Abstract
The Fatality Analysis Reporting System (FARS) and the Web-based Injury Statistics Query and Reporting System (WISQARS) were used to compare fatal pedestrian crashes in American Indians and Alaskan Natives (AI/AN) between urban and rural locations for 2000–2001. There were significant differences between urban and rural crashes for driver, pedestrian, environmental, and engineering factors. Rural pedestrian crashes more often occurred on highways (p<0.0001) lacking traffic control devices (p<0.0001) and artificial lighting (p<0.0001). Alcohol was a significant cofactor in both environments (40% urban vs. 55% rural; p=0.0239). Prevention of AI/AN deaths should include engineering countermeasures specific to the needs of rural (lighting) and urban (medians with barriers) environments and address drinking behavior in both populations.
In the United States, motor vehicle crashes are a major public health concern for American Indians/Alaskan Natives (AI/AN). According to the National Center for Health Statistics (NCHS) Web-based Injury Statistics Query and Reporting System (WISQARS), American Indians and Alaskan Natives had the highest pedestrian death rate of any ethnic group in the United States in 2000 (Hoyert et al., 2001). Pedestrian crashes were the third leading cause of all deaths among AI/AN age 1–44 years, preceded by motor-vehicle driver, and motor-vehicle occupant crash related fatalities. In 2000, the age-adjusted pedestrian crash mortality rate was 4.73 per 100,000 population for AI/AN, significantly higher than the national age-adjusted mortality rate of 1.67 per 100,000 population for all races and ethnicities as reported by the National Center for Health Statistics, Web-based Injury Statistics Query and Reporting System (WISQARS). This rate reflects 311 pedestrian-related deaths among approximately 2.4 million AI/AN individuals living in the United States. Population estimates used to calculate the AI/AN pedestrian fatality rate were taken from the United States Census 2000 data.
Several factors may explain the high rates of AI/AN death. Factors previously identified to be associated with an increased risk of pedestrian mortality include poverty, lack of education, rurality (in terms of population density), and alcohol use (Campos-Outcalt, 1997). Other factors include roadway types, roadway characteristics, motorist behaviors and pedestrian behaviors (Porvaznik, 1988). In particular, rurality has been recognized to increase the pedestrian crash case fatality rate due to the nature of roadways in rural areas, elevated travel speeds, and isolation from emergency medical services and hospital trauma services (Gallaher et al., 1992).
Most fatal pedestrian crashes in the U.S. occur in urban areas (NHTSA, 2003) However, data from pedestrian crashes from 1990–1996 in Arizona show that there were differences in fatality rates among racial/ethnic groups with regard to urban and rural settings. Although other racial/ethnic groups showed similar urban and rural rates, American Indians in Arizona were found to have not only a markedly higher overall pedestrian fatality rate (6–13 times the rate for non-Hispanic whites), but the rural fatality rate exceeded the urban rate by almost 50%, or an additional 6.1 deaths per 100,000. This disparity persisted even after adjusting for the tendency toward rural residency in American Indians (Campos-Outcalt, 2002).
Because AI/AN pedestrian fatalities occur in both urban and rural settings at rates that exceed those of other races/ethnicities, we were interested in whether the high overall AI/AN pedestrian death rates would mask important differences in mortality between urban and rural settings, and whether there might be unique issues in each setting that could be addressed as a possible preventive strategy (Campos-Outcalt, 2002). We hypothesized that rural crashes would differ from urban crashes in regards to environmental and pedestrian factors, and would tend to occur in unlit areas on rural highways off the reservation. In addition, consistent with off-reservation drinking behaviors identified in past research (Watson, 1992), we hypothesized that rural pedestrian fatalities would occur along rural highways and that a larger proportion of these crashes would involve alcohol on the part of the pedestrian. We also hypothesized that alcohol would be a prominent factor in both urban and rural AI/AN pedestrian crashes. Through an analysis of the differences in AI/AN pedestrian mortality in urban and rural areas, we hoped to be able to focus specific prevention strategies for each group through engineering countermeasures, and to address pedestrian and driver factors through education and behavioral interventions.
The purpose of our research was to describe the crash characteristics of and individual and environmental differences between rural and urban fatal American Indian and Alaskan Native pedestrian crashes occurring within the United States from 2000–2001. Through descriptive analyses of these fatal AI/AN crashes we hope to identify points of intervention in rural and urban settings.
METHODS
A cross-sectional, retrospective study design was used to address the research question. Fatal pedestrian crash data involving either American Indians or Alaskan Natives were obtained from the National Highway and Transportation Administration (NHTSA) Fatality Analysis Reporting System (FARS) and National Center for Health Statistics (NCHS) Web-based Injury Statistics and Reporting System (WISQARS). FARS data consist of motor vehicle traffic crashes occurring in public roads which result in the death of an occupant of a vehicle or a non-motorist within 30 days of the crash and are collected from police accident reports, State Highway Department data, vital statistics, death certificates, coroner reports, and hospital medical records. FARS data were collected on single and multiple vehicle pedestrian crashes for 2000–2001, years for which ethnic and race information was available. The FARS data elements utilized included crash characteristics, roadway characteristics, and demographic information of both the driver and the pedestrian. Analyses were conducted to elicit characteristics of the individuals involved in the crash, driver and pedestrian, as well as the characteristics of the environment surrounding the crash. The individual characteristics for pedestrian and motorist that were analyzed included alcohol involvement at the time of the crash, previous motorist citations, blood alcohol levels, gender, ethnicity, age and age group. Additional analyses investigated environmental conditions surrounding the AI/AN fatal crash in urban and rural locales. The environmental crash conditions that were analyzed included crash location, roadway characteristics, roadway design elements, lighting conditions, month, day of week and time of day.
Data were limited to those records that were not missing pedestrian race information and whose race was equal to American Indian/Alaskan Native (n=263). For a two year period, 311 AI/AN pedestrian fatalities were identified using the NCHS WISQARS data. Hence, comparing death certificates to AI/AN individuals in FARS, approximately 15% records were missing race information. Throughout the analyses, records which were missing data for variables analyzed were dropped from those specific analyses. Urban and rural designations were assigned using the FARS Roadway Function Class variable which describes the urban or rural designation of the highway, street, or road at which the crash occurred. Using the FARS definition for rurality, communities with populations under 5,000, or roadway segments which lay outside of incorporated municipalities with greater than 5,000 residents, were defined as rural. Urban areas were defined as those municipal roads or highway segments associated with populations above 5,000 residents.
FARS variables were regrouped and condensed to facilitate more meaningful comparisons. Age groups were assigned using stem plotting methods. For this analysis, five year intervals of age were chosen for pedestrian age groups and 20 year intervals were chosen for driver age groups.
Additionally, blood alcohol concentration (BAC) associated with pedestrian fatalities was assessed using the FARS imputed alcohol related data. NHTSA has developed a statistical model estimating the driver and non-occupant BAC test result. The statistical model is based on important characteristics of the crash including crash factors (e.g., time of day, day of week, type of crash, location), vehicle factors (e.g., vehicle type and role in crash), and person factors (e.g., age, sex, restraint use, previous driving violations) as well as state factors such as liquor laws. The multiple imputation method employs 3-level linear discriminant models to estimate the probability that a particular driver or non-occupant has a BAC of 0.0 – 0.1 or greater based upon the aforementioned crash characteristics (NHTSA, 1998). Recent research has demonstrated that multiple imputation increased alcohol-involved counts between 2% – 5% per year for 1992–2000 (NHTSA, 2002).
Age adjustment was used in order to compare fatality rates without concern that the observed differences are related to differences in the age distribution between different populations. Age adjustment was conducted using the direct method wherein adjusted rates are derived by applying category-specific rates observed in each of the populations to a single standard population (2000 AI/AN US Population). For all rates, 2000 was selected as a standard year. WISQARS injury-related mortality data is compiled annually by the NCHS. These data are derived from the ‘multiple cause of death’ data received from all states in the United States as part of the mandatory reporting required by the National Vital Statistics System (NVSS). NVSS protocols require the submission of demographic information as well as details surrounding the death for every death certificate issued in the US (Hoyert et al., 2001). Age-adjusted fatality rates were calculated by applying age-specific data for each age strata to the corresponding US AI/AN population to obtain a rate within each strata. These were then summed and divided by the total US AI/AN population to obtain the age-adjusted rate.
Data were analyzed using the SAS version 8.2 software. Chi square test for significance and proportion and Student’s t statistical measures of difference were applied where appropriate. Descriptive statistics were used to describe which demographic, crash, and roadway characteristics were associated with the incidence of AI/AN pedestrian crashes at rural and urban locations. Relative rates of injury were calculated within a 95% confidence levels. We used a two-tailed type I error rate of 5% to determine significance.
RESULTS
Using data collected from both the NCHS WISQARS system and FARS, we calculated the AI/AN pedestrian fatality rate per 100,000 population by rural and urban designations using the county variable in FARS data. Rural locations were found to have a higher AI/AN pedestrian fatality rate per population than urban areas (6.1 deaths vs. 3.9 deaths per 100,000 population, p<0.0001). The relative rate of pedestrian death per 100,000 population among AI/AN compared to all other races was 2.56 (95% confidence level).
Characteristics surrounding the environment of the pedestrian crash were analyzed. Differences in crash characteristics associated with fatal AI/AN pedestrian crashes in the United States during 2000–2001 are represented in Tables 1 and Table 2. Crash characteristics common among urban AI/AN fatalities included their occurrence within the roadway, on municipal roads and major arterials. Urban crashes were also more likely associated with a level roadway profile, two lanes of traffic, and a clearly marked median with no barrier. Urban crashes occurred most often outside of reservation jurisdictions, in the dark, without alcohol involvement by either the driver or the pedestrian, and without a hit and run crash designation. Urban crashes occurred most frequently on the weekend. Rural AI/AN fatal pedestrian crashes occurred on segments of state highways and interstates, within the roadway. Rural crashes were most likely to be associated with a level roadway profile, two lanes of traffic, and an undivided highway segment. Rural crashes tended to occur in the dark, with noted alcohol involvement on the part of either the driver or the pedestrian, and without a hit and run designation. Rural crashes also tended to occur most frequently on the weekend.
Table 1.
Selected roadway characteristics of AI/AN fatal pedestrian crashes, United States, 2000 – 2001. Source: FARS.
| URBAN CRASHES
|
RURAL CRASHES
|
p value | |||
|---|---|---|---|---|---|
| N | (%) | N | (%) | ||
| Roadway Characteristics | 103 | 185 | |||
| Route Signing | |||||
| US Highway/Interstate | 22 | (21.4) | 62 | (33.5) | < 0.0001 |
| State Highway | 18 | (17.5) | 84 | (45.4) | < 0.0001 |
| County Roads | 13 | (12.6) | 34 | (18.4) | = 0.2051 |
| Municipal Roads | 50 | (48.5) | 5 | (2.7) | < 0.0001 |
| Roadway Class | |||||
| Arterial | 60 | (58.3) | 29 | (15.7) | < 0.0001 |
| Collector | 15 | (14.6) | 59 | (31.9) | = 0.0012 |
| Interstate/Highway | 21 | (20.4) | 64 | (34.6) | = 0.0113 |
| Local Street/Road | 7 | (6.7) | 33 | (17.8) | = 0.0094 |
| Relation to Roadway | |||||
| On Roadway | 100 | (97.1) | 156 | (84.3) | = 0.0009 |
| Shoulder / Roadside | 2 | (1.9) | 14 | (7.6) | = 0.0457 |
| Off Roadway | 1 | (1.0) | 15 | (8.1) | = 0.0117 |
| Roadway Profile | |||||
| Level | 83 | (80.6) | 129 | (69.7) | = 0.0451 |
| Grade | 16 | (15.5) | 50 | (27.0) | = 0.0261 |
| Hillcrest | 4 | (3.9) | 6 | (3.2) | = 0.7761 |
| Lanes of Travel | |||||
| 1 - 2 Lanes | 61 | (59.2) | 161 | (87.0) | < 0.0001 |
| 3 - 4 Lanes | 37 | (35.9) | 7 | (3.8) | < 0.0001 |
| 5 or More Lanes | 5 | (4.9) | 17 | (9.2) | = 0.1843 |
| Traffic Way Flow | |||||
| Undivided | 40 | (38.8) | 128 | (69.2) | < 0.0001 |
| Median - No Barrier | 43 | (41.7) | 54 | (29.2) | = 0.0306 |
| Median - With Barrier | 20 | (19.4) | 3 | (1.6) | < 0.0001 |
Table 2.
Selected crash characteristics of AI/AN fatal pedestrian crashes, United States, 2000 – 2001. Source: FARS.
| URBAN CRASHES
|
RURAL CRASHES
|
p value | |||
|---|---|---|---|---|---|
| N | (%) | N | (%) | ||
| Crash Characteristics | 103 | 185 | |||
| Reservation Jurisdiction | |||||
| Yes | 2 | (1.9) | 85 | (45.9) | < 0.0001 |
| No | 101 | (98.1) | 100 | (54.1) | |
| Time of Day (Crash) | |||||
| Dark | 71 | (68.9) | 131 | (70.8) | = 0.0989 |
| Daylight | 25 | (24.3) | 30 | (16.2) | = 0.0955 |
| Dawn / Dusk | 7 | (6.8) | 24 | (13.0) | = 0.1049 |
| Alcohol Involvement* | |||||
| Yes | 35 | (40.2) | 90 | (55.2) | = 0.0239 |
| No | 52 | (59.8) | 73 | (44.8) | |
| Hit and Run Crash | |||||
| No | 77 | (74.8) | 151 | (81.6) | = 0.1691 |
| Yes | 26 | (25.2) | 34 | (18.4) | |
| Day of Crash | |||||
| Weekday (Mon - Thur) | 44 | (42.7) | 86 | (46.5) | = 0.5379 |
| Weekend (Fri - Sun) | 59 | (57.3) | 99 | (53.5) | |
NOTE: Alcohol Involvement refers to police assessment at crash scene indicating signs of alcohol involvement on part of driver and/or pedestrian
A comparison of urban to rural crash characteristics showed qualitative similarities in many categories. However, statistically significant differences in magnitude were observed between rural and urban groups in all crash characteristics except time of day, hit and run designation, and day of the week. Table 1 also shows a comparison of the environmental factors associated with rural and urban AI/AN pedestrian crashes demonstrated differences in the type of roadway, with rural crashes tending to occur more frequently on state highways while urban crashes tended to occur on municipal roads (p<0.0001). Higher percentages of rural deaths occurred on the roadway shoulder and off the roadway compared to urban crashes (p=0.0142). A higher percentage of rural pedestrian deaths showed alcohol involvement on the part of the either the driver or pedestrian (p=0.0239).
In addition to comparisons found in Table 1, presence or absence of traffic control devices between urban and rural crash sites was explored. Urban crashes more often involved artificial lighting than rural crashes (79% vs. 13%; p<0.0001). Traffic control devices related to intersections were also more common among urban AI/AN pedestrian fatalities. Rural AI/AN pedestrian crash locations often lacked intersections, crosswalks, and signals (p<0.0001).
In addition to environmental factors associated with rural and urban AI/AN pedestrian fatalities, individual characteristics related to the pedestrian were also found to have distinct differences between rural and urban crash locations. Figures 1 and 2 allow a visual comparison of the age and sex distribution of rural and urban AI/AN pedestrian victims. These figures show the disproportionate nature of pedestrian fatality with regards to sex in both urban and rural settings. A greater percentage of AI/AN pedestrian fatalities involved males (76%). In addition, distinct differences in at-risk age groups between rural and urban AI/AN populations emerged. Within urban areas, more children and elderly fatalities occurred. Among rural AI/AN populations, young males 15–29 predominated as at-risk. Furthermore, the mean age of an AI/AN pedestrian fatality was lower among rural crashes than urban crashes (37.3 years vs. 42.3 years, p=0.0308).
Figure 1.
Rural AI/AN Pedestrian Fatalities, 2000–2001. Source: CDC WISQARS.
Figure 2.
Urban AI/AN Pedestrian Fatalities, 2000–2001. Source: CDC WISQARS.
Table 3 describes pedestrian characteristics associated with fatal crashes. Male pedestrians are overrepresented in each group (urban 73.8%, rural 78.9%). There was no significant difference between urban and rural groups with regards to pedestrian alcohol involvement as indicated by blood alcohol content analysis postmortem (urban 46.6% alcohol-involved, rural 52.4% alcohol-involved; p=0.1662). The top contributing factor in the crash attributed to the pedestrian for urban crashes was improper crossing. The top contributing factor in the crash attributed to the pedestrian for rural pedestrian fatalities was walking in the roadway.
Table 3.
Selected AI/AN pedestrian characteristics associated with fatal AI/AN pedestrian crashes, United States, 2000–2001. Source: FARS.
| URBAN CRASHES
|
RURAL CRASHES
|
p value | |||
|---|---|---|---|---|---|
| N | (%) | N | (%) | ||
| Pedestrian Characteristics | |||||
| Total Pedestrians | 103 | 185 | |||
| Males | 76 | (73.8) | 146 | (78.9) | = 0.3206 |
| Pedestrian Drinking | |||||
| Yes | 48 | (46.6) | 97 | (52.4) | = 0.1662 |
| Top Contributing Factor | |||||
| Walking in Road | 20 | (16.7) | 90 | (53.3) | < 0.0001 |
| Improper Crossing | 43 | (35.8) | 18 | (10.7) | < 0.0001 |
| Not Visible to Driver | 9 | (7.5) | 32 | (18.9) | = 0.0463 |
| Dart / Run into Road | 13 | (10.8) | 14 | (8.3) | = 0.1584 |
| Failure to Yield | 19 | (15.8) | 5 | (3.0) | = 0.0001 |
| Other | 16 | (13.3) | 10 | (5.9) | = 0.0040 |
NOTE: Pedestrian drinking refers to the presence of a BAC associated with the fatality.
Factors associated with the driver in fatal pedestrian crashes are described in Table 4. (below) There were significantly more male drivers involved in rural pedestrian crashes than urban crashes (71.9% vs. 64.1%; p<0.0001). The age distribution of the drivers, the state of residence of the driver, and the frequency of issued citations for the crash were not significantly different between urban and rural groups. No difference in mean age was observed among rural and urban drivers. Urban drivers involved in fatal AI/AN pedestrian crashes were significantly more likely to have received previous traffic citations than rural drivers (urban 26.2%, rural 15.7%; p<0.0001). Rural drivers were significantly more likely to have been cited for drinking at the time of the crash than urban drivers (urban 9.7%, rural 14.0%; p=0.0285). The top contributing factor in the crash attributed to the driver in both urban and rural crashes was hit-and-run; however, there were significantly more hit and run crashes in urban settings (urban 39.3%, rural 21.7%; p =0.0218).
Table 4.
Selected driver characteristics associated with fatal AI/AN pedestrian crashes, United States, 2000 – 2001. Source: FARS.
| URBAN CRASHES
|
RURAL CRASHES
|
p value | |||
|---|---|---|---|---|---|
| N | (%) | N | (%) | ||
| Driver Characteristics | |||||
| Total Drivers | 103 | 185 | |||
| Males | 66 | (64.1) | 133 | (71.9) | < 0.0001 |
| Age group (in years) | |||||
| 18 – 25 | 16 | (15.5) | 30 | (16.2) | = 0.8795 |
| 26 – 35 | 32 | (31.1) | 59 | (31.9) | = 0.8853 |
| 36 – 45 | 34 | (33.0) | 55 | (29.7) | = 0.5637 |
| 46 – 65 | 15 | (15.6) | 28 | (15.1) | = 0.8961 |
| 66 and older | 6 | (4.9) | 13 | (7.0) | = 0.6937 |
| Driver State of Residence | |||||
| In-State Driver | 76 | (73.8) | 114 | (64.4) | = 0.1051 |
| Out-of-State Driver | 27 | (26.2) | 63 | (35.6) | = 0.2213 |
| Driver Cited for Crash | |||||
| Yes | 21 | (20.4) | 55 | (29.7) | = 0.0847 |
| No | 82 | (79.6) | 130 | (70.3) | |
| Previous Citation | |||||
| Yes | 27 | (26.2) | 29 | (15.7) | < 0.0001 |
| No | 76 | (73.8) | 156 | (84.3) | |
| Driver Drinking* | |||||
| Yes | 10 | (9.7) | 26 | (14.0) | = 0.0285 |
| No | 93 | (90.3) | 159 | (86.0) | |
| Top Contributing Factor | |||||
| Hit and Run | 22 | (39.3) | 26 | (21.7) | = 0.1108 |
| Driving Too Fast | 6 | (10.7) | 19 | (15.8) | = 0.1991 |
| Driver Inattention | 8 | (14.3) | 11 | (9.2) | = 0.5507 |
| Run Off Road/Lane | 0 | (0.0) | 19 | (15.8) | = 0.0008 |
| Glare | 3 | (5.4) | 8 | (6.7) | = 0.5491 |
| Failure to Yield | 4 | (7.1) | 4 | (3.3) | = 0.3942 |
| Drowsy / Asleep | 1 | (1.8) | 6 | (5.0) | = 0.3942 |
| Weather | 2 | (3.6) | 5 | (4.2) | = 0.2300 |
| Other | 10 | (7.9) | 22 | (18.3) | = 0.3240 |
NOTE: Driver drinking refers to the presence of a BAC associated with the fatality. BAC was imputed in 21% of records.
Table 5 (below) describes the mean driver and pedestrian BAC for fatal AI/AN pedestrian crashes. Mean BAC was calculated using imputed BAC information. FARS imputation of BAC data is described above. For all AI/AN pedestrian fatalities BAC was imputed in only 16% of the total records. Generally, drivers issued a citation for driving while impaired received a BAC test. Drivers sometimes refuse a BAC test, the test is not administered, or the information is missing from the data. Similarly, pedestrian crash victim BAC is also obtained, often more frequently because of coroner protocol. No significant differences between urban and rural BAC for either the driver or the pedestrian were observed.
Table 5.
Comparison of mean driver and pedestrian blood alcohol concentration for fatal AI/AN pedestrian crashes, United States, 2000–2001. Source: FARS.
| Blood Alcohol Concentration(BAC) | URBAN CRASHES
|
RURAL CRASHES
|
p value | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| n | Mean | Min | Max | SD | n | Mean | Min | Max | SD | ||
| Pedestrian | 80 | 18.9 | 0 | 48 | 13.6 | 92 | 20.8 | 0 | 44 | 12.5 | = 0.3355 |
| Driver | 32 | 15.6 | 0 | 31 | 12.2 | 48 | 18.6 | 0 | 36 | 16.3 | = 0.2341 |
NOTE: Values are given as a percentage of alcohol (e.g. a BAC of 24 reads .24 mg/dL on above table).
DISCUSSION
AI/AN have a significant risk of dying as a victim of a pedestrian crash in both urban and rural environments, with an overall pedestrian fatality rate that is 2.5–3 times higher than all other races/ethnicities. Unlike national pedestrian injury patterns, rural AI/AN communities have higher rates of pedestrian fatality than their urban counterparts. (NHTSA, 2003) Although this national fatality rate disparity is not as dramatic as the disparity noted in Arizona for this population, it still represents a significant risk to AI/AN in the U.S. as a whole. In addition, these data, like the data from Arizona, show that there are significant differences in the risk in each environment. Our results show that compared to urban AI/AN crashes, rural crashes were more likely to involve a pedestrian who was walking within the roadway and a driver who had been drinking. Similar to the results from Arizona (Campos-Outcalt, 2002), on a national level rural AI/AN pedestrians who die in a crash are more likely to have been drinking than their urban counterparts. However, there was no significant difference in mean blood alcohol concentration between rural and urban pedestrians and rural and urban drivers. In both locations, males in mid-adulthood appear to be at highest risk although pedestrians over 70 years of age also are at substantial risk. Elderly pedestrians tend to succumb to injuries that might not kill younger pedestrians because they die of complications of hospitalization or lack substantial physiological reserve to respond to the injury insult (Sklar et al., 1989). Further research is needed to determine the urban-rural disparity in pedestrian deaths in races/ethnicities other than AI/AN.
Urban pedestrian victims are almost always hit while improperly crossing the road, on roads with medians and one to two lanes and within intersections that have functioning traffic control devices. These roadway characteristics suggest that mapping of locations for pedestrian injury and the implementation of pedestrian safety engineering countermeasures might be productive interventions for the prevention of urban AI/AN pedestrian deaths (Middaugh, 1989). Since rural areas often lack such engineering countermeasures and traffic control devices, the application of such devices may reduce rural pedestrian crashes that occur either within small towns or on reservation communities.
Engineering countermeasures for rural AI/AN crashes would likely be targeted to segments of rural state highway and interstate. These countermeasures should emphasize the lighting of roadways and adjacent roadway shoulders, division of highways with barriers, and shoulder separation from roadways in selected high risk areas. These measures are most needed in the smaller one to two lane roads which are the site of most fatal AI/AN pedestrian crashes. Interstate highways and state highways with high rates of AI/AN pedestrian injury should be identified and mapped such that specific high risk areas can be identified and visited to assess possible traffic engineering changes that would prevent pedestrians from utilizing the roadway.
AI/AN have unacceptably high pedestrian fatality rates in both urban and rural locations, although the rural rate is higher. The comparison of rural and urban AI/AN pedestrian fatalities highlights an issue that transcends location. The common theme in both locations is the presence of alcohol among the pedestrian and/or the driver. Usually the pedestrian was an adult male who was walking along the roadway in the dark on a weekend. These crashes likely are associated with pedestrian drinking behavior, whether in urban or rural locations. The level of intoxication in both the driver and the pedestrian, in both urban and rural locations, is substantial. For reference, in most states the “legal limit” of BAC is 0.08 g/dL. These data show that levels of 0.30 g/dL or higher occurred in both pedestrians and drivers. This level of profound intoxication is a substantial contributing factor in all types of automobile crashes, including pedestrian death. Behavioral countermeasures with measured levels of success have included the increased enforcement of alcohol server laws (Watson, 1992). Rural crashes are compounded by the long distance from medical care and the higher speeds of traffic moving along highways.
It should be considered that the BAC data reviewed for this study was from the FARS alcohol data file which uses a multiple imputation technique to assign alcohol involvement, and that a limited number of records contained imputed alcohol information. Furthermore, mean BAC comparisons between rural and urban fatal AI/AN pedestrian crashes show little difference.
Efforts to prevent AI/AN pedestrian deaths will require both an urban and rural component. The urban component should be a part of a comprehensive pedestrian safety program that targets all at-risk pedestrians, as well as drinking drivers. Other urban prevention measures should include the mapping of high risk areas and engineering modifications. The rural component should focus on drinking drivers, as well as pedestrian behaviors of walking in the roadway and traffic engineering measures in high risk areas including highway lighting. Educational campaigns would also be best directed at the youth or elderly AI/AN in on and off-reservation urban communities.
Accuracy of rural and urban fatal AI/AN pedestrian crash comparisons is impacted by the infrequent collection of race information in the FARS data. This study was limited to the 2000 – 2001 period because previous years were missing up 53% of the race information for each unique crash record. By comparing FARS data to the WISQARS death certificate data, it was found that the FARS database is missing about 15% of the records found in the WISQARS database. Many reasons for this discrepancy exist. Often tribes do not use coroner services or generate standardized police accident reports. Hence, on-reservation crash counts are typically lower than reported by other sources like local area trauma registries or the Indian Health Service. Lowered on-reservation crash counts would impact the number of rural crashes considered in this research. Although in New Mexico, it is known that this effect is minimal (Gallaher, 1992).
CONCLUSION
AI/AN pedestrian fatality analysis demonstrated high death rates in both urban and rural environments. Although alcohol intoxication of pedestrian and/or driver occurs commonly in both settings and represents an opportunity to intervene through law enforcement and alcohol abuse treatment programs, other traffic engineering countermeasures can also be identified. These engineering countermeasures should be tailored to the specific data concerning risk in each environment, both urban and rural. Dangerous rural highway segments require lane separation, physical dividers, and other traffic control devices in addition to artificial lighting. Dangerous urban areas require increased traffic calming measures, and behavior modification programs targeted at alcohol-involved crashes and unsafe pedestrian behavior.
REFERENCES
- Baltes Michael R. A Study of Fatal Pedestrian Crashes in Florida. Center for Urban Transportation Research; 1996. [Google Scholar]
- Campos-Outcalt D, Prybylski D, Watkins AJ, Rothfus G, Dellapenna A. Motor-Vehicle Crash Fatalities among American Indians and Non-Indians in Arizona, 1979 through 1988. American Journal of Public Health. 1997;87(2):282–284. doi: 10.2105/ajph.87.2.282. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Campos-Outcalt D, Bay C, Dellapenna A, Cota Marya. Pedestrian Fatalities by Race/Ethnicity in Arizona, 1990–1996. American Journal of Preventive Medicine. 2002;23(2):129–135. doi: 10.1016/s0749-3797(02)00465-8. [DOI] [PubMed] [Google Scholar]
- Gallaher M, Fleming DW, Berger L, Sewell M. Pedestrian and Hypothermia Deaths Among Native Americans in New Mexico Between Bar and Home. JAMA. 1992;267(10):1345–1347. [PubMed] [Google Scholar]
- Leden Lars. Pedestrian risk decrease with pedestrian flow. A Case Study Based on Data from Signalized Intersections in Hamilton, Ontario. Accident Analysis and Prevention. 2002;34:457–464. doi: 10.1016/s0001-4575(01)00043-4. [DOI] [PubMed] [Google Scholar]
- Macias D, Olson L, Sandvig J, Collins K, Sklar DP. Program for Reducing Pedestrian and Motor Vehicle Crashes. Academic Emergency Medicine. 1997 March 1998;5(3):252–257. doi: 10.1111/j.1553-2712.1998.tb02623.x. [DOI] [PubMed] [Google Scholar]
- Middaugh JP. Pedestrian Fatalities in Alaska. Alaska Medicine. 1989;31(3):108–113. [PubMed] [Google Scholar]
- National Highway Traffic Safety Administration [NHTSA] Technical Report, DOT HS 809 456. 2003. Pedestrian Roadway Fatalities. National Center for Statistics and Analysis, 400 7th Street, S.W., Washington, D.C., 20590. [Google Scholar]
- National Highway Traffic Safety Administration [NHTSA] National Center for Statistics and Analysis. 2002. Transitioning to Multiple Imputation–A New Method to Impute Missing Blood Alcohol Concentration (BAC) Values in FARS. Technical Report, DOT HS 809 403. 400 7th Street SW, Washington, D.C., 20590. [Google Scholar]
- National Highway Traffic Safety Administration [NHTSA] National Center for Statistics and Analysis. 1998. Multiple Imputation of Missing Blood Alcohol Concentration (BAC) Values in FARS. Research Note, DOT. 400 7th Street, SW, Washington, D.C., 20590. [Google Scholar]
- Preusser DF, Wells JK, Williams AF, Weinstein HB. Pedestrian Crashes in Washington, DC and Baltimore. Accident Analysis and Prevention. 2002;34:703–710. doi: 10.1016/s0001-4575(01)00070-7. [DOI] [PubMed] [Google Scholar]
- Sklar DP, Demarest GB, McFeeley P. Increased Pedestrian Mortality Among the Elderly. American Journal of Emergency Medicine. 1989;7(4):387–390. doi: 10.1016/0735-6757(89)90044-2. [DOI] [PubMed] [Google Scholar]
- Watson CS. Injury Prevention Practitioners Fellowship Project. IHS; Ft. Defiance District Office: 1992. Pedestrian Fatalities on the Navajo Reservation, 1986 – 1991. [Google Scholar]