Abstract:
Background
Spatial organization in cities has often taken a special attention to motor vehicles
requirements, and neglecting pedestrians and cyclists’ needs. Current emphasis in city
planning for health and sustainable development has led to shifting local transport policies
emphasis to the promotion of active transport modes and to the reassignment of urban space
to these modes. However, to increase the share of active transport modes while meeting road
safety policy means improvements in the safety of vulnerable road users are needed.
Accident occurrence in urban environments represents a critical safety issue for all countries in
the world, as the opportunities for crashes involving pedestrians or cyclists and motorized
vehicles increases with urbanization growth and those occurrences tend to involve severe
injuries. According to the CARE database (the European Union’s injury road accident database),
in 2015 21% of European road accident fatalities were pedestrians, of which 69% occurred
inside urban areas and only 31% occurred in rural areas.
In the same year, in Portugal pedestrian fatalities corresponded to 24% of all road accident
deaths, a percentage that is slightly below the corresponding European value. However, when
analysed in detail, 76% of these pedestrian fatalities occurred in urban areas, which highlights
a higher relative importance of the pedestrian road safety challenges in Portugual than in
Europe.
A safe environment is indispensable to promote walking and cycling. However, motorized
vehicle priority urbanization created a built environment unfriendly for pedestrians, making
walking activities too much vulnerable, even in developed countries.
Improved knowledge on the underlying factors involved in crash occurrence and elements
influencing the severity of resulting injuries are needed to improve pedestrian safety. The
consideration of variables describing build environment in explaining crashes and injury
outcomes may help to improve urban planning and street environment design. Several studies
have examined the relationship between built environment factors and pedestrian crash
frequency and risk.
Aim
This study aimed at investigating factors associated with injury severity levels that pedestrians
experienced in the city of Lisbon, Portugal, such as urban infrastructure, population and other
exposure indicators, and urban characteristics. To accomplish this analysis, a geocoded
database on road accidents and victims that occurred in Lisbon between 2008 and 2011 was
used. The analysis was conducted using the multinomial logit (MNL) model to estimate
pedestrian and driver injury outcomes, by severity level.
Method or methodological issues
In crash severity analysis, several models can be applied; MNL being a frequently used one.
MNL models are traditional discrete outcome models that may consider several outcome
levels and that do not explicitly consider the ordering that may be present in these outcomes.
These models require the assumption that the unobserved terms are independent of the injury
severity level. If there are injury severity levels that share unobserved terms (so being
correlated), coefficients and severity probabilities would be erroneously estimated; in this
cases, these models should not be used.
The data collected for this study concerns Lisbon municipality. This study uses four data sets,
namely pedestrian crash data, land use information, population census data, and pedestrian
exposure proxies. Land use information is disaggregated by area type: industrial, green,
residential, historical, services, special uses, mixed-use, railways, roads, and buildings. Census
data includes the number of housing units, number of families, and number of inhabitants per
gender, age group, and main activity. Finally, pedestrian exposure proxy data were calculated
using a numeric scale conversion of the pedestrian potential maps (MAPPe) developed within
the Pedestrian Accessibility Plan of Lisbon.
Results obtained or expected
A MNL model was fitted to identify the possible street geometric, road user, environmental,
vehicle, and land use predictors of pedestrian injury severity in Lisbon. In the analysis, data
were used, from 2006 pedestrian crashes that occurred in Lisbon in the period 2008 to 2011.
Altogether, 19 variables were calibrated and used to identify the potential effects of different
factors related to the categories listed above.