policies
(Mann et al., 2001; Ruhm, 1996; Whetten-Goldstein et al., 2000; 1. Review the literature concerning macro-level collision predic-
tion models.
2. Estimate the relationship between urban street networks and
⇑ Corresponding author. Tel.: +60 129410543.
the number of transport fatalities across different cities.
E-mail address: mmehdi2@live.utm.my (M. Moeinaddini).

0925-7535/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ssci.2013.08.015
 M. Moeinaddini et al. / Safety Science 62 (2014) 114–120 115

3. Demonstrate the use of macro-level models in a city level plan- Marshall and Garrick (2011) focused on how some aspects of
ning process. overall community design might affect road safety at city level.
The aim in this research was to identify how the characteristics
Only a few efforts have been undertaken to quantify and assess of the street network, such as street network density, street con-
traffic safety impacts in the planning process at the city level nectivity and street network patterns, can have an impact on road
(De Leur and Sayed, 2001). Collision prediction models at the com- safety outcomes. This study concluded that road safety outcomes
munity-based macro-level would improve the road safety analyses are affected by street network characteristics. They proposed a
that are used by engineers and planners (Hadayeghi et al., 2003; spatial analysis of crash data in 24 medium-sized California cities.
Guevara et al., 2004). Some studies have developed macro-level Noland (2003) also found that increases in the total lane miles, the
collision prediction models to provide empirical tools for engineers average number of lanes on collector roads, and the percent of col-
and planners to conduct proactive road safety planning (Hadayeghi lectors with lanes 12 feet or wider was correlated with an increase
et al., 2003; Guevara et al., 2004; Hadayeghi et al., 2010a). In one of in total fatalities in the US.
the early studies on urban form factors and traffic safety, Augustus Some researchers have developed empirical road safety
(2012) proposed a model that used the length of roads (in km), the planning tools and macro-level collision prediction models at
presence of a road safety corps (special traffic police in Nigeria) and neighborhood level (e.g., Hadayeghi et al., 2003; Lovegrove, 2007;
population as independent variables and the number of road traffic Lovegrove and Sayed, 2006a). Hadayeghi et al. (2010b) and
accidents as the dependent variable in Lagos from 1970 to 2001. Hadayeghi (2009) considered different types of data, including
Although this model suggests considering two urban design fac- street networks, land use, traffic demands, socioeconomic and
tors, these indicators are not strong enough to illustrate urban demographic characteristics, dwelling units, and employment, in
street patterns on a macro-level scale at the city level. the city of Toronto to reduce the number of collisions and, conse-
Rifaat et al. (2011) proposed the effect of street patterns on the quently, improve safety. Hadayeghi (2002) proposed that the num-
severity of crashes involving vulnerable road users. In this study, ber of households, major road kilometers, vehicle kilometers
street patterns were classified into four categories: grid-iron, traveled, intersection density, posted speed and volume/speed
warped parallel, loops and lollipops, and mixed patterns (refer to ratio can impact accident occurrence. He proposed models for pre-
Fig. 1). This classification divided the city into different precincts. dicting the number of accidents per zone in the city of Toronto.
The results of this research showed that, compared to other street Wei and Lovegrove (in press) proposed an empirical tool based
patterns, the loops and lollipops design increased the probability of on four groups of data as independent variables including expo-
an injury but reduced the probability of fatalities and property sure, socio-demographics, transportation demand management
damage-only incidents in the event of a crash. However, it should and road network in evaluating the safety of cyclists. These vari-
be considered that the areas of these precincts may affect the re- ables are not usually considered at the micro level (Khondakar
sults regardless of street patterns. et al., 2010). Detailed descriptions of these four groups are given

Fig. 1. Examples of street patterns (Rifaat et al., 2011).

116 M. Moeinaddini et al. / Safety Science 62 (2014) 114–120

in different studies, such as Lovegrove (2007) and Lovegrove and cannot be illustrated in the street network at the city level. These
Sayed (2006a). In addition, Guevara et al. (2004) found that the sig- indicators also cannot represent the shape of real urban street net-
nificant factors of injury and property-damage crash models in works. Because the real shapes of street networks are illustrated in
Tucson, Arizona, are the number of employees, population density, street maps, this research also uses the OSM data source to include
intersection density, percentage of miles of principal arterial, per- indicators of urban street network shape.
centage of miles of minor arterials, and percentage of miles of ur- Because block size and intersection density are important fac-
ban collectors. tors in street networks, connectivity and shape indicators are illus-
Hadayeghi et al. (2006) examined the temporal transferability trated in this study by blocks per area and nodes per selected areas
of the zonal accident prediction models for the city of Toronto in (refer Fig. 2). The number of blocks per area is the number of poly-
1996 and 2001. Hadayeghi et al. (2010a) considered land use, traf- gons that are bounded by street lines in a selected area. UITP data-
fic, demographic, dwelling units, employment types and street net- base defined Metropolitan and central business district for each
works, including number of rail stations, total rail kilometer, city in the database. The selected area is chosen based on UITP
number of schools, total arterial road, expressway, collector, lane- Metropolitan definition. The number of nodes per selected areas
way, local road and ramp kilometers, number of 4-legged and is the number of nodes in a selected area per the number of poly-
3-legged signalized intersections, as independent variables in gons that are bounded by street lines in that area. The number of
developing planning level transportation safety tools using a geo- blocks per area can show the size of the street network blocks,
graphically weighted Poisson regression. Although these studies and the number of nodes per selected areas can introduce the
evaluated different effective factors, they cannot show street net- shape of them. Blocks per area and nodes per selected areas can
work and transport fatality relationships across different cities. be achieved for urban streets by modifying and analyzing OSM
There have also been other efforts to classify street patterns. data.
Lovegrove (2007) proposed that, in residential areas, cul-de-sac OSM data are digital maps of street networks with various lay-
road networks are popular for limiting through-traffic but accessi- ers. Because only the street layer is needed in this research, these
bility for emergency vehicles and transit is significantly less than a maps should be edited in JOSM software (Java open street map).
grid network. Lovegrove and Sayed (2006b) suggested that 3-way JOSM is a free program for modifying OSM maps that is provided
offset road networks are safer than grid networks and cul-de-sac by the OSM web site. This software is user friendly, and the streets
networks. Sun (2009) proposed that the 3-way offset and Fused layer can be selected easily. This selected layer can be saved as a
Grid road networks were the safest road patterns. The 3-way offset new .gpx file, and then this file can be analyzed in ArcGIS software.
pattern also features significantly better road safety and accessibil- There are three different layers in the .gpx file when it is opened in
ity than other street patterns (Ogden, 1996). Wei and Lovegrove ArcGIS. These layers are displayed as streets that are lines, intersec-
(2012) reviewed some studies of neighborhood road patterns that tions and junctions that are nodes and the selected area that is a
can save the lives of vulnerable road users. However, these efforts polygon. Nodes also include the points at which a street changes
cannot illustrate urban street patterns across different cities. direction (mid-block nodes in Fig. 2 are used to show these nodes).
Many researchers have proposed traditional micro-level The streets layer is converted to polygons using Arc Toolbox to cre-
collision prediction models, but less attention has been given to ate polygons that are bounded by street lines. The intersect tool in
the transferability of macro-level collision prediction models Arc Toolbox can provide street polygons in the selected area and
(Khondakar et al., 2010). Although the effects of different factors the number of nodes in that area. Therefore, the block (polygon)
have been explored in previous research, limited research has been density and nodes per selected areas are the result of open street
conducted on exploring the impacts of street network factors on map analysis in ArcGIS software after editing the maps in JOSM
transport fatalities in different contexts (at the city level) with a software. Fig. 2 illustrates the result of converting the OSM data
macro-level scale. As a part of continuing efforts in this area, this to block density and nodes per selected areas.
study examines how these factors, including length of roads, blocks The number of blocks per area can illustrate the size of street
per area, nodes per selected areas and length of motorways, contrib- blocks, and nodes per selected areas show the average road seg-
ute to crash fatalities. This study uses data from 20 cities around the ment density for blocks. Although the type of connection (e.g.,
world to estimate this relationship in different contexts. 4-way, 3-way, interchange, etc.) is not considered in this macro-
level study, more nodes can indicate more routes. Therefore, these
indicators also represent connection density. The length of motor-
2. Materials and methods way per ten thousand inhabitants, the length of road per million
inhabitants and the number of passenger transport fatalities per
This paper focuses on the relationship between street network million inhabitants are collected from UITP.
factors and transport fatalities in different contexts with a There are also data on the length of road and length of motor-
macro-level scale by considering street network variables, such way per urban hectare in the UITP database. Because the depen-
as blocks per area, nodes per selected areas and length of roads dent variable is transport fatalities per inhabitant, the length of
and motorways, as independent variables and the number of pas- road and length of motorway per inhabitant are also selected in or-
senger transport fatalities per million inhabitants as the dependent der to illustrate the relationship between transport fatalities and
variable (refer Table 1). The data that are used in this study are col- street length in the context of population. This issue shows the im-
lected from the mobility in cities database (UITP, 2006) and the pacts of population growth, length of road and length of motorway
Open Street Map (OSM) web site. The UITP database has 120 urban on road safety. In addition, blocks per area and nodes per selected
mobility indicators in 52 cities from various parts of the world, but areas can indicate land development. The roads in this research are
the number of urban street network indicators is limited to 20 cit- different from motorways. The roads can be used by motorized
ies in this database. The cities are chosen based on availability of vehicles, pedestrians and cyclists but the motorways can be used
data for all indicators after removing the outliers (refer Table 1). by motorized vehicles while walking and cycling users do not
In addition, there are limited indicators for street networks in the experience benefits from them. Table 1 presents the data that are
UITP data source. There are only four indicators for street networks used in this study.
at the city level that indicate the length of roads and length of Different methods have been successfully used by previous
motorways in different units (per hectare and per inhabitant). researchers in different approaches to crash studies. Clark and
Other UITP street indicators, such as road investments and speed, Cushing (2004) used linear regression. Linear regression with a
 M. Moeinaddini et al. / Safety Science 62 (2014) 114–120 117

Table 1
Research data.

Cities Blocks per Nodes per Length of road per Length of motor way Passenger transport fatalities
area selected areas million inhabitants (m) per ten thousand inhabitants (m) per million inhabitants
Amsterdam 53.6 6.71 2.75 – 33
Athens – – 2.31 3.9 77
Barcelona 54.97 4.89 2.1 8.97 84.3
Berlin – – 1.57 2.02 19.2
Bern – – 3.92 15.7 34.8
Bilbao – – 4.36 12 54.3
Bologna 18.2 8.55 2.49 15.2 99.1
Brussels – – 1.94 3.63 45.6
Budapest – – 2.43 1.36 63.1
Chicago – – 4.77 9.53 79.7
Clermont Ferrand – – 3.4 4.17 114
Copenhagen 21.96 6.48 3.85 12.5 46.9
Dubai – – 3.1 39.5 203
Dublin – – 4.26 – 47.2
Geneva 26.96 8.26 4.9 10.3 40.5
Ghent – – 5.48 14.6 –
Glasgow 35.26 6.91 5.8 11.1 52.7
Graz 20.03 9.8 4.4 7.96 39.8
Hamburg – – – – 34.5
Helsinki – – 3.61 9.08 20.6
Hong Kong 18.08 9.51 0.284 1.65 30
Krakow – – 1.45 2.24 –
Lille – – 3.48 14.4 24.5
Lisbon 37.61 7.01 0.889 8.68 73.1
London 53.16 8.48 2.03 0.99 41.8
Lyons 34.96 5.11 2.47 6.81 22.1
Madrid 59.48 5.43 4.87 9.83 71
Manchester – – 3.7 7.09 42.2
Marseilles 23.14 8.52 1.63 4.13 62.5
Melbourne – – – – 77.4
Milan – – – – –
Moscow – – 0.406 1.14 47.1
Munich 52.5 6.3 1.83 4.78 22.3
Nantes 32.16 10.44 5.41 12.3 64.9
Newcastle – – 4.12 12.9 23.2
Oslo – – 5.86 17 38.7
Paris 46.12 6.12 1.98 6.95 65.9
Prague – – 2.91 6.44 57.5
Rome 58.03 6.89 2.8 11.1 108
Rotterdam 62.11 5.88 4.07 – 33.9
Sao Paulo – – 1.96 2.02 109
Seville 52.55 5.8 2.02 11.6 103
Singapore – – 0.94 4.52 58.4
Stockholm 38.08 7.83 – 26.9 36.4
Stuttgart 29.06 11.77 1.19 4.92 57.2
Tallinn – – 2.17 2.51 65.2
Tunis – – – – –
Turin 35.09 4.24 2.71 17.2 97.9
Valencia – – 2.87 2.87 33.8
Vienna 46.37 5.77 1.81 2.64 26.4
Warsaw 20.38 8.77 1.68 4.5 –
Zurich – – 4.7 10.9 44.6

spatial lag parameter was used by Levine et al. (1995). Chatterjee regression models with a normally distributed error structure are
et al. (2004) utilized Poisson models. Negative binomial (NB) mod- not suitable for modeling road collisions. Poisson and NB regres-
els were also used by numerous studies (e.g., Lovegrove and Sayed, sions in the generalized linear model (GLM) framework have been
2006b; Wei and Lovegrove, in press; Abdel-Aty and Radwan, 2000; more successfully adopted for this issue (Hadayeghi, 2009). Kim
Hadayeghi et al., 2003 and Hadayeghi et al., 2007; Maher and et al. (2002) found that Poisson Gamma and Poisson lognormal
Summersgill, 1996; Miaou and Lord, 2003; Mitra and Washington, models can be more successful for the prediction of collisions.
2007; Noland and Quddus, 2004; Memon, 2006). Zero-inflated The Poisson distribution has been shown to be suitable for model-
Poisson and NB models (ZIP/ZINB) were used by Qin et al. (2004) ing crash data for a single site, but crash data for different sites of-
and Shankar et al. (1997). Guevara et al. (2004) utilized simulta- ten exhibit large variances and small means that lead to
neous estimation of NB models. Full Bayes hierarchical models overdispersion when there is a variance-to-mean ratio greater than
with spatial effects were used by Aguero-Valverde and Jovanis 1 (Hadayeghi, 2009). Therefore, the NB distribution, also known as
(2006) and Quddus (2008). Hadayeghi et al. (2003) proposed a geo- the Poisson–gamma distribution, has become the most commonly
graphically weighted regression (GWR). Log-linear models using used method for modeling crash data at a series of sites
ordinary least squares (OLS) estimation were used by Washington (Hadayeghi, 2009). In previous collision prediction models studies,
et al. (2006) and Wier et al. (2009). GLMs were commonly used. They have been successful in inter-
Road collisions are random and rare events. Moreover, count preting random, rare, sporadic, and non-negative collision data
data are non-negative and discrete. Therefore, conventional linear (Lovegrove et al., 2010; Wei and Lovegrove, in press).
118 M. Moeinaddini et al. / Safety Science 62 (2014) 114–120

Table 5
Parameter estimates.

Parameter B Std. 95% Wald Hypothesis test

error confidence
interval
Lower Upper Wald df Sig.
Chi-
Square
(Intercept) 1.922 .5343 .874 2.969 12.933 1 .000
Blocks per area .018 .0054 .008 .029 11.329 1 .001
Nodes per .114 .0429 .030 .198 7.058 1 .008
selected
areas
Length of road .140 .0453 .229 .051 9.547 1 .002
per
million
inhabitants
Fig. 2. An illustration of blocks and nodes. Length of .114 .0181 .079 .150 39.563 1 .000
motor
way per ten
thousand
Table 2
inhabitants
Continuous variable information.
(Scale) .068a .0214 .037 .126
Indicators N Minimum Maximum Mean Std.
Dependent variable: passenger transport fatalities per million inhabitants.
deviation
Model: (Intercept), blocks per area, nodes per selected areas, length of road per
Passenger transport 20 22 108 60.47 27.435 million inhabitants, length of motor way per ten thousand inhabitants.
fatalities per million a
Maximum likelihood estimate.
inhabitants
Blocks per area 20 18.1 59.5 37.785 14.1208
Nodes per selected areas 20 4.24 11.77 7.3140 2.02433 Table 6
Length of road per 20 .28 5.80 2.7732 1.56602 Omnibus testa.
million inhabitants
Length of motor way per 20 .99 17.20 8.4805 4.40316 Likelihood ratio Chi-Square df Sig.
ten thousand 23.742 4 .000
inhabitants
Dependent variable: passenger transport fatalities per million inhabitants.
Model: (Intercept), blocks per area, nodes per selected areas, length of road per
million inhabitants, length of motor way per ten thousand inhabitants.
a
Compares the fitted model against the intercept-only model.
Table 3
Case processing summary.
Table 7
N Percent (%)
SD and Pearson Chi-Square goodness of fit.
Included 20 38.5
Excluded 32 61.5 Value df Value/df
Total 52 100.0 Deviance 1.382 15 .092
Scaled deviance 20.228 15
Pearson Chi-Square 1.356 15 .090
Scaled Pearson Chi-Square 19.851 15
Table 4
Dependent variable: passenger transport fatalities per million inhabitants.
Correlation matrix.
Model: (Intercept), blocks per area, nodes per selected areas, length of road per
Pearson correlation R M P N million inhabitants, length of motor way per ten thousand inhabitants.

R 1 .510 .052 .012

M .510 1 .101 .276
P .052 .101 1 .554 the NB model, the Gamma model assumes a log link (refer to Eq.
N .012 .276 .554 1 (1)):
R = length of road per million inhabitants. !
M = length of motorway per ten thousand inhabitants.
X
i
Y ¼ EXP b0 þ bi  X i ð1Þ
P = blocks per area.
i¼1
N = nodes per selected areas.

where Y is the dependent variable, i the subscript showing the num-

ber of independent variables, X the independent variable, b0 the
In this study, the dependent variable is the number of passenger constant, calculated in the calibration process and bi is the coeffi-
transport fatalities per million inhabitants. It is considered to be cient of the independent variable, calculated in the calibration pro-
similar to count data, where the number of deaths only takes posi- cess of the model.
tive and discrete values. On the other hand, this variable is a scaled
factor because it is divided by inhabitants. Lognormal and Gamma 3. Results
with log link models are used to scale data in the GLM framework.
Lognormal is not in the exponential family (e.g., Poisson, Gamma, In this section, the results of the street network GLM analysis
Binominal, Inverse, Negative Binominal, etc.). Therefore, GLMs are discussed. Tables 2 and 3 indicate descriptive statistics for
with a gamma distribution are recommended because they per- the variables included in the model. Table 4 indicates that there
formed slightly better than the log transformed model. Similar to is no strong correlation between the independent variables
 M. Moeinaddini et al. / Safety Science 62 (2014) 114–120 119

included in the model. Table 5 indicates the variables included in illustrate this relationship regardless of context effects. In addition,
the model, their parameter estimates, and the significance of the this study also attempts to use some factors, such as nodes and
parameters (5% level). The Omnibus test, likelihood ratio Chi- blocks, to describe the shapes of the street networks at a macro
Square test statistics, the scaled deviance (SD) and the Pearson scale. Moreover, it attempts to address street network indicators
Chi-Square statistic show the model goodness of fit (refer to Tables at the city level (instead of the neighborhood level) that have not
6 and 7). Therefore, the final model can be defined as follows: been sufficiently studied in the literature.
Currently, more sustainable urban areas are needed. To pro-
PF ¼ EXPð1:992  0:140R þ 0:018P þ 0:114N þ 0:114MÞ ð2Þ mote these kinds of areas, having fewer traffic-related fatalities is
an important goal. Therefore, this research tries to indicate the
where PF is the passenger transport fatalities per million inhabit-
relationship between urban street indicators and transport fatali-
ants, R the length of road per million inhabitants, P the blocks per
ties. This relationship can be used in fatality reduction strategies
area, N the nodes per selected areas and M is the length of motor-
in different cities. In addition, because the urban street network
way per ten thousand inhabitants
is the result of macro-scale planning decisions, considering the
This model shows that passenger transport fatalities in this
relationship of urban street indicators and fatalities can lead to
model are significantly affected by street network indicators.
better planning decisions for new cities.
Among these indicators, the number of nodes per selected areas
This research is limited to selected UITP and open street map
and the length of motorway per inhabitant have higher positive
web site data in selected cities. However, the universal scale rela-
parameters, so these indicators have greater effects on passenger
tionship models may be replicated by future studies for other ur-
transport fatalities in this model. The third effective indicator with
ban structure indicators in order to identify effective factors for
a positive relationship is blocks per area. The length of road per
sustainable travel patterns regardless of context effects. In addi-
inhabitant has negative parameter and thus it is inversely associ-
tion, this research uses OSM data sources that are free editable
ated with transport fatalities in this model.
maps because we had financial limitations. Further studies may
Overall, fewer nodes per selected areas and blocks per area,
enhance the model by using the original versions of urban maps.
shorter motorways per inhabitant and more roads per inhabitants
Moreover, urban structure indicators change rapidly; future stud-
are effective urban street network indicators that are associated
ies may evaluate these changes by updating their data sources
with fewer transport fatalities. This suggests that macro-scale
and examining the relationship between urban structure indicators
street network planning that reduces block density, nodes per se-
and fatalities in various parts of the world to address the changing
lected areas and motorway density and increases road density
results.
could contribute to fewer transport fatalities across different cities.

4. Discussion and conclusions

References
Macro-scale urban street network indicators as independent Abdel-Aty, M., Chundi, S., Lee, C., 2007. Geo-spatial and log-linear analysis of
variables and passenger transport fatalities as the dependent vari- pedestrian and bicyclist crashes involving school-aged children. J. Saf. Res. 38,
able are analyzed in this research using GLM analysis. From this 571–579.
Abdel-Aty, M.A., Radwan, A.E., 2000. Modeling traffic accident occurrence and
study, it can be concluded that the length of road per million involvement. Accident Anal. Prevent. 32, 633–642.
inhabitants, blocks per area, nodes per selected areas and length Aguero-Valverde, J., Jovanis, P.P., 2006. Spatial analysis of fatal and injury crashes in
of motorway per ten thousand inhabitants are the urban street net- Pennsylvania. Accident Anal. Prevent. 38, 618–625.
Augustus, A., 2012. Determinants of road traffic accident occurrences in Lagos state:
work factors that are associated with passenger transport fatalities. some lessons for Nigeria. Int. J. Humanities Soc. Sci. 2, 252–259.
The model shows that more blocks per area are correlated with Chatterjee, A., Everett, J.D., Reiff, B., Schwetz, T.C., Seaver, W.L., Wegmann, F.J., 2004.
more passenger transport fatalities. Distributing more blocks in ur- Tools for assessing safety impact of long-range transportation plans in urban
areas, In: 80th Annual Meeting of the Transportation Research Board, TRB,
ban areas can produce more nodes. These nodes produce more Washington, DC.
conflicts. On the other hand, more and smaller blocks create more Clark, D.E., Cushing, B.M., 2004. Rural and urban traffic fatalities, vehicle-miles and
connectivity, and this arrangement produces more direct routes for population density. Accident Anal. Prevent. 36, 967–972.
Dee, T.S., 1999. State alcohol policies, teen drinking, and traffic fatalities. J. Publ.
alternative travel modes. However, the model shows that more
Econ. 72, 289–315.
blocks are also correlated with more passenger transport fatalities De Leur, P., Sayed, T., 2001. Using claims prediction model for road safety
involving vehicle transport. This suggests that considering more evaluation. Can. J. Civ. Eng. 28, 804–812.
connected routes to be used exclusively by non-motorized modes Eluru, N., Bhat, C., Hensher, D., 2008. A mixed generalized ordered response model
for examining pedestrian and bicyclist injury severity level in traffic crashes.
is more effective at allowing fewer conflicts and transport fatalities Accident Anal. Prevent. 40, 1033–1054.
involving vehicle transport than creating more blocks and nodes to Evans, L., 1986. The effectiveness of safety belts in preventing fatalities. Accident
increase connectivity for all modes. In addition, the model shows Anal. Prevent. 18, 229–241.
Grimm, M., Treibich, C., 2012. Determinants of Road Traffic Crash Fatalities across
that more nodes per selected areas are associated with more pas- Indian States. Health Economics.
senger transport fatalities. Consequently, more simple, large blocks Guevara, F.L.D., Washington, S.P., Oh, J., 2004. Forecasting crashes at the planning
with fewer nodes are correlated with fewer fatalities. level: simultaneous negative binomial crash model applied in Tucson, Arizona.
Transp. Res. Rec. 1897, 191–199.
The model also indicates that longer motorway per inhabitant is Hadayeghi, A., 2002. Accident Prediction Models for Safety Evaluation of Urban
correlated with more passenger transport fatalities, while road Transportation Network. Department of civil engineering. University of
length has a negative relationship with fatalities. More motorways Toronto, Ontario, Canada.
Hadayeghi, A., 2009. Use of Advanced Techniques to Estimate Zonal Level Safety
can increase convenience for private motorized vehicle users. This Planning Models and Examine Their Temporal Transferability. Department of
type of road belongs to motorized vehicles users with high speed, Civil Engineering. University of Toronto, Ontario, Canada.
while walking and cycling users do not experience benefits from Hadayeghi, A., Shalaby, A., Persaud, B., 2003. Macro-level accident prediction
models for evaluating the safety of urban transportation systems. Transp. Res.
them. Therefore, streets patterns with more facilities, such as
Rec. 1840, 87–95.
motorways for high speed car users, do not contribute to fewer Hadayeghi, A., Shalaby, A., Pesuad, B., 2007. Safety prediction models: a proactive
fatalities. tool for safety evaluation in urban transportation planning applications. Transp.
This study evaluates the relationship of urban street indicators Res. Rec. 2019, 225–236.
Hadayeghi, A., Shalaby, A., Persaud, B., 2010a. Development of planning level
and passenger transport fatalities in different cities within various transportation safety tools using geographically weighted poisson regression.
contexts at the city level. Therefore, the results of this study can Accident Anal. Prevent. 42, 676–688.
120 M. Moeinaddini et al. / Safety Science 62 (2014) 114–120

Hadayeghi, A., Shalaby, A., Persaud, B., 2010b. Development of planning-level Permpoonwiwat, C.K., Kotrajaras, P., 2012. Pooled time-series analysis on traffic
transportation safety models using full bayesian semiparametric additive fatalities in Thailand. World Rev. Business Res. 2, 170–182.
techniques. J. Transp. Saf. Security 2, 45–68. Qin, X., Ivan, J.N., Ravishanker, N., 2004. Selecting exposure measures in crash rate
Hadayeghi, A., Shalaby, A., Persaud, B., Cheung, C., 2006. Temporal transferability prediction for two-lane highway segments. Accident Anal. Prevent. 36, 183–
and updating of zonal-level accident prediction models. Accident Anal. Prevent. 191.
38, 579–589. Quddus, M.A., 2008. Modelling area-wide count outcomes with spatial correlation
Khondakar, B., Sayed, T., Lovegrove, G., 2010. Transferability of community-based and heterogeneity: an analysis of London crash data. Accident Anal. Prevent. 40,
collision prediction models for use in road safety planning applications. J. 1486–1497.
Transp. Eng. 136, 871–880. Rifaat, S.M., Tay, R., De.Barros, A., 2011. Effect of street pattern on the severity of
Kim, H., Sun, D., Tsutakawa, R.K., 2002. Lognormal vs. gamma: extra variations. crashes involving vulnerable road users. Accident Anal. Prevent. 43, 276–283.
Biometrical J. 44, 305–323. Ruhm, C.J., 1996. Alcohol policies and highway vehicle fatalities. J. Health Econ. 15,
Kopits, E.-A., 2004. Traffic Fatalities and Economic Growth. Department of 435–454.
Economics. University of Maryland, College Park. Sass, T.R., Zimmerman, P.R., 2000. Motorcycle helmet laws and motorcyclist
Lapparent, M., 2005. Individual cyclists’ probability distributions of severe/fatal fatalities. J. Regul. Econ. 18, 195–215.
crashes in large French urban areas. Accident Anal. Prevent. 37, 1086–1092. Shankar, U., 2003. Pedestrian Roadway Fatalities. National Center for Statistics and
Levine, N., Kim, K.E., Nitz, L.H., 1995. Spatial analysis of Honolulu motor vehicle Analysis Advanced Research and Analysis. U.S. Department of Transportation
crashes. II. Zonal generators. Accident Anal. Prevent. 27, 675–685. National Highway Traffic Safety Administration.
Lovegrove, G., 2007. Road Safety Planning, New Tools for Sustainable Road Safety Shankar, V., Milton, J., Mannering, F., 1997. Modeling accident frequencies as
and Community Development. Verlag Dr, Müller, Germany. zeroaltered probability processes: an empirical inquiry. Accident Anal. Prevent.
Lovegrove, G., Lim, C., Sayed, T., 2010. Community-based, macro-level collision 29, 829–837.
prediction model use with a regional transportation plan. J. Transp. Eng. 136, Snyder, D., 1989. Speeding, coordination, and the 55 MPH speed limit: comment.
120–128. Am. Econ. Rev. 79, 922–925.
Lovegrove, G., Sayed, T., 2006a. Macro-level collision prediction models for Sun, J., 2009. Sustainable Road Safety: Development, Transference and Application
evaluating neighbourhood traffic safety. Can. J. Civ. Eng. 33, 609–621. of Community-Based Macro-Level Collision Prediction Models, Civil
Lovegrove, G., Sayed, T., 2006b. Using macro-level collision prediction models in Engineering. University of British Columbia, Okanagan.
road safety planning applications. Transp. Res. Rec.: J. Transp. Res. Board 1950, UITP, 2006. Mobility in cities Database. International Association of Public
73–82. Transport (UITP).
Maher, M.J., Summersgill, I., 1996. A comprehensive methodology for the fitting of Wagenaar, A., Maybee, R., Sullivan, K., 1988. Mandatory seat belt laws in eight
predictive accident models. Accident Anal. Prevent. 28, 281–296. states: a time-series evaluation. J. Saf. Res. 19, 51–70.
Mann, R.E., MacDonald, S., Stoduto, G., Bondy, S., Jonah, B., Shaikh, A., 2001. The Washington, S., Schalwyk, I.V., Meyer, M., Dumbaugh, E., Zoll, M., 2006.
effects of introducing or lowering legal per se blood alcohol limits for driving: Incorporating Safety into Long-Range Transportation Planning. NCHRP Report
an international review. Accident Anal. Prevent. 33, 569–583. 546, TRB, Washington, DC.
Marshall, W.E., Garrick, N.W., 2011. Does street network design affect traffic safety? Wei, F., Lovegrove, G., (in press). An empirical tool to evaluate the safety of cyclists:
Accident Anal. Prevent. 43, 769–781. community based, macro-level collision prediction models using negative
McCarthy, P., 2001. Effect of speed limits on speed distributions and highway binomial regression. Accident Analysis and Prevention.
safety: a survey of recent literature. Transp. Rev. 21, 31–50. Wei, V.F., Lovegrove, G., 2012. Sustainable road safety: a new (?) neighbourhood
Memon, A.Q., 2006. Road Accident Prediction Models Developed from a National road pattern that saves VRU lives. Accident Anal. Prevent. 44, 140–148.
Database: Poisson and Negative Binomial Regressions, In: Transportation Whetten-Goldstein, K., Sloan, F., Stout, E., Liang, L., 2000. Civil liability, criminal law,
Research Board 85th Annual Meeting CD-ROM Washington DC, United States. and other policies and alcohol-related motor vehicle fatalities in the United
Miaou, S.P., Lord, D., 2003. Modeling traffic crash-flow relationships for States: 1984–1995. Accident Anal. Prevent. 32, 723–733.
intersections: dispersion parameter, functional form, and Bayes versus Wier, M., Weintraub, J., Humphreys, E.H., Seto, E., Bhatia, R., 2009. An area-level
empirical Bayes. Transp. Res. Rec. 1840, 31–40. model of vehicle-pedestrian injury collisions with implications for land use and
Mitra, S., Washington, S., 2007. On the nature of over-dispersion in motor vehicle transportation planning. Accident Anal. Prevent. 41, 137–145.
crash prediction models. Accident Anal. Prevent. 39, 459–468. Wilkinson, J.T., 1987. Reducing drunken driving: which policies are most effective?
Noland, R.B., 2003. Traffic fatalities and injuries: the effect of changes in Southern Econ. J. 54, 322–334.
infrastructure and other trends. Accident Anal. Prevent. 35, 599–611. World Health Organization (WHO), 2009. Global Status Report on Road Safety. Time
Noland, R.B., Quddus, M.A., 2004. Improvements in medical care and technology for Action. World Health Organization, Geneva.
and reductions in traffic-related fatalities in Great Britain. Accident Anal.
Prevent. 36, 103–113.
Ogden, K.W., 1996. Safer Roads: A Guide to Road Safety Engineering. Avebury
Technical.