Pedestrian injury is second only to cancer as the leading
cause of death of children between the ages of five and nine years
1,2
. Furthermore, pedestrian injury accounts for 31% to 61% of all admissions
of children to the hospital for treatment of injuries
3-5
. In the United States, rates of pedestrian injuries among children
are estimated to be 111 per 100,000 and account for approximately
18,000 hospital admissions each year
6-8
. In 1996, 1191 children under the age of sixteen were killed in pedestrian-motor
vehicle collisions
9
. Operative intervention is required after approximately 11% of
pedestrian injuries, and hospital admission is required after 36%
10,11
. Pedestrian injuries are more likely to be severe than are injuries
sustained by motor-vehicle occupants, and they are associated with
a higher rate of mortality (4% compared with 0.5%)
12
. Mortality is caused by multisystem trauma (80%) or isolated head
injury (20%)
13
.
There are three phases of injury in a pedestrian-motor vehicle collision
14
. The initial impact from the bumper often results in lower-extremity
injuries. Next, head and torso injuries are sustained as the victim
strikes the hood or windshield. Finally, additional head, torso,
and upper-extremity injuries occur as the child falls to the ground.
There is a high incidence of head injury with attendant increases
in morbidity and mortality
15,16
. In one series of more than 1900 children who had been struck by a
motor vehicle, 31% had intracranial injuries, 14% had lower-extremity
fractures, and 2% had pelvic fractures
10
. Given the height of a bumper strike, children under three years of
age sustain femoral fractures twice as often as they sustain tibial
or fibular fractures, whereas lower-extremity injuries in older patients
are most often tibial and fibular fractures
17
. Importantly, pedestrian-motor vehicle collisions are often preventable.
Decreasing pediatric pedestrian morbidity and mortality would have
major social and economic benefits, given the young age of the patients
and the long-term morbidity of some musculoskeletal injuries.
In 1994, we examined the frequency and location of pediatric pedestrian
injuries in New Haven, Connecticut, over an eighteen-month period.
Six years later, we reexamined these injury rates. In the interim,
five interventions were implemented that may account for the substantial decreases
in pediatric pedestrian injuries that we found.
Data on pediatric pedestrian injuries were accumulated from two
sources, the New Haven Police Department accident database and the
Children's Hospital at Yale-New Haven Hospital Emergency Department
records. Both sources were used to obtain a complete sample, as
some pedestrians struck by a motor vehicle are brought to the emergency
room without a police report being filed, and conversely, some cases
are reported to the police, but medical attention is not required
or sought. The population was defined as individuals with an age
of less than twenty years. Data were initially collected for the
eighteen-month time-period from June 1992 through December 1993.
Six years later, exactly the same protocol was followed to collect
information on pedestrian injuries during an equivalent eighteen-month
period from June 1998 through December 1999. An eighteen-month period was
chosen to increase the sample size and to reduce the potential for
inherent periodic variation.
Police department data:
Fields available from this database included the date, time, location
(nearest cross street), person's name, and date of birth. Police
department records differentiate a pedestrian-motor vehicle collision
from a bicycle-motor vehicle collision. Only data on pedestrian
incidents were included. The sex of the injured person was inferred
from the patient's name. There were no changes in the amount, type,
or procedures of data collection or data entry between the two-time periods
studied. Additionally, to examine year-to-year variability and the
value of interventions over time, police records were also reviewed
retrospectively to identify all motor vehicle-pedestrian collisions
(adult and pediatric) in the years 1993 through 1999.
Hospital data:
Emergency department triage logs from the Children's Hospital at
Yale-New Haven Hospital and data from its trauma registry were examined
retrospectively for each day during the two eighteen-month time-periods
to identify pediatric pedestrian injuries. In New Haven, all children requiring
treatment for a traumatic injury are taken to the Children's Hospital
at Yale-New Haven Hospital according to emergency medical service
protocol. This was confirmed by a review of the records of the other community
hospital in the city. The charts of all patients who were suspected
to have been injured in a motor vehicle-pedestrian collision were
reviewed. Detailed information, including the time, date, and location
of the incident and the age and sex of the patient, was recorded
from the hospital charts. Patients were excluded from the study
if the collision had occurred outside the city of New Haven, if
they had been riding a bicycle at the time of the incident, if their charts
did not identify the location of the collision, and if the location
of the accident had not also been identified in police records.
There were no changes in the type of triage logs kept or the level
of detail of hospital records between the two-time periods of interest.
Environmental, educational, and structural change data:
The city's planning department provided information on overall city
population. Changes in the pediatric population were derived from
public school enrollment figures from the local school board as
well as census data. Statistics on changes in traffic laws, traffic
volume, and structural improvements were provided by the city's
department of traffic and parking. data on public housing projects
were provided by the department of public housing. Information on average
incomes was obtained from the Census Bureau and the State of Connecticut's
Department of Economic Development. The New Haven Police Department
provided information on changes in police enforcement. The Dattco
Bus Company and the New Haven School District provided facts on
school busing and traffic safety education.
Locations of pedestrian-motor vehicle collisions were mapped to
one of twenty-six census tracts and correlated with census demographics.
These demographics included population density, race, and income
and were obtained from the latest census data available (1990).
On the basis of the number of collisions occurring in each, the
census tracts were defined as low-frequency (zero to six collisions),
moderate-frequency (seven to eleven collisions), or high-frequency
(twelve or more collisions). The 1992-1993 data were used exclusively
for this analysis, as a different definition of a high-frequency
tract would have been required for the 1998-1999 data to accommodate the
overall drop in collision rates, making interyear comparisons invalid.
The demographic variables for the high, moderate, and low-frequency
tracts were compared with analysis of variance, with alpha = 0.05.
Pedestrian Injuries
There were 223 pediatric pedestrian-motor vehicle collisions
in the eighteen-month time-period from June 1992 through December
1993. Although this number is much higher than the 111 per 100,000
that is the United States mean, New Haven is likely to have more
risk factors for pediatric pedestrian-motor vehicle collisions than
the average United States city. Over the eighteen-month time-period
from June 1998 through December 1999, eighty-seven collisions were recorded.
Children between the ages of five and nine years were most at risk,
with 35% of the incidents in the 1992-1993 period and 31% of those
in the 1998-1999 period occurring in this age-group. Males were
more commonly injured than were females, by a ratio of approximately
2:1 (
Table I
). The fewest number of collisions (8%) occurred during the winter
months, whereas spring, summer, and fall had similar collision rates.
Most collisions (53% and 54%) occurred during the afternoon, between
3 pm and 8 pm (
Table II
). The temporal distribution of collisions was similar between the
1992-1993 and 1998-1999 data sets.
The location of each incident was mapped to one of twenty-six census
tracts. The rate of collisions was higher in tracts with (1) a high
child population density (p < 0.001), (2) a high overall population
density as defined by the percentage of households with more than
one person per room (p < 0.03), (3) a high number of occupied
housing units per acre (p < 0.04), and (4) a high percentage
of nonwhites (p < 0.02). Median income and the percentage of
families living below the poverty line were not associated with
collision frequency (
Table III
).
A review of police records of all motor vehicle-pedestrian collisions
(adult and pediatric) in each year revealed a drop from 281 in 1996
to eighty-nine in 1999. Child pedestrian-motor vehicle collisions
decreased from 118 in 1993 to forty-three in 1999 (
Table IV
).
Changes in City Statistics
A number of factors that traditionally affect the frequency of pediatric
pedestrian injuries had not changed meaningfully from 1992 to 1999.
These included population, traffic volume, speed limits, laws allowing
right turns at red lights, parking regulations, public parks, one-way
streets, crosswalks, walk signals, and crossing guards
2,18-21
.
There was no change in population statistics. On the basis of 1990
census data and other estimates, the 1992 population of the city
was between 125,000 and 130,000 people. The 1998 population was
estimated to be 130,000 people. A 1999 population estimate was not
available at the time of writing. On the basis of public school
enrollments, there was a small increase in the number of school-age
children between 1992 (18,418 children) and 1999 (19,883). Average
family income, adjusted for inflation, increased 7% from 1993 to 1999.
Traffic volume increased approximately 3% from 1992 to 1999. There
were no changes in speed limits or in parking allowed on the sides
of streets. One street that had been one-way for a one-block section
was made two-way. Interestingly, there were no pedestrian-motor
vehicle collisions on this street in 1992, when it was one-way,
but there were two such collisions in 1999, after conversion to
two-way traffic. (One-way streets have been shown to reduce pedestrian-motor
vehicle collisions
22
.) No new parks or playgrounds had been constructed. There had been
few changes in the number or nature of crosswalks, walk signals,
or crossing guards.
There were five interventions between the two data-collection periods
that may account for some of the changes observed: (1) improved
traffic safety education for children, (2) increased school busing,
(3) a public relations campaign to promote safe driving, (4) decentralization
of public housing projects leading to a decrease in child population density,
and (5) an increase in traffic-related tickets issued by the police
department.
Although many factors known to influence the frequency of pedestrian-motor
vehicle collisions did not change between the two time-periods of
the study, the five interventions noted above may account for the decrease
in the frequency of collisions. Unfortunately, because of the dynamic
environment of the city, it is difficult to isolate the effects
of any single intervention. There may have been other factors, such
as variations in the local economy and even differences in local
weather patterns, that affected children's exposure to injury by
motor vehicles.
Traffic safety education was completely transformed in the New
Haven public school system. Prior to 1991, traffic safety education
was not a formal part of the curriculum and it occurred sporadically
at best. Beginning in 1991, traffic safety education was provided
twice a year for every elementary schoolchild who rode a bus to
school (approximately 30% of all schoolchildren). In 1995, this education
was expanded to include all grade levels.
School busing increased greatly in the 1990s as schools were integrated
citywide and magnet schools were created. In 1992, approximately
35% of public school children rode a bus to school compared with
73% in 1999. Additionally, there was an increase in the number of
door-to-door pick-ups as opposed to the use of bus stops at which
a large number of children were picked up. We speculate that this most
likely had two effects on pedestrian safety. First, the risk of
children being hit by a motor vehicle is associated with the number
of streets that they have to cross
23
. An increase in busing and in door-to-door pickups substantially
decreases the number of streets crossed by a child. Second, when
a child who previously had a relatively short walk to school begins
to ride the bus instead, he or she may get home approximately thirty
to forty-five minutes later, decreasing the number of daylight hours
of exposure to traffic hazards.
In 1996, the city sponsored an initiative to promote safe driving. This
campaign included flyers in public areas, mailings, and billboards.
A revised version, called "Drive Smart�Do Your Part," was launched
in 1999 with a similar mechanism of message distribution.
New Haven began decentralizing its public housing in 1990. The largest
complex was gradually vacated and was demolished in 1999. Some residents
were transferred to other projects, and others were given vouchers
for subsidized housing in the private rental market. This follows
similar nationwide trends of decentralization in order to reduce
the concentration of poverty and associated crime. In the 1992-1993
time-period, an intersection adjacent to the large housing project
had more pedestrian injuries (five) than any other single location,
whereas there were no pediatric pedestrian-motor vehicle collisions
at this intersection during the 1998-1999 time-period. Clearly,
the local decrease in population may account for this decline in
collisions. In the first portion of this study, we found a significant relationship
between child population density and pedestrian injuries. However,
after New Haven changed from a strategy of concentrated public housing to
one of more even distribution of low-income families around the
city, the overall incidence of pedestrian injuries has remained low.
Beginning in 1999, police officers, as part of a department-wide initiative,
were encouraged to increase the number of traffic tickets and warnings
that they issued. Police resources are concentrated at known high-risk areas
during times associated with a high frequency of collisions. The
number of tickets issued in the first five months of 1999 increased
22% compared with that in the previous year. The effects of this
campaign may or may not be a causal factor in the decreased number
of pedestrian-motor vehicle collisions, since it was started relatively
recently.
This study had a few weaknesses, including an inability to link
cause and effect with certainty and only two eighteen-month periods
of detailed review. Additional studies are necessary to determine
if the improvements observed are lasting and to help identify the
interventions that were most beneficial. The size of the decrease
in pedestrian-motor vehicle collision rates well exceeded the expectations
of community leaders and the authors of this study. It is possible
that other, currently unidentified factors also played a role in this
reduction.
The incidence of injuries in this study is higher than the incidence
reported in other studies
6-8
. The low economic status of large areas of the city, which has often
been shown to correlate with an increased incidence of pediatric
pedestrian injuries
2,6,7,19
, may contribute to these differences.
Although a number of studies have documented risk factors for pedestrian
injury
2,18,19,24
, few have identified cities that have substantially reduced the number
of pedestrian injuries. Fortenberry and Brown reported a 34% decline
in pedestrian injuries of six and seven-year-olds in four Alabama
cities after institution of a school-based pedestrian-safety program
25
. Blomberg et al. described an 18% decrease in midblock "dart and
dash" injuries in Los Angeles and Milwaukee and a 36% decrease in
Columbus, Ohio, after institution of a school-based education and
awareness campaign
21
. The results of that study appear to be inconsistent with actual changes
in behavior, as the number of children looking for oncoming motor
vehicles before crossing the street increased 8% in Los Angeles,
increased 12% in Milwaukee, and decreased 4% in Columbus. Additionally,
only about 20% to 33% of elementary school children received the
traffic safety education.
Although some of the interventions discussed above were instituted
for reasons other than child pedestrian safety, our study demonstrated
that a multifactorial approach may have substantially decreased
child pedestrian injuries. It also demonstrated that improvements
can be made without changing some of the intractable problems that
also correlate with child pedestrian injury, such as poverty. Furthermore,
with the exception of the demolition of one housing project, expensive structural
changes were not necessary to improve children's safety.
There are other environmental and community interventions that
might help to reduce the occurrence of child pedestrian injuries
and that were not initiated in this community during the period
of this study. These include:
- posting crossing guards at high-risk corners during high-risk times
- posting highly visible signs warning motorists of areas at high
risk for pedestrian-motor vehicle collisions
- providing well-lit, brightly colored crosswalks with reflectors or
small lights embedded in the street pavement
- establishing self-flagging crosswalks
- creating one-way streets
- banning parking on one side of the street
- adding speed bumps
- erecting pedestrian overpasses
- resetting traffic signals to give children more time to cross
busy intersections and to slow vehicular speeds between signals
- providing playgrounds that are safe and clean
- adding, improving, or fencing sidewalks
- banning cellular telephone use by drivers
Lastly, our study suggests an opportunity for orthopaedic surgeons
to provide a stimulus for change in their communities. This reported
experience could be used as a model to collect data and suggest
changes.