Article Text
Abstract
Objective: To examine the urban and rural variation in walking patterns and pedestrian crashes.
Design: The rates of pedestrians being struck by motor vehicles was estimated according to miles walked and resident years.
Setting: New York State, USA during 2001 through 2002.
Participants: 35 732 pedestrians struck by vehicles.
Main outcome measures: The adjusted rate ratio (aRR) of pedestrian–vehicle crash and pedestrian injury based on resident years and miles walked according to urban and rural areas.
Results: Compared with rural areas, the aRR for a pedestrian–vehicle collision, based on resident years, was 2.0 (95% CI 1.7 to 2.3) in small urban areas, 1.8 (95% CI 1.5 to 2.3) in mid-size urban areas, and 4.2 (95% CI 3.6 to 4.8) in the large urban area. The aRR based on miles walked was 2.3 (95% CI 1.6 to 3.2) in small urban areas, 2.0 (95% CI 1.4 to 2.9) in mid-size urban areas, and 1.9 (95% CI 1.4 to 2.7) in the large area. The aRR for a fatal pedestrian injury, based on miles walked, was 2.1 (95% CI 1.3 to 3.6) in small urban areas, 1.9 (95% CI 1.3 to 2.9) in mid-size urban areas, and 0.9 (95% CI 0.6 to 1.3) in the large urban area.
Conclusions: The rate of pedestrian crashes and injuries in small and mid-size urban areas was twice that in rural areas, whether based on resident years or miles walked. The high rate of pedestrian crashes in the large urban area based on resident years could be partly explained by the fact that residents in such areas walk about twice as much as residents in rural areas. The rate of fatal pedestrian injury based on miles walked was similar in the large urban area and rural areas.
Statistics from Altmetric.com
Pedestrians represent more than one-third of the 10 million traffic injuries and 1.2 million traffic fatalities every year worldwide.1 In the USA in 2004, 4641 pedestrians died and 68 000 were injured in motor vehicle crashes; pedestrians accounted for 11% of all traffic deaths.2 There were 328 pedestrian deaths in New York State in 2004, 21% of all traffic deaths.3
In the USA, most pedestrian deaths and injuries occur in urban areas; rates for all pedestrian–motor vehicle collisions per resident year are usually higher in urban areas, but rates for pedestrian deaths are sometimes higher in rural areas.4–6 A limitation of resident-year-based measures of pedestrian–vehicle crash and injury is that it does not account for the time pedestrians spend on roads and the distance walked.6–8 The 2001 National Household Travel Survey (NHTS) can provide estimates of annual miles walked and walking patterns,9 which can help to generate walking-based measures.
Our objective was to examine the urban and rural difference in walking, and estimate how pedestrian collision rates per resident year and per mile walked varied between urban and rural areas. In particular, we wished to know whether variations per resident year might be explained, in part, by differences in the amount people walk in rural and urban areas.
MATERIALS AND METHODS
Pedestrian injury data
To obtain counts of pedestrians who were hit by vehicles, we used police crash reports for New York State during 2001 and 2002. These reports are routinely collected by law enforcement.10 They provided police-gathered information about injury severity, classified as none, complaint of pain, visible, incapacitating, and fatal.
Population estimates
We used data from Claritas Incorporated for 2001 and 2002 to obtain estimates of resident population by county, sex, and age.11
Information about walking
Estimates of miles walked by New York State residents were obtained from the 2001 NHTS in the USA.9 This survey used computer-assisted telephone interviews to collect information about personal and household characteristics, daily trips, and long-distance travel. A total of 30 268 people were interviewed in New York State from April 2001 through May 2002. For daily trips, respondents were instructed to keep a written diary of all the trips made during a randomly assigned 24 h travel day; information included trip purpose, transportation means, and trip length. A total of 26 401 people recorded the travel diary for daily trips. Walk trips as short as 0.1 mile were recorded. Each respondent was assigned a sampling weight or a series of replicate weights to reflect their selection probability and to adjust for non-response, undercoverage, and multiple phones in a household. The respondents were a weighted sample of all New York State residents.9
We estimated average annual miles walked per person according to the population density of the geographic region (described below), sex, and four age categories (described below). We accounted for the sampling weights, sampling strata, and sampling units used by NHTS using SUDDAN software V9.0. Jackknife replication methods were used to estimate variances of annual miles walked.12
Classification of geographic regions
The 2001 NHTS data allowed us to classify geographic regions into four levels of population density based on Metropolitan Statistical Area (MSA) categories from the US Census.9 An MSA was a county or group of adjacent counties that had either one city with at least 50 000 population, or two cities with a combined population of 50 000 or more. Adjacent MSA counties were grouped into a single MSA. For this study, we considered all non-MSA counties to be rural. Urban areas were classified into three categories: small areas with MSAs less than one million population, mid-size areas with MSAs of one to less than three million, and a large urban area with one MSA larger than three million. The small urban category included six MSA regions (Albany, Binghamton, Elmira, Glen Falls, Syracuse, and Utica-Rome), the mid-size category included two MSAs (Buffalo-Niagara Falls and Rochester), and the large category covered the New York City area.
Statistical analysis
Assuming Poisson distributions, we estimated rates per resident year and per miles walked and their 95% CIs with profile likelihood methods.13 To compare rates across geographic regions, rate ratios were estimated using Poisson regression. To relax the assumption that the counts fit a Poisson distribution, we used a robust variance estimator.1415 For the model based on resident years, the dependent variable was the number of pedestrian crashes or injuries by the pedestrian victim’s age (0–15, 16–24, 25–64, 65 years or older) and sex in the geographic region. The independent variables included the urbanicity of the geographic region and the pedestrian victims’ age and sex. The offset was the total number of resident years by the age and sex in the geographic region. For the miles walked model, the offset was the total number of miles walked by age and sex in the geographic region. The adjusted rate ratio estimates were based on information from 33 551 pedestrian victims (93.9% of the total), as sex was missing for 109 (0.3%) subjects and age data missing for 2072 (5.8%). We conducted subgroup analyses to examine whether the adjusted rate ratios comparing the four regions varied by age category or sex. We used the SAS PROC GENMOD procedure to fit Poisson models (SAS V9.1).
RESULTS
The population density was 63, 164, 346, and 42 123 per square mile for rural, small urban, mid-size urban, and large urban areas, respectively. People in rural areas made an average of 4.9 walking trips in the week before their interview, compared with 4.3 for the small urban category, 3.8 for mid-size urban areas, and 6.3 for the New York city region (table 1). Few respondents said that they walked to work, although this commute method was most common (6.6%) in the New York City region. Use of public transportation for commuting to work was least common in rural areas (0.5%) and most common (40.7%) in the large urban region. The average number of miles walked per year by residents was 133 in rural areas, 112 in the small urban regions, 116 in the mid-size urban areas, and 278 in the largest urban region.
There were 35 732 pedestrians struck by vehicles in New York State during 2001–2002 (table 2). The pedestrian–vehicle crash rate per 100 000 resident years was 28.1 in rural areas, 55.5 in the small urban areas, 51.3 in the mid-size urban areas, and 116.9 in the large urban area. Compared with rural areas, the adjusted rate ratio for a pedestrian–vehicle collision, based on resident years, was 2.0 in small urban areas, 1.8 in mid-size urban areas, and 4.2 in the large urban area. The pedestrian–vehicle crash rate per million miles walked was 2.1 in rural areas, 5.0 in small urban areas, 4.4 in mid-size areas, and 4.2 in the large area. Compared with rural areas, the adjusted rate ratio for a vehicle hitting a pedestrian, based on miles walked, was 2.3 in small urban areas, 2.0 in mid-size areas, and 1.9 in the large area.
For the state as a whole, the rate of pedestrian–vehicle crashes per 100 000 resident years was 82.4 for females (95% CI 81.1 to 83.7), 105.5 for males (95% CI 104.0 to 107.0), 99.9 for persons 0–15 years (95% CI 97.7 to 102.1), 115.5 for persons 16–24 years (95% CI 112.4 to 118.7), 80.7 for those 25–64 years (95% CI 79.5 to 82.0), and 73.9 for those 65 years and older (95% CI 71.5 to 76.3). Subgroup analyses revealed that the adjusted rate ratios comparing the four geographic regions did not vary greatly by age category or sex.
Injury severity was unknown for 1047 pedestrians. Among the remaining 34 685 pedestrians, 1106 (3%) had no injury, 21 420 (62%) complained of pain, 6956 (20%) had visible injuries, 4563 (13%) had incapacitating injuries, and 640 (1%) died. The adjusted rate ratios for a visible, incapacitating, or fatal pedestrian injury, based on resident years, were about two in the small and mid-size urban areas and nearly three in the large urban area (table 3). Based on miles walked and compared with rural areas, the adjusted rate ratios were about two in the small and mid-size urban areas, but about one in the New York City area.
The adjusted rate ratio for fatal pedestrian injury, based on resident years, was about twice as great in all the urban regions as in the rural areas (table 4). The pattern was different, however, for deaths per mile walked; compared with rural areas, the adjusted rate ratio was about twice as great in the small and mid-size urban areas, but in the New York City area the rate per mile was nearly the same as that in rural areas.
DISCUSSION
We found that in the New York City area, the rate of pedestrian crashes per resident year was about four times that in rural areas; in small and mid-size urban areas it was double that of rural areas. However, pedestrian–vehicle collision rates based on miles walked were similar in all urban areas and about twice those in rural areas.
In small and mid-size urban areas, the pedestrian crash and fatal injury rates were about twice as great as those in rural areas, whether measured per resident year or per mile walked. Pedestrian–vehicle crashes originate from the interplay of pedestrian, vehicle driver, and environment, and multiple factors are involved, such as pedestrian’s age, gender, and annual walking distance, driver’s age and gender, vehicle speed, traffic density, traffic signs, availability of sidewalks, crosswalks, and refuge islands, the width of the road, and other factors.6716–25 Our analysis considered the pedestrian’s walking distance, gender, and age. The increased pedestrian crash rate in urban areas found in our study might be attributed to the higher vehicle density in urban areas, compared with rural areas; but many other factors may play a role and we do not have data to identify each of these factors.
What is already known on this topic
Most pedestrian deaths and injuries occur in urban areas.
Rates for pedestrian crashes and injuries per resident year are usually higher in urban areas.
What this study adds
The rate of pedestrian crashes and injuries in small and mid-size urban areas was twice that in rural areas, whether based on resident years or miles walked.
The high rate of pedestrian crashes based on resident years in the large urban area may partly be explained by the fact that residents in the large urban area walked about twice as much as residents in rural areas.
The rate of fatal pedestrian injury based on miles walked was similar in the large urban area and rural areas.
The higher rate of pedestrian involvement in collisions in the New York City area was partly explained by the fact that its residents walk about twice as much as other people in New York State. In the New York City area, the availability of convenient mass transit (subway and buses) and the inconvenience and expense of driving, probably encourages many people to travel on foot. Furthermore, the average severity of a pedestrian–vehicle crash was less in the New York City area, compared with rural and all other urban areas: the pedestrian crash rate per mile walked in New York City area was twice that of rural areas, but the fatal pedestrian rates per mile walked were nearly the same in New York City area and rural areas. These findings suggest that pedestrians in New York City area are more likely to be hit by a car when walking compared with rural areas, but not more likely to die. This may be attributed to lower vehicle speed in the New York City area. One source reports that compared with areas with a posted speed limit of 30 miles per hour or less, the ratio of deaths to injuries is about 2.5 times as high where the speed limit is 31–40 miles/h and 7 times as high where the speed limit is 41–54 miles/h.4
Our finding of higher pedestrian crash rates in urban compared with rural areas is consistent with previous studies.4571821 For example, Mueller and colleagues5 analyzed Washington State pedestrian crashes in 1981–1983 and reported that the rate per resident year in urban areas was 2.8 times that in rural areas. Paulozzi6 reported that the fatal pedestrian injury rate per miles of vehicle travel in urban states was 1.5 times that in rural states in 2003.
A strength of our study was the use of rates based on miles walked and rate ratios to take account of the distance traveled by foot, which is usually not available for crash data. A previous analysis of pedestrian injury based on resident years reported that New York City residents experienced a higher pedestrian injury rate than the rest of the state.26 However, our analysis revealed that, relative to rural areas, the adjusted rate ratio for incapacitating and fatal pedestrian injury based on resident years was twofold greater in the New York City area, but the rates based on miles walked were nearly the same in the New York City area and rural areas; the higher rate of pedestrian injury per resident year in New York City can be explained, in part, by the extra walking done by its residents.
A limitation of our study is that our classification of rural and urban areas was based on MSA classifications of counties, which may not reflect the rural or urban environment of each vehicle–pedestrian crash location. We classified areas based on MSA size because this allowed us to merge data from police crash reports, which are coded by county and city, and miles walked data from the NHTS, which are coded by MSA size or census tract. Nationwide, 14% of residents in metropolitan counties would be defined as rural according to Census Bureau definitions.27 This modest degree of misclassification is not likely to affect our estimates greatly. Another potential limitation is that residents in an MSA may be involved in pedestrian vehicle crashes while working in another MSA. According to the 2000 Census, 3.9% of New York City MSA residents and 6.2% of Albany MSA residents did not work in their own MSAs.28 This suggests any commuting effect on our estimates would be small. Furthermore, our analysis considered pedestrian walking distance, age, and gender, but pedestrian–vehicle crashes originate from the interplay of pedestrian, vehicle driver, and environment, and multiple factors are involved. The urban and rural variation in pedestrian crash and injury rate observed in our study can be attributed to many unmeasured factors. Lastly, ecological bias may exist. Our analysis were based on the urbanicity at the residential county level. We cannot rule out the possibility that those injured pedestrians in urban counties were those who walked less than the average person in the county.
CONCLUSIONS
We found that pedestrian crash rates per resident year and per mile walked were greater in small and mid-size urban areas than in rural areas. However, in the largest urban area, New York City area, higher rates for all vehicle–pedestrian crashes were partly explained by the extra walking done by its residents. Per resident year, the fatal pedestrian injury rate for New York City area residents was little different from that of small and mid-size urban areas; per mile walked, the New York City area’s rate of fatal pedestrian injury was less than that of other urban areas and similar to that of rural residents.
REFERENCES
Footnotes
Funding: This project was funded by a grant from the National Highway Traffic Safety Administration (NHTSA).
Competing interests: None.