Scholarly article on topic 'Age-related differences in fatal intersection crashes in the United States'

Age-related differences in fatal intersection crashes in the United States Academic research paper on "Animal and dairy science"

CC BY
0
0
Share paper
Academic journal
Accident Analysis & Prevention
OECD Field of science
Keywords
{Fatal / Intersection / Crashes / Accidents / Involvement / Aging}

Abstract of research paper on Animal and dairy science, author of scientific article — David A. Lombardi, William J. Horrey, Theodore K. Courtney

Abstract Objective Given the aging U.S. population and resulting number of older drivers in the coming years, it is important to understand the factors leading to their involvement in vehicle crashes and develop counter-measures to reduce their frequency and severity. This is also useful for helping older adults “age in place” in terms of accessibility, mobility, quality of life and safety. Thus, the objective of this study was to provide up-to-date data on differences in age-related risks and rates for involvement in fatal intersection motor-vehicle crashes in the US. Methods Pooled data for the years 2011–2014 from the FARS, a census of fatal traffic crashes within the 50 States, the District of Columbia, and Puerto Rico, created by the US National Highway Traffic Safety Administration (NHTSA) were used to calculate summary statistics including annualized crash rates. Multivariate logistic regression models were used to evaluate age and gender-related differences in fatal intersection crash risk, controlling for covariates. An induced exposure analysis was conducted to calculate crash involvement ratios (CIRs) for all two-vehicle fatal intersection crashes. Older and younger drivers were compared with respect to the presence of factors related to intersection crashes using a multivariate Poisson regression model. Results During the period of 2011–2014, among the reported 120,809 fatal accidents in the US involving 178,489 drivers of vehicles, 48,733 (28%) were drivers involved in fatal intersection crashes. Age-adjusted annualized fatal intersection crash rates per 100,000 licensed drivers were highest for drivers aged 85 or older (9.89/100,000), followed by 20 years of age (8.93/100,000). Teen and older drivers (55+ years of age) were over-involved in fatal intersection crashes, drivers from 20 to 54 years old were under-involved. Male and female drivers, 70–74 years of age, were 20% and 21%, respectively, more likely to be involved in a fatal intersection crash than 20–24year olds (of same gender). By age 85, fatal intersection crash risk for all drivers was almost doubled. Significant differences in factors related to crashes involving younger (<65) and older (65+ years) drivers were time of day, lighting and weather conditions, day of week, roadway type and number of lanes, presence of visible traffic controls, speed limit and estimated driving speed, and whether the driver was deemed at fault for the crash Conclusion The results provide the most up-to-date analysis of aging and fatal intersection crash risk in the US, and underscore several trends worthy of further investigation. Older adults face a number of challenges associated with natural aging, including sensory, perceptual, cognitive and motor declines that may impact their driving. As with younger drivers, expanded or renewed approaches to driver training at licensing renewals, as well as safety-based technological advances are viable avenues toward improving the safety outlook for older adults.

Academic research paper on topic "Age-related differences in fatal intersection crashes in the United States"

Contents lists available at ScienceDirect

Accident Analysis and Prevention

journal homepage www.elsevier.com/locate/aap

ACCIDENT ANALYSIS

Age-related differences in fatal intersection crashes in the United States

CrossMark

David A. Lombardia,c'*, William J. Horreyb, Theodore K. Courtney

a Center for Injury Epidemiology, Liberty Mutual Research Institute for Safety, Hopkinton, USA b Center for Behavioral Sciences, Liberty Mutual Research Institute for Safety, Hopkinton, USA c Department ofEnvironmental Health, Harvard T.H. Chan School ofPublic Health, Boston, USA

ARTICLE INFO

Article history:

Received 13 April 2016

Received in revised form 6 October 2016

Accepted 27 October 2016

Keywords: Fatal

Intersection

Crashes

Accidents

Involvement

ABSTRACT

Objective: Given the aging U.S. population and resulting number of older drivers in the coming years, it is important to understand the factors leading to their involvement in vehicle crashes and develop counter-measures to reduce their frequency and severity. This is also useful for helping older adults "age in place" in terms of accessibility, mobility, quality of life and safety. Thus, the objective of this study was to provide up-to-date data on differences in age-related risks and rates for involvement in fatal intersection motor-vehicle crashes in the US.

Methods: Pooled data for the years 2011-2014 from the FARS, a census of fatal traffic crashes within the 50 States, the District of Columbia, and Puerto Rico, created by the US National Highway Traffic Safety Administration (NHTSA) were used to calculate summary statistics including annualized crash rates. Multivariate logistic regression models were used to evaluate age and gender-related differences in fatal intersection crash risk, controlling for covariates. An induced exposure analysis was conducted to calculate crash involvement ratios (CIRs) for all two-vehicle fatal intersection crashes. Older and younger drivers were compared with respect to the presence of factors related to intersection crashes using a multivariate Poisson regression model.

Results: During the period of 2011-2014, among the reported 120,809 fatal accidents in the US involving 178,489 drivers of vehicles, 48,733 (28%) were drivers involved in fatal intersection crashes. Age-adjusted annualized fatal intersection crash rates per 100,000 licensed drivers were highest for drivers aged 85 or older (9.89/100,000), followed by 20 years of age (8.93/100,000). Teen and older drivers (55+ years of age) were over-involved in fatal intersection crashes, drivers from 20 to 54 years old were under-involved. Male and female drivers, 70-74 years of age, were 20% and 21%, respectively, more likely to be involved in a fatal intersection crash than 20-24year olds (of same gender). By age 85, fatal intersection crash risk for all drivers was almost doubled. Significant differences in factors related to crashes involving younger (<65) and older (65+ years) drivers were time of day, lighting and weather conditions, day of week, roadway type and number of lanes, presence of visible traffic controls, speed limit and estimated driving speed, and whether the driver was deemed at fault for the crash

Conclusion: The results provide the most up-to-date analysis of aging and fatal intersection crash risk in the US, and underscore several trends worthy of further investigation. Older adults face a number of challenges associated with natural aging, including sensory, perceptual, cognitive and motor declines that may impact their driving. As with younger drivers, expanded or renewed approaches to driver training at licensing renewals, as well as safety-based technological advances are viable avenues toward improving the safety outlook for older adults.

© 2016 Elsevier Ltd. All rights reserved.

* Corresponding author at: Principal Research Scientist Center for Injury Epidemiology Liberty Mutual Research Institute for Safety 71 Frankland Road, Hopkinton, MA 01748, USA.

E-mail address: david.lombardi@libertymutual.com (D.A. Lombardi).

1. Introduction

Each year fatal crashes in the U.S. lead to an estimated societal burden of more than $230 billion from medical and other costs (NHTSA, 2010a,b). After six consecutive years of declining rates on U.S. highways, crash and fatality rates increased in 2012 (NHTSA,

http://dx.doi.org/10.1016/j.aap.2016.10.030 0001-4575/© 2016 Elsevier Ltd. All rights reserved.

2013a,b), in which there were 33,561 people killed in roadway crashes, 3.3% higher compared with 32,479 in 2011. This increase was consistent across most crash characteristics - such as vehicle type, alcohol impairment, and location of the crash. Accordingly, there were an estimated 2.36 million people injured in motor vehicle traffic crashes (increase of 6.5% compared with 2.22 million in 2011) according to NHTSA's National Automotive Sampling System (NASS) General Estimates System (GES). Collectively, these numbers underscore the reality that, although we have made many gains in traffic safety, crash and fatality rates remain high.

Intersections are a particularly dangerous location for drivers, pedestrians and bicyclists. Intersection-related motor vehicle crashes accounted for 44.8% of all crashes and 21.5% of all fatal crashes in 2007 (Federal Highway Administration [FHWA], 2009). An earlier study (for the years 1994-1995) reported that the risk of a fatal crash for older drivers aged 65-69, compared with the risk for drivers aged 40-49 in the U.S. was 2.26 times higher for multiple-vehicle involvements at intersections and 1.29 times higher in all other situations (Preusser et al., 1998).These statistics demonstrate why intersection safety and understanding the mechanism of these crashes are priorities - and especially in considering differences across the life span. A study examining the characteristics of an estimated 787,236 motor vehicle intersection between 2005 and 2007, using data from the National Motor Vehicle Crash Causation Survey (NMVCCS), showed that 96% of intersection crashes had critical factors attributed to drivers, while in less than 3%, the factors were related to the vehicle or environment (NHTSA, 2010a,b).

While the proximities and conflicting trajectories of vehicles contribute greatly to the dangers associated with intersections, these are also areas that are often densely packed with critical driving-related (and often driving-unrelated) information that drivers must contend with. In short, intersections are complex features of the environment. Complexity can be influenced by the road geometry and lane configurations, presence and nature of traffic control devices, the volume of other road users and the density of other objects, buildings, advertisements, among many other elements. Drivers at intersections must therefore be attentive and vigilant with respect to traffic and other road users in their own and adjacent lanes as well as on cross roads, not to mention traffic control devices, signage and other road information. Various studies have examined the impact of intersections or driving scenes of varying complexity on drivers' visual scanning, decision-making and performance (e.g., Werneke and Vollrath, 2012; Edwards et al., 2003; Ho et al., 2001), all of which have some bearing on overall safety. The role of the complexity of intersections leading to increased crash risk is of particular concern among older adults (Braitman et al., 2007). Older adults tend to exhibit sensory, perceptual, cognitive and motor declines (e.g., Salthouse, 2004), all of which can impact their ability to deal with the complexities of intersections. Crash data corroborates some of these known deficiencies; older drivers have been identified as having a higher frequency of intersection crashes involving vehicles crossing paths prior to the collision, compared with their involvement in all crash types (Viano and Ridella, 1996). Moreover, in documenting critical driver errors involved in serious crashes in the NMVCSS, Cicchino and McCartt (2015) found that over 70% of older drivers' (aged 70 and over) surveillance errors involved attentional failures, such as looking but failing to see vehicles or traffic control devices.

After age 60, with a sharp increase after age 80, driver fragility and over-involvement in crashes is estimated to account for 34-45% of fatal crash risk (Li et al., 2003). The authors suggested that fragility is of "over-riding importance in explaining the increased fatality risk per unit of travel among older drivers" (p. 233), however excess crash involvement became a clear contributing factor among older drivers at ages 75-79.

There have been a number of studies that have utilized Fatality Analysis Reporting System (FARS) data that have reported important and actionable findings. Examples include studies of the effects of seating position, combined with restraint use and air bag status, on children's risk of dying in crashes (Braver et al., 1998; Berg et al., 1999), studies that examined the independent contribution of driver age and gender (Preusser et al., 1998; Stutts et al., 2009; Sifrit et al., 2011 and Tefft, 2012), and also studies examined crash and vehicle characteristics (Bedard et al., 2002; Li et al., 2003) that found significant differences by age and gender suggesting specific safety needs of older drivers and female drivers. However, much of the literature concerning older adults and intersection safety is based on older data and so is in need of updating in order to provide a more immediate picture of the state of driver safety at intersections. In this study, we sought to (1) provide an up-to-date analysis of age and gender differences in fatal intersection crash involvement rates in the US for the period of 2011-2014, and (2) identify which specific risk factors were associated with age-related risks in fatal intersection crashes. Given the aging of the U.S. population and the number of older drivers expected in the coming years, it is important to understand the factors leading to these crashes and to develop counter-measures to reduce their frequency and severity. Naturally, this effort is also relevant for helping older adults "age in place" in terms of accessibility, mobility, quality of life and safety.

2. Methods

2.1. Design and data sources

FARS data for all crash-related fatalities for the years 2011-2014 were downloaded from the National Highway Traffic Safety Administration (NHTSA) website (ftp://ftp.nhtsa.dot.gov/FARS). Individual FARS datasets were first merged by year and then pooled across all years. FARS has collected data based on a complete census of fatal traffic crashes within the 50 US states, the District of Columbia and Puerto Rico since 1975 (NHTSA, 2013a,b). To assure high quality and consistency of these data, NHTSA has a cooperative agreement with agencies in each state to provide comprehensive information on all qualifying fatal crashes in the state (managed by 10 regional offices). FARS data are obtained from police accident reports, death certificates, state vehicle registration files, coroner/medical examiner reports, state driver licensing files, hospital medical reports, state highway department data, emergency medical service reports, vital statistics and other state records. FARS state analysts are responsible for gathering, translating, and transferring data to the National Center for Statistics and Analysis (NCSA) in a standard format. The FARS data do not include personal identifying information (e.g., names, addresses, and social security numbers) and are made available to the public fully conforming to the Privacy Act. The current study was approved by the New England Independent Review Board (NEIRB).

2.2. Inclusion criteria

For the current analysis, inclusion criteria for a crash from FARS to be utilized include: (1) crash must have involved a motor vehicle traveling on a traffic way customarily open to the public, (2) must have resulted in the death of an occupant of a vehicle or a non-occupant within 30 days (720 h) of the crash, and (3) the vehicle involved was in transport and the driver was present at the time of the crash.

2.3. Data analysis

The annualized number of fatal intersection crashes during the period of 2011-2014 (per 100,000 licensed drivers) was

calculated across five-year age strata. The numerator was the sum of crash counts from the FARS data within each category of age, and the denominator was the number of total licensed drivers, based on FHWA estimates for 2013 (the most recent year of data available) for each US state and age group (FHWA, 2013). Annualized estimates were derived as four-year fatal crash averages. Characteristics of each crash and the drivers involved were summarized using descriptive statistics (e.g., N, percentage, cumulative percentage, mean, standard error, and 95% CIs).

2.4. Multivariate modelling

2.4.1. Fatal intersection crash risk

Multivariate logistic regression models were used to estimate the risk (adjusted odds-ratios and 95% Wald confidence intervals) of involvement in a fatal intersection crash compared to a non-intersection crash across categories of age. Drivers aged 20-24 were used as the referent group, as they had the highest annualized count of drivers involved in fatal intersection crashes). Models were stratified by gender and controlled for statistically significant factors (p-value<0.05) that included the time of day, day of week, type of road, type of trafficway, road alignment, weather conditions, lighting conditions, and whether the driver was deemed at fault for the crash or not (see Induced exposure analysis below for details regarding this determination). Other variables were evaluated, but not included in the final model, due to either the high number of missing values (e.g., travel speed at the time of the crash), or variables that were evaluated as co-linear (and less-populated) with other variables in the model.

The form ofthe multivariate logistic regression equation used in our analyses, was estimated using the maximum likelihood method (Hosmer and Lemeshow, 2000) as follows:

logit (p (x)) = log

L1 -p (x)J

= bo + biXi +b2X2... + bjXj

In this model, the logits (log odds) is the result of the b coefficients (slope values) of the regression equation, where p(x) is the probability of driver involvement in a fatal intersection crash, b0 is a constant, and each b,Xj is coefficient predictor value.

All statistical analyses were conducted with SAS version 9.2 (SAS Inc., Cary, North Carolina).

2.4.2. Induced exposure analysis

An induced exposure analysis was conducted to calculate crash involvement ratios (CIRs) using methods previously published by Reinfurt et al. (2000) and Stutts et al. (2009), as well as 95% confidence intervals of each CIR (Altman, 1991). In our analysis, we selected all two-vehicle crashes, with non-missing age and gender, and assigned whether each driver was "at-fault" or "not-at-fault", based upon whether they had a contributing factor or moving violation (a complete list can be found in Appendix A of Stutts et al., 2009) during the course of the crash that was recorded in the FARS database. Common driving related factors (coded using the FARS analytical codebook, 2013) included citations such as "failure to keep in proper lane", "failure to yield right-of-way", "operating the vehicle in an erratic, reckless or negligent manner", "operating at erratic or suddenly changing speeds", and "careless driving" and moving violations such as "driving while intoxicated (alcohol or drugs) or BAC above limit", "Inattentive, careless, improper driving", "speeding" and "failure to yield or stop". The next step was to then calculate a CIR and 95% confidence interval for each age category, and among the exposure of interest which in this study is intersection crashes. CIRs are calculated as simply the ratio of at-fault to not-at-fault crashes. If a CIR value is greater than 1.0, the more likely that age-group is over-involved in an intersection

crash, conversely where a value less than 1.0 that would suggest an under-involvement.

2.4.3. Age differences in fatal intersection crashes

To evaluate the significant differences in factors associated with younger (<65) and older (65+ years) drivers involved in fatal intersection crashes, drivers were split into two groups according to these age ranges. This age group cut-point was chosen as it was the first age category in which either gender (in this case females) showed a statistically significant increase in fatal intersection crash risk (see Table 3). A multivariate Poisson regression model was then used to compare older and younger drivers with respect to differences in factors related to their intersection crashes that simultaneously included time of day, day of week, type of road, type of trafficway, road alignment, weather conditions, lighting conditions, and whether the driver was deemed at fault for the crash.

3. Results

3.1. Crash and driver characteristics

For the 4-year period (2011-2014), there were a reported 120,809 fatal crashes in the US, involving 178,489 drivers of vehicles (referred to as "fatal crash involvements"), of which 48,733 (27.8%) were identified as occurring at an intersection (see Table 1A). Among these intersection crashes, the majority occurred primarily at four-way (64.2%), followed by T-intersections (32.6%) and Y-intersections (2.5%) (Table 1B). Most were on two-way, non-divided trafficways (61.3%), defined as a highway on which vehicles travel in opposite directions but the opposing travel lanes are not physically separated by more than an easily traversable centerline (Federal Motor Carrier Safety Administration), followed by crashes occurring on two-way, divided, trafficways with an unprotected (painted <4 feet) median (21.8%). The number of travel lanes prior to the vehicle's critical pre-crash event was most frequently two (65.6%) or four (18.0%). Most roadways were arterial, with 44.3% in urban areas, and 21.3% in rural. Lighting conditions at the time of the crash were during daylight (61.4%) or dark conditions that were lighted (22.1%), and during primarily clear or cloudy conditions (92.2%). Visual traffic controls (e.g., control signals, signs, and devices) were present in 57.6% of all intersections crashes, and the speed limit was most frequently 50+mph (40.1%), followed by 35-45 mph (29.2%).

With regards to driver characteristics (see Table 1A), the mean (95% confidence interval) age of drivers involved in fatal intersection crashes was 44.6 (44.5-44.8) years old, whereas for non-intersection fatal crashes, drivers were almost three years younger at 41.8 (41.7-41.9). Males accounted for 71.0% of all drivers involved in fatal intersection crashes (see Table 1B), and according information from the police-reports, 40.9% of drivers were deemed at fault for the crash. Peak days for intersection crashes were Friday and Saturday (16.3% and 16.1%, respectively), and peak times of the day when crashes occurred were from 12:00 p.m. to 7:00 p.m. (43.8% of all crashes, data not shown).

3.2. Fatal intersection crash rates by age

Fatal intersection crash counts and annualized rates across five-year categories of age for the period of 2011-2014 are presented in Table 2 and in Fig. 1. Over this 4-year period, there were 47,997 drivers involved in fatal intersection crashes (with known driver age), or an average of 11,999 driver intersection crashes involving a fatality per year. The number of licensed drivers in 2013 (most recent available data for this period) was 212,159,728 according to the US DOT. Fatal intersection crash rates

Table 1A

Age of driver involved by type of fatal crash (FARS 2011-2014).

Age of Driver1

Type of Crash Drivers Involved (N) Mean 95% Confidence Interval

Intersection Crash2 48,733 44.64 44.47-44.82

Non-Intersection Crash 129,430 41.81 41.71-41.90

Intersection Status Unknown 326 42.70 -

Total 178,489 42.58 42.50-42.67

Traffic Circle and Roundabout, and Five-Point+ Intersections.

1 Age was missing for3052 drivers (Non-intersection = 2316, Intersection» 736).

2 Intersection crashes include Four-Way Intersection, T-Intersection, Y-Intersection.

per 100,000 licensed drivers was highest for drivers aged 85 or older (9.89/100,000 licensed drivers), followed by drivers under 20 years of age (8.93/100,000), the latter possibly reflecting the dangers faced by young and inexperienced drivers (e.g., Foss et al., 2011 ).The average fatal intersection crash rate across all categories of age was 5.66/100,000 licensed drivers. In general, the shape of these rates (see Fig. 1) for drivers followed a bathtub curve, which was highest at both extremes of age categories, sloping toward a constant rate for middle-aged drivers (below average), with rates steadily increasing again at approximately 70 years of age.

3.3. Multivariate modelling

3.3.1. Fatal intersection crash risk

Table 3 presents estimated adjusted odds-ratios from multivari-ate logistic regression analysis comparing fatal intersection crash risk by age and gender. This analysis controlled for all significant risk factors associated with fatal intersection crashes included in the multivariate model that included the time of day, day of week, type of road, type of trafficway, road alignment, weather and lighting conditions, and at fault determination for the crash. For males, 123,581 crashes were used that did not have missing values for whether the crash occurred at an intersection, age, or any of the covariates in the model, for females, 43,299 crashes were analyzed.

There was no significant increase in the likelihood of a fatal intersection crash for males until ages 60-64 (compared to males ages 20-24), OR =1.08 (95% CI; 1.01-1.16). However risk steadily increased, and remained significant with increasing age, peaking

for male drivers ages 85 or older; OR =1.94 (95% CI; 1.74-2.16). Similarly, for female drivers, significant fatal intersection crash risk also began at ages 60-64; OR = 1.15 (95% CI; 1.03-1.29), and steadily rose with increasing age until reaching a maximum risk at age 85 or older; OR= 1.92 (95% CI; 1.64-2.25).

3.3.2. Induced exposure analysis

Crash involvement ratio's (CIRs) are presented in Table 4 for all two-vehicle crashes where fault of both involved drivers could be determined using the method previously described. For the period of 2011-2014, there were 120,809 fatal US crashes involving 178,489 drivers of vehicles. Determination of driver of vehicle at fault was made in 120,442 or 99.7% of crashes. In addition, driver age and sex could not be missing to be included in these analyses. Among these crashes, there were 76,391 two-vehicle crashes with non-missing driver age and gender 30,227 (39.6%) of which occurred at an intersection. Similar to the results of the multiple logistic regression, both male and female drivers under 20 years of age were over-involved in fatal intersection crashes, 1.13 and 1.29, respectively. However, drivers from the ages of 20-54 years old, were underrepresented in fatal intersection crashes. Except for females, who had some age categories with small cell sizes, CIRs increases steadily with increasing age after 55 years old and reached a maximum of 1.61 for males and 1.63 for females at ages 85+, suggesting significant over-involvement in fatal intersection crashes in this age group.

Fig. 1. Annualized intersection fatality rate per 100,000 licensed drivers (FARS 2011-2014).

Table 1B

Fatal intersection crash characteristics for involved drivers (N = 48,733).

Characteristic

Percentb (%)

Gender of Driver Male Female Total

Day of Week Sunday Monday Tuesday Wednesday Thursday Friday Saturday Total

Type ofIntersection Four-Way Intersection T-Intersection Y-Intersection Traffic Circle Roundabout Five Point, or More L-Intersection Total

Roadway Type Rural-Arterial

Rural-Local Road or Street or Collector Urban-Arterial

Urban-Local Road or Street or Collector Total

Trafficway Description Two-Way, Not Divided

Two-Way, Divided, Unprotected (Painted >4 Feet) Median Two-Way, Divided, Positive Median Barrier One-Way Trafficway

Two-Way, Not Divided, Continuous Left-Turn Lane Total

Number of Lanes One lane Two lanes Three lanes Four lanes Five or more lanes Total

Traffic Controls1 No Traffic Controls Present Visual Traffic Control Present Total

Speed Limit 0-25 mph 25-35 mph 35-45 mph 50+ mph Total

Lighting Condition Daylight

Dark - Not Lighted Dark - Lighted Dawn Total

Weather Condition Clear/Cloudy Rainy

Snow/Sleet/Hail Other (Fog, Smoke, Sand) Total

At Fault Status Driver Not at Fault Driver at Fault Total

34,138 13,927 48,065

6341 6424 6515 6695 6963 7939 7856 48,733

31,265

15,883

48,733

10,294

21,438

48,431

29,564

10,503

48,254

31,699

48,350

20,646 28,087 48,733

11,309

14,206

19,521

48,733

29,888

10,771

48,647

44,702

48,491

28,782 19,926 48,708

71.02 28.98 100.00

13.01 13.18

13.37 13.74 14.29 16.29 16.12 100.00

64.16 32.59 2.50 0.06 0.11 0.48 0.09 100.00

21.25 16.55 44.27 17.93 100.00

61.27 21.77 7.15 2.06 7.76 100.00

0.95 65.56 9.81 18.01 5.66 100.00

42.37 57.63 100.00

7.59 23.21 29.15 40.06 100.00

61.44 12.25 22.14 4.17 100.00

100.00

59.09 40.91 100.00

1 Visual traffic controls include traffic signals, and traffic or regulatory signs. b Unknowns and missing values are not represented in the percentages.

Table 2

Fatal intersection crash rates by age (FARS 2011-2014).

Fatal Intersection Crashes 2011-141 2013 Number of Annualized Fatal Intersection

Licensed Drivers1 Crash Rate per 100,000

Licensed Drivers

Driver Ages 4-Year Total Annualized Total Annualized Percent Total

<20 3209 802 6.69 8,982,187 8.93

20-24 5677 1419 11.83 17,668,252 8.03

25-29 4860 1215 10.13 18,341,228 6.62

30-34 4200 1050 8.75 18,356,676 5.72

35-39 3626 907 7.55 17,273,905 5.25

40-44 3763 941 7.84 18,744,887 5.02

45-49 3884 971 8.09 19,299,319 5.03

50-54 4020 1005 8.38 20,607,806 4.88

55-59 3559 890 7.42 19,398,515 4.59

60-64 2917 729 6.08 16,656,737 4.38

65-69 2203 551 4.59 13,227,162 4.16

70-74 1787 447 3.72 9,307,315 4.80

75-79 1518 380 3.16 6,420,197 5.91

80-84 1398 350 2.91 4,398,882 7.95

85+ 1376 344 2.87 3,476,660 9.89

Total 47,997 11,999 100.00 212,159,728 5.66

1 U.S. Department of Transportation, Federal Highway Administration (FHWA, Office of Highway Policy Information, Highway Statistics Series, Table DL-22-Highway Statistics 2013). http://www.fhwa.dot.gov/policyinformation/statistics/2013/dl22.cfm (accessed 04/01/2016).

3.3.3. Differences in factors related to age

Table 5 compares differences in factors related to the intersection crashes of younger (<65) and older (65+ years) drivers over the 4-year period. There were statistically significant differences (p-value < 0.05) in several factors. With regards to the time of day and lighting conditions, for older driver crashes, peak crash times were

at 12:00-04:00 p.m., with 79.8% occurring during daylight hours, whereas younger drivers peak crash times were at 03:00-07:00 p.m., with 58.4% occurring during daylight. The day of the week was also significantly different. For older drivers peak crash days were fairly uniform from Tuesday-Friday, whereas for younger drivers, frequencies were highest on Friday and Saturday.

Table 3

Multivariate logistic regression, adjusted odds ratio (risk) estimates of a fatal intersection crash by age and gender (FARS 2011-2014).

Adjusted1 Odds Ratio Estimates for Males

Age Category (Referent, Ages 20-24) Point Estimate 95 Percent Wald Confidence Limits

Ages<20 0.987 0.923 1.054

Ages 25-29 1.011 0.955 1.071

Ages 30-34 1.034 0.974 1.098

Ages 35-39 1.040 0.977 1.107

Ages 40-44 1.025 0.963 1.090

Ages 45-49 1.041 0.979 1.107

Ages 50-54 1.023 0.962 1.087

Ages 55-59 1.051 0.986 1.119

Ages 60-64 1.083* 1.012 1.159

Ages 65-69 1.038* 0.961 1.121

Ages 70-74 1.203* 1.105 1.309

Ages 75-79 1.348* 1.226 1.482

Ages 80-84 1.450* 1.309 1.606

Ages 85+ 1.935* 1.736 2.158

Adjusted1 Odds Ratio Estimates for Females

Age Category Point Estimate 95 Percent

(Referent, Ages 20-24) Wald Confidence Limits

Ages<20 1.038 0.937 1.150

Ages 25-29 1.027 0.934 1.130

Ages 30-34 0.986 0.892 1.089

Ages 35-39 0.960 0.865 1.066

Ages 40-44 1.043 0.941 1.156

Ages 45-49 1.097 0.990 1.216

Ages 50-54 1.083 0.977 1.200

Ages 55-59 1.081 0.971 1.204

Ages 60-64 1.151* 1.026 1.290

Ages 65-69 1.294* 1.147 1.461

Ages 70-74 1.206* 1.057 1.376

Ages 75-79 1.548* 1.345 1.782

Ages 80-84 1.861* 1.605 2.158

Ages 85+ 1.921* 1.640 2.250

1 Multivariate logistic models includes 123,581 male crashes and 43,299 female crashes that had no missing values for intersection, age, or any of the covariates adjusted for in the model which include time and day of week, type of road and trafficway, road alignment, weather and lighting conditions, and at fault determination for the crash. * Statistically significant P-value at <0.05.

Table 4

Induced exposure analysis: at-fault crash involvement ratios (CIRs) and 95% confidence intervals for all two-vehicle crashes1 (FARS 2011-2014).

AH Crashes Intersection Involvement Crashes

Age Category Two-Vehicle Crashes Two-Vehicle Crashes Male Female

(N = 76,391) (N = 30,227) (N = 21,442) (N = 8783)

Ages<20 1.17 (1.09-1.26) 1.19 (1.04-1.37) 1.13 (0.96-1.33) 1.29 (1.00-1.67)

Ages 20-24 1.01 (0.96-1.06) 0.90 (0.81-0.99) 0.92 (0.82-I.03) 0.84 (0.70-1.02)

Ages 25-29 0.96 (0.91-1.02) 0.87 (0.78-0.96) 0.89 (0.79-I.01) 0.80 (0.65-0.99)

Ages 30-34 0.93 (0.88-0.98) 0.87 (0.78-0.98) 0.86 (0.76-0.98) 0.93 (0.74-1.17)

Ages 35-39 0.88 (0.83-0.94) 0.84 (0.75-0.95) 0.85 (0.74-0.97) 0.84 (0.66-1.08)

Ages 40-44 0.92 (0.87-0.98) 0.97 (0.85-1.09) 0.97 (0.84-1.11) 0.98 (0.76-1.26)

Ages 45-49 0.93 (0.87-0.98) 0.94 (0.83-1.06) 0.92 (0.80 -1.05) 1.04 (0.80-1.34)

Ages 50-54 0.88 (0.83-0.93) 0.92 (0.82-1.03) 0.92 (0.81-1.05) 0.92 (0.72-1.17)

Ages 55-59 0.99 (0.93-1.06) 1.07 (0.94- 1.22) 1.09 (0.94-1.26) 1.04 (0.79-1.36)

Ages 60-64 0.95 (0.89-1.02) 0.98 (0.85-1.12) 1.06 (0.90-1.25) 0.79 (0.62-1.02)

Ages 65-69 1.03 (0.95-1.12) 1.21 (1.02-1.43) 1.17 (0.96-1.42) 1.28 (0.93-1.77)

Ages 70-74 1.27 (1.15-1.41) 1.35 (1.12-1.64) 1.32 (1.05-1.65) 1.45 (0.99-2.11)

Ages 75-79 1.40 (1.25-1.58) 1.39 (1.13-1.71) 1.41 (1.09-1.82) 1.29 (0.91-1.83)

Ages 80-84 1.58 (1.38-1.80) 1.36 (1.10-1.67) 1. 1.54 (1.16-2.03) 1.06 (0.77-1.46)

Ages 85+ 1.94 (1.67-2.25) 1.64 (1.31-2.06) 1.61 (1.22- 2.13) 1.63 (1.10-2.41)

1 Includes all two-vehicle crashes where "at fault" of both involved drivers involved in the crash could be determined, and driver age and gender was non-missing.

The presence of visible traffic controls (e.g., signals, signs) at the crash intersection was significantly higher for older drivers (65.7%), compared to younger drivers (56.0%), as was the speed limit of the roadway where the intersection crash occurred. However estimated average crash speed (mph) was significantly lower for older driver crashes compared to younger drivers; 35.0 vs. 28.0 mph, respectively. Significant differences were also observed roadway type, where older driver crashes occurred more frequently on rural roads, and younger on urban roads, and also number of lanes. In comparing whether the driver was deemed "at fault", older drivers were significantly more likely to be at fault in fatal intersection crashes, than younger drivers (55.9% vs. 38.0%), respectively.

4. Discussion

The results of this study provide up-to-date data on motor vehicle fatal intersection crash risks and rates by age and gender in the US pooled data from the FARS for the period of 2011-2014. The major findings show that during this period, among the reported 120,809 fatal accidents in the US that involved 178,489 drivers ("fatal crash involvements), 48,733 (28%) were drivers in fatal intersection crashes. Most crashes were at four-way intersections, on two-way, non-divided trafficways, and equally split between daylight and night conditions. Drivers were typically male, 45 years of age; however similar to previous descriptions of the age-distribution of fatal and non-fatal crashes (Tefft, 2012), the shape of the age distribution of fatal intersection crash rates per 100,000 licensed drivers followed a bathtub curve, with the highest rates at the ends of the age distribution, and below-average rates for middle-aged drivers. Teenage drivers (i.e. drivers < 20 years old) had the second highest annualized intersection fatality rate of 8.93, as well as significant over-involvement in all two-vehicle fatal crashes and at intersections. Although in the past two-decades in the US, we have experienced significant declines in fatal and non-fatal teenage crashes (McCartt and Teoh, 2015), preventable crash-related morbidity and mortality rates for this age group remain high, accounting for one-third of all teenage deaths (Minino, 2010).

Although drivers under the age of 65 were involved in 82.7% of all fatal intersection crashes, older female drivers (65-69 years of age), and older male drivers (70-74 years of age), were 25% and 14%, respectively, more likely to be involved in a fatal intersection crash compared to drivers aged 20-24 (of the same gender), after controlling for all significant factors that included time of day, day of week, type of road, type of trafficway, road alignment, weather

conditions, lighting conditions, or being deemed at fault for the crash. By the age of 85 years, fatal intersection crash risk for female and male drivers was almost doubled, however these adjusted odds ratios are somewhat attenuated compared to those presented by Preusser et al. (1998) that were based on fatal crash data from 1994 and 1995. They observed that drivers aged 65-69 were 2.3 times more likely to be involved in multiple-vehicle crashes at intersections while those aged 85+ were 10.6 times more likely. However, it is important to underscore that the Preusser et al. (1998) analysis used a reference group (drivers aged 40-49) with a lower crash risk in their study, leading to potentially higher age-related relative crash risks. Additionally, these two studies did not statistically control for similar factors, thus the rates cannot be directly compared. The reduction in relative fatal crash risk could be due, in part also to more advanced safety features and countermeasures that help offset the inherent frailty of older drivers (e.g., Li et al., 2003; Kahane, 2013; and Cicchino, 2015).

In comparing the results of our induced exposure analysis and associated CIRs for two-vehicle crashes at intersections to another study that used the FARS data for the years 2002-2006 (Sifrit et al., 2011), we generally found similar results. Male and female drivers younger than 20 years old were over-involved in fatal intersection crashes, and males and female drivers from the ages of 20 to 54 years old were under-involved. CIRs steadily increased after the age of 55 years old. However, our study examined the years 2011-2104, and the maximum CIR was 1.61 for males and 1.63 for females at ages 85+, whereas in the Sifrit et al. (2011) study CIRs were 5.4 for ages 80+ years. Differences between the two studies at this highest age group may be due to small cell-size variability, bin-size differences, or other exposure distribution cohort differences.

We found that the significant differences in factors related to intersection crashes of younger (<65) and older (65+ years) drivers were the time of day, lighting conditions, day of week, weather conditions, type of road, number of lanes, presence of visible traffic controls, speed limit, estimated driving speed, and being deemed at fault for the crash. These results were similar to those from a previous study by Chen et al. (2012) of intersection crashes involving casualties in Victoria, Australia for the period 2000-2009. In that study, they reported that seven factors that were significantly associated with the severity of intersection crashes that included driver age and gender, speed zone, traffic control type, time of day, crash type, and seat belt usage. However, the current study focused on fatal crashes which did not allow for an examination across a range of severity.

Table 5

Age differences in fatal intersection crashes (FARS 2011-2014).

Intersection Crashes with Known Driver Age (N = 47,997)

Ages < 65 (n=39,715) N %a Ages 65+ (n=8282) N %a P-value1

Gender

Male 28,672 72.20 5405 65.26 <0.0001

Female 11,038 27.80 2877 34.74

Day of Week

Sunday 5274 13.28 940 11.35 0.0101

Monday 5226 13.16 1113 13.44

Tuesday 5153 12.98 1289 15.56

Wednesday 5347 13.46 1267 15.30

Thursday 5580 14.05 1304 15.75

Friday 6502 16.37 1297 15.66

Saturday 6633 16.70 1072 12.94

Type of Intersection

Four-Way Intersection 25,371 63.88 5402 65.23 0.3894

T-Intersection 13,016 32.77 2642 31.90

Y-Intersection 1025 2.58 180 2.17

Traffic Circle 23 0.06 7 0.09

Roundabout 46 0.12 10 0.12

Five Point, or More 194 0.49 35 0.42

L-Intersection 40 0.10 6 0.07

Roadway Type

Rural-Arterial 8222 20.83 2039 24.78 <0.0001

Rural-Local Road or Street or Collector 6553 16.60 1418 17.24

Urban-Arterial 17,510 44.36 3485 42.36

Urban-Local Road or Street or Collector 7191 18.22 1285 15.62

Trafficway Description

Two-Way, Not Divided 23,840 60.58 5341 65.25 0.7730

Two-Way, Divided, Unprotected (Painted >4 Feet) Median 8723 22.17 1645 20.10

Two-Way, Divided, Positive Median Barrier 2845 7.23 514 6.28

One-Way Trafficway 848 2.16 117 1.43

Two-Way, Not Divided With a Continuous Left-Turn Lane 3096 7.87 568 6.94

Number of Lanes

One lane 365 0.93 87 1.06 0.0003

Two lanes 25,570 64.86 5764 70.20

Three lanes 3946 10.01 697 8.49

Four lanes 7238 18.36 1287 15.67

Five or more lanes 2305 5.85 376 4.58

Traffic Controls2

No Controls Present 17,457 43.96 2841 34.30 <0.0001

Visual Traffic Control Present 22,258 56.04 5441 65.70

Level 31,010 78.08 6495 78.42 0.1418

Graded 8705 21.92 1787 21.58

Pavement Type

Blacktop, Bituminous, or Asphalt 27,887 70.22 5966 72.04 0.2975

Non-Blacktop 11,828 29.78 2316 27.96

Roadway Alignment

Straight 36,121 93.04 7617 94.23 0.4430

Curved 2701 6.96 466 5.77

Speed Limit

0-25 mph 2893 7.28 705 8.51 0.0019

25-35 mph 9211 23.19 1818 21.95

35-45 mph 11,683 29.42 2328 28.11

50+ mph 15,928 40.11 3431 41.43

Estimated Driving Speed (MPH)

Mean ± Std. Error of Mean 34.97 ±0.20 27.96 ±0.37 <0.0001

Lighting Condition

Daylight 23,158 58.42 6602 79.82 <0.0001

Dark - Not Lighted 5259 13.27 577 6.98

Dark - Lighted 9479 23.91 839 10.14

Dawn 1748 4.41 253 3.06

Table 5 (Continued)

Intersection Crashes with Known Driver Age (N = 47,997)

Ages < 65 Ages 65+ P-value1

(n = 39,715) (n=8282)

N %a N %a

Weather Conditions

Clear/Cloudy 36,448 92.24 7628 92.45 0.8715

Rainy 2293 5.80 494 5.99

Snow/Sleet/Hail 280 0.71 64 0.78

Other (Fog, Smoke, Sand) 492 1.25 65 0.79

Driver at Fault

No 24,592 61.95 3649 44.08 <0.0001

Yes 15,102 38.05 4629 55.92

a Unknowns and missing values are not represented in the percentages.

1 P-value based on Poisson Regression model comparing age (<65 vs. 65+ years) controlling for all other variables in the model.

2 Visual traffic controls include traffic signals, and traffic or regulatory signs.

As previously described, important factors related to higher fatal crash rates among older drivers was reported to be fragility and over-involvement in crashes (Li et al., 2003). Although these two factors may be difficult to de-confound, one recent study sought to quantify the most frequent type(s) of error made by older drivers in serious crashes using data from NMVCCS (Cicchino and McCartt, 2015). That study compared critical errors made by drivers aged 70 years and older to errors made by drivers aged 35-54 (middle-aged) and found that among the older group critical driver error was the reason for the large majority (97%) of crashes. These errors included inadequate surveillance (33% vs. 22% for middle-aged), misjudgment of the gap length between vehicles or the other vehicle's speed, illegal maneuvers, medical events, and concentration. However, it was also found that 71% of older drivers' surveillance errors (compared to 40% of middle-aged) were due to looking but failing to see another vehicle, or traffic control, and occurred while turning left at intersections.

With regards to the over-involvement of older drivers in crashes, one potential hypothesis is that complex intersections can present drivers with too much dynamic information to be processed in the time available. Drivers must selectively prioritize and attend to only certain areas of interest and, thus, run the risk of missing (potentially critical) information that appears in unattended - and possibly unexpected - areas. As such, drivers might make poorer decisions in complex intersections, based on incomplete information, and stand a higher risk of motor vehicle crash involvement as a result. Previous work in laboratory settings has documented age-related differences in visual scanning of road information, the detection of critical (and changing) information, and the quality of subsequent decisions (e.g., Maltz and Shinar, 1999; Pringle et al., 2001; Caird et al., 2005).

There are several limitations of our study, and although FARS is considered a complete annual census of fatal crashes on US public roadways, some of these limitations may influence the findings of this study (Braver et al., 1998). An important one is the underreporting of the prevalence or incomplete data on driver behavioral factors that are used in this study for both the multiple logistic and Poisson regressions and the induced exposure analyses. For example, the variable calculated to determine whether a driver is 'at fault' integrates driving related factors, based upon driver citations such as "failure to keep in proper lane" and "operating the vehicle in an erratic, reckless or negligent manner", and moving violations such as "Inattentive, careless, improper driving", and "speeding". Often these driving behaviors are subjectively observed and determined as citations, or simply unobserved and thus un-reported. Although this may be likely, we believe that for most cases, these reporting biases are non-differential on whether the crash occurred at an intersection or not, or regardless of the age of the drivers. However, we recognize that for some factors, for example

identifying fatigued drivers, there may be clear investigating police officer biases in the assessment of crashes that occur at different times of the day (e.g., more likely under suspicion for a 3:00 a.m. crash). In addition, we caution on interpreting the results of the Poisson regression analyses that compare age differences in factors related to fatal intersection crashes. These comparisons are made between proportions for these two groups without knowledge of the underlying exposure distributions. For example, a significant difference was observed in roadway type where older drivers crashed more frequently at intersections on rural-arterial roads compared to younger drivers. However it is not known if older drivers more frequently drive on rural roads compared to urban roads in these FARS data.

5. Conclusions and implications for prevention

The primary aim of this study was to update and document the incidence and characteristics of motor vehicle crashes occurring at intersections as well as the risk of involvement (using adjusted odds ratios) for different driver demographics and environmental characteristics. The results of this study serve to underscore several trends that are worthy of further investigation as well as targeted remediation. One of the most noteworthy outcomes, when considering the annualized rates of fatal intersection crashes (per licensed driver), concerns the age effects at both ends of the lifespan. As documented in several studies, young drivers are at high risk of motor vehicle crash and of dying in those crashes (e.g., Mayhew et al., 2003; McCartt et al., 2003). While factors related to the physical, psychological and social development will be ever-present in teens and young drivers, there are many potential inroads towards ameliorating the outlook for this group, including (but not limited to) enhanced risk and hazard anticipation training, expanded graduated driver's licensing requirements, and new onboard safety monitoring systems that can be used for behavior modification and coaching, to name a few. After controlling for several factors in our study, such as time of day, and "at-fault" violations (e.g., speeding, alcohol and drug impairment), the adjusted risk of young drivers involvement in fatal intersection crashes was similar to that of middle-aged drivers; it follows that these factors represent a potentially important starting point (or a continued focal point) for targeted remediation (see e.g., Mann et al., 1986).

The current study's accounting of fatal intersection crashes provides an important spot check for road safety in the US; however, there remains considerable work to be done. Older adults face a number of challenges associated with natural aging, including sensory, perceptual, cognitive and motor declines (e.g., Salthouse, 2004). Although these declines may be inevitable, medical and pharmacological technologies and developments have raised the quality of treatment in older adults and, consequently, have led to

increases in life span, including increases in the driving lifespan. Driving and the mobility and freedom it affords have a tremendous impact on quality of life, and the decision to cease driving is difficult (Dugan, 2006). As with younger drivers, expanded or renewed training approaches, new approaches to licensing renewals, and safety-based technological advances (including vehicle-based automation and vehicle to infrastructure applications) are all viable avenues toward improving the safety outlook for older adults.

References

Altman, D.G., 1991. Practical Statistics for Medical Research. Chapman & Hall, London.

Bedard, M., Guyatt, G.H., Stones, J.J., Hirdes, J.P., 2002. The independent

contribution of driver, crash: and vehicle characteristics to driver fatalities. Accid. Anal. Prev. 34, 717-727. Berg, M., Cook, L., Corneli, H., Vernon, D., Dean, M., 1999. Effect of seating position and restraint use on injuries to children in motor vehicle crashes. Pediatrics 106 (4), 831-835.

Braitman, K.A., Kirley, B.B., Ferguson, S., Chaudhary, N.K., 2007. Factors leading to

older drivers' intersection crashes. Traffic Inj. Prev. 8 (3), 267-274. Braver, E., Whitfield, R., Ferguson, S., 1998. Seating positions and children's risk of

dying in motor vehicle crashes. Inj. Prev. 4,181-187. Caird, J.K., Edwards, C.J., Creaser, J.I., Horrey, W.J., 2005. Older driver failures of attention at intersections: using change blindness methods to assess turn decision accuracy. Hum. Factors: J. Hum. Factors Ergon. Soc. 47 (2), 235-249. Chen, H.1, Cao, L., Logan, D.B., 2012. Analysis of risk factors affecting the severity of

intersection crashes by logistic regression. Traffic Inj. Prev. 13 (3), 300-307. Cicchino, J.B., McCartt, A.T., 2015. Critical older driver errors in a national sample of

serious U: S. crashes. Accid. Anal. Prev. 80, 211-219. Cicchino, J.B., 2015. Why have fatality rates among older drivers declined? The relative contributions of changes in survivability and crash involvement. Accid. Anal. Prev. 83, 67-73. Dugan, E., 2006. The Driving Dilemma: The Complete Resource Guide for Older

Drivers and Their Families. Edwards, C.J., Creaser, J.I., Caird, J.K., Lamsdale, A.M., Chisholm, S.L., 2003. Older and younger driver performance at complex intersections: implications for using perception-response time and driving simulation. Proceedings of the Second International Driving Symposium on Human Factors in Driver Assessment, Training and Vehicle Design, 33-38. Federal Highway Administration, U.S. Department of Transportation, "The National Intersection Safety Problem". Publication FHWA-SA-10-005, March 2008. http://safety.fhwa.dot.gov/intersection/other_topics/fhwasa10005/brief_2.cfm (accessed 11.10.16.). Federal Highway Administration, U.S. Department of Transportation, (FHWA, Office of Highway Policy Information, Highway Statistics Series, Table DL-22-Highway Statistics 2013).

http://www.fhwa.dot.gov/policyinformation/statistics/2013/dl22. cfm (accessed 04.01.16.). Foss, R.D., Martell, C.A., Goodwin, A.H., O'Brien, N.P., Center U. H. S. R, 2011. Measuring Changes in Teenage Driver Crash Characteristics During the Early Months of Driving. AAA Foundation for Traffic Safety, Washington DC. Ho, G., Scialfa, C.T., Caird, J.K., Graw, T., 2001. Visual search fortraffic signs: the effects ofclutter, luminance, and aging. Hum. Factors: J. Hum. Factors Ergon. Soc. 43 (2), 194-207.

Kahane, C.J., 2013. Injury Vulnerability and Effectiveness of Occupant Protection Technologies for Older Occupants and Women. National Highway Traffic Safety Administration, Washington, DC (Publication No. DOT HS 811 766).

Li, G., Braver, E.R., Chen, L., 2003. Fragility versus excessive crash involvement as determinants of high death rates per vehicle-mile of travel among older drivers. Accid. Anal. Prev. 35, 227-235.

Maltz, M., Shinar, D., 1999. Eye movements of younger and older drivers. Hum. Factors: J. Hum. Factors Ergon. Soc. 41 (1), 15-25.

Mann, R.E., Vingilis, E.R., Leigh, G., Anglin, L., Blefgen, H., 1986. School-based programmes for the prevention of drinking and driving: issues and results. Accid. Anal. Prev. 18 (4), 325-337.

Mayhew, D.R., Simpson, H.M., Pak, A., 2003. Changes in collision rates among novice drivers during the first months of driving. Accid. Anal. Prev. 35 (5), 683-691.

McCartt, A.T., Teoh, E.R., 2015. Tracking progress in teenage driver crash risk in the United States since the advent of graduated driver licensing programs. J. Saf. Res. 53,1-9.

McCartt, A.T., Shabanova, V.I., Leaf, W.A., 2003. Driving experience, crashes and traffic citations of teenage beginning drivers. Accid. Anal. Prev. 35 (3), 311-320.

Minino, A., 2010. Mortality among teenagers aged 12-19 years: United States, 1999-2006. NCHS Data Brief, 1-8.

NHTSA, 2010a. Report to Congress NHTSA's Crash Data Collection Program: DOT HS 811 337. National Highway Traffic Safety Administration, Washington, DC (April 2010).

NHTSA, 2010b. Crash Factors in Intersection-Related Crashes: An On-Scene

Perspective: DOT HS 811 366. National Highway Traffic Safety Administration, Washington, DC (September 2010).

NHTSA, 2013a. Fatality Analysis Reporting System (FARS) Analytical User's Manual 1975-2011: DOT HS 811 693. National Highway Traffic Safety Administration, Washington, DC (September 2013.).

NHTSA, 2013b. Traffic Safety Facts: Research Note. 2012 Motor Vehicle Crashes: Overview: DOT HS 811 856. National Highway Traffic Safety Administration, Washington, DC.

Preusser, D.F., Williams, A.F., Ferguson, S.A., Ulmer, R.G., Weinstein, H.B., 1998. Fatal crash risk for older drivers at intersections. Accid. Anal. Prev. 30 (2), 151-159.

Pringle, H.L., Irwin, D.E., Kramer, A.F., Atchley, P., 2001. The role of attentional breadth in perceptual change detection. Psychon. Bull. Rev. 8 (1), 89-95.

Reinfurt, D.W., Stewart, J.R., Stutts,J.C., Rodgman, E.A., 2000. Investigations of Crashes and Casualties Associated With Older Drivers. Project G.8. Chapel Hill NC: University of North Carolina Highway Safety Research Center (November).

Salthouse, T.A., 2004. What and when of cognitive aging. Curr. Dir. Psychol. Sci. 13 (4), 140-144.

Sifrit, K.J., Stutts, J., Martell, C., Staplin, L., 2011. Intersection Crashes Among Drivers in Their 60s, 70s, and 80s. National Highway Traffic Safety Administration, Washington, DC (Publication No. DOT ^ 811 495).

Stutts, J., Martell, C., Staplin, L., 2009. Identifying Behaviors and Situations

Associated with Increased Cras Risk for Older Drivers Report No: DOT HS 811 093.

Tefft, B.C., 2012. Motor Vehicle Crashes, Injuries, and Deaths in Relation to Driver. Age: United States, 1995-2010. AAA Foundation for Traffic Safety.

Viano, D., Ridella, S., 1996. Significance of Intersection Crashes for Older Drivers (SAE Technical Paper 960457).

Werneke, J., Vollrath, M., 2012. What does the driver look at? The influence of intersection characteristics on attention allocation and driving behavior. Accid. Anal. Prev. 45, 610-619.