Are Pedestrian Crashes Becoming More Severe? A Breakdown of Pedestrian Crashes in Urban Tennessee

Abstract

Similar to the nationwide trend of declining pedestrian safety, pedestrian fatalities in Tennessee increased by 117 % between 2009 and 2019. During the same time, pedestrians involved in traffic crashes have only increased by 26 %, suggesting that the reason for the increased number of fatalities is that pedestrian crashes are becoming more severe. The media, and safety advocates speculate about the different scenarios that could be responsible for the decrease in pedestrian safety. Past studies have failed to integrate pedestrian involvement into their studies and confirm whether pedestrian crashes are becoming more severe. Our study collected all reported pedestrian crashes and their outcomes from crash data kept by the Tennessee police, thus providing fatality and involvement numbers. Excluding crashes on controlled access roads, we measured the pedestrian fatality rate (PFR) for essential variables for each year from 2009 to 2019. An upward PFR graph shows that pedestrian crashes are becoming more severe compared with the past. To substantiate our claims with adequate statistical significance, we employed separate Poisson regression models for the pedestrian fatality count and PFR associated with each attribute of pedestrian crashes. The results show that an increase in severity is linked with multilane urban arterials with speeds above 35 mph at midblock locations. The study results also confirm that cars are still the cause of the high fatality rate in Tennessee rather than sport utility vehicles or other large vehicles. We recommend that cities reduce design speeds to 35 mph, increase safe pedestrian crossing opportunities, and install more pedestrian-scale lighting infrastructure on urban arterials.

Keywords

pedestrians safety pedestrian fatality trend design visualization in transportation equity (justice)

One of the significant contributors to fatalities caused by injury in the U.S.A. is traffic crashes ( 1 ), which claim the lives of thousands of motorized and nonmotorized road users, including pedestrians, in the country each year. Nationally, pedestrian safety had improved for 40 years until the last decade, when pedestrian fatalities rose by 51% ( 2 ). With pedestrian deaths declining in other developed countries, this increase is unique to the U.S.A. The European Union observed a steady decline of 19% in overall pedestrian fatalities between 2010 and 2018 ( 3 ). In the United Kingdom and Australia, the trends were relatively constant during a similar period ( 4 , 5 ). Pedestrian deaths in the U.S.A. also increased against the backdrop of an overall decline in traffic deaths. This has resulted in the share of pedestrian deaths among overall traffic deaths growing from 12% to 17% between 2009 and 2019 ( 2 , 6 ).

Tennessee, like most U.S. states, follows the rising pedestrian fatality trend. Pedestrian deaths have more than doubled in Tennessee, with 156 deaths in 2019 compared with 72 in 2009. However, overall pedestrian involvement in crashes has increased only slightly, with a 26% rise from 1,687 to 2,126 during the same period ( 7 ). The huge disparity between the growth in pedestrian involvement and fatalities suggests that crashes involving pedestrians are becoming more severe. Reports on pedestrians from 2009 and 2019 produced by the National Highway Traffic Safety Administration reveal no significant change in pedestrian injuries nationwide, although there is a distinct trend in fatalities ( 2 , 6 ). However, a 51% national growth seems underwhelming compared with the figures in Tennessee. Figure 1 shows how pedestrian involvement and fatalities have progressed in the state during the last 13 years.

Figure 1.

Pedestrian fatality and involvement trends in Tennessee.

This study aims to assess the factors correlated with this increase to recommend policies that can reverse the trend of increasing fatalities. The findings here are probably similar to those for other states, but our study is unique in that most other studies use fatality-only (FARS) data. We include fatal and nonfatal crashes and aim to identify trends in the increasing fatality rate, particularly for the past several years.

Factors Determining Severe Outcome in a Pedestrian Crash

Years of research have identified the variables associated with fatal outcomes in pedestrian crashes. Some factors involve user behavior, and others involve the design of the road. Children ( 1 ), intoxicated adults ( 8 , 9 ), and elderly ( 10 ) pedestrians are the most vulnerable pedestrian groups. In part, this vulnerability is because of the developing, impaired, and declining cognitive and perceptual faculties of children, pedestrians under the influence, and aging pedestrians, respectively ( 11 – 14 ). Minority populations, including African Americans and other people of color, the homeless, and lower-income households, are other vulnerable pedestrian groups because of overexposure and their incompatibility with the infrastructure ( 15 – 17 ). In addition, pedestrian behavior such as midblock crossing ( 18 ) and walking under the influence are also known for producing fatal outcomes ( 19 ). On the other hand, driver behavior, such as distracted driving on curved and sloped roads, using communication devices on straight roads, and speeding, are linked with severe outcomes in pedestrian crashes ( 20 ).

An important factor dictating the severity of a pedestrian crash outcome is the transfer of kinetic energy, which is a function of impact speed and vehicle mass, from the vehicle involved to the pedestrian ( 21 ). The injury outcome is highly sensitive to speed, because, on average, a 1 km/h (0.6 mph) increase in impact speed can boost the probability of fatality by 11%. There is a 5% chance of pedestrian death at a vehicle speed of 30 km/h (18.6 mph), a 10% chance at 37 km/h (23 mph), a 50% chance at 59 km/h (36.7 mph), a 75% chance at 69 km/h (42.9 mph), and a 90% chance at 80 km/h (49.7 mph) ( 22 ). However, it is difficult to measure impact speeds during collisions. Studies depend on posted speed limits for a rough estimation because these are highly correlated with impact speeds and higher crash severity ( 10 , 18 , 23 ). Vehicle type and size correlate with the vehicle mass, which is the other element of the kinetic energy. Unsurprisingly, the literature almost unanimously agrees that larger vehicles, ranging from heavy trucks to minivans, sport utility vehicles (SUVs), and pickups, are more responsible for severe outcomes than sedans and coupes ( 10 , 18 , 24 ). Innovative technology, such as automatic pedestrian detection, automatic braking, and evasive steering in vehicles, has developed gradually and helped to reduce the severity of the outcomes ( 25 , 26 ), although it still requires substantial improvement to be effective in nighttime driving and high-speed conditions ( 27 ).

Incompatible infrastructure and road designs pose a great safety risk to pedestrians. Roads without proper and adequate pedestrian infrastructure increase pedestrians’ exposure to the traffic environment. Wide road crossings and random traffic flow patterns contribute to severe outcomes in crashes involving pedestrians ( 28 ). Straight and wide urban arterials in suburban areas significantly contribute to the massive kinetic energy in the system, encouraging higher speeds compared with traffic-calmed roads in a compact downtown area ( 28 – 30 ). Other factors that affect injury outcomes are adverse weather conditions and traveling in the dark. Slippery surfaces amplify driver errors, and loss of visibility increases reaction times, both of which affect the injury outcomes ( 23 , 31 ). Land use patterns, business activities, urban and rural settings, and so forth, also play some role in determining pedestrian injury outcomes ( 15 ).

Pedestrian Fatality Trend Studies in the U.S.A

Many pedestrian safety studies have examined the determinants of severe pedestrian crashes and the means of predicting them. Some studies have incorporated the temporal and spatial heteroskedasticity associated with pedestrian crashes ( 18 , 32 ), and although these have succeeded in understanding pedestrian crashes and their significant aspects, they have only focused on the cross-sections of time. They have not elucidated how pedestrian crash severity has varied over time. The media, and safety advocates tend to link the most critical findings of these severity studies with the ongoing U.S. pedestrian safety crisis, which might not necessarily be correct. Proponents of transportation safety often relate the suburbanization of poverty, growth in vehicle size, speed, lack of pedestrian facilities, intoxication, distraction in drivers, an aging population, and other factors to the increase in pedestrian fatalities. Likewise, others believe pedestrians’ negligence, distraction, intoxication, and failure to pay attention to the traffic control devices can all play a part in increasing pedestrian fatalities ( 33 ).

Only a handful of studies have explored the longitudinal effects of the factors involved, as mentioned earlier. Schneider (2020) performed trend analyses on fatality data from 1977 to 2016, and found that pedestrian fatalities associated with larger vehicles, such as minivans, SUVs, and pickup trucks, increased by 100%. The study also reported a growth of 26% in pedestrian fatalities on roads with a 35 mph or higher speed limit and a growth of 41% in the same on roads with more than three lanes ( 34 ). A couple of studies reported an 80% increase in pedestrian fatalities associated with SUVs from 2009 to 2016 ( 35 ) and 2009 to 2018 ( 36 ). Another recent study has found that replacing SUVs, minivans, and pickup trucks with smaller cars (sedans and coupes) would have avoided the deaths of 8,000 pedestrians between 2000 and 2019 ( 37 ). Tefft et al. ( 36 ) tackled the pedestrian safety problem by cross-tabulating various variables every year from 2009 to 2018. The study findings corroborated previous findings with regard to increased fatalities on urban nonfreeway arterials, crashes during the night, and crashes involving cars and SUVs ( 36 ). Ferenchak and Abadi ( 38 ) explored crashes during the night only, because these accounted for more than 85% of total pedestrian fatalities, and explored the relationship between nighttime crash trends and variables such as the pedestrian, the driver, the built environment, the vehicle, and situational factors. However, this study was unable to determine whether pedestrian crashes were becoming more severe ( 38 ). Pedestrian safety studies in Tennessee have implemented methods such as data mining ( 39 ) and spatial analysis and detection of hotspots ( 40 ) to predict pedestrian crashes using pedestrian crash data from the Tennessee Integrated Traffic Analysis Network (TITAN) database. However, these studies focus on cross-sectional data and do not break down the crash types to identify the increase in severity over time.

Research Gaps and Objectives

A critical limitation of these trend studies is that they only use FARS as the base data set for their analyses and only visualize the trends associated with fatal crashes. Hu and Cicchino ( 35 ) and Ferenchak and Abadi (38) went on to utilize secondary data sources such as the National Automobile Sampling System and General Estimates System for the exposure data, which is still a significant limitation of their studies ( 35 , 38 ). Inspired by these two studies, this research extends the horizon of pedestrian safety trend analyses because it overcomes the extant limitation by using a single database for both pedestrian involvement and fatalities in Tennessee, along with a novel variable derived from the home location of the pedestrians. This study is also among the first to investigate pedestrian crash trends at the state level.

This study focuses on the seriousness of pedestrian safety issues in the U.S.A., especially in Tennessee, and aims to assess the factors that contribute to declining pedestrian safety in the state. There are two fundamental questions that need to be asked. First, what has happened in Tennessee to cause fatal crashes to more than double in a decade, with no evidence for the trend? Second, what can the state and city governments and other partners do to address the problem? With the help of comprehensive visualizations and trend analysis tools, this study attempts to answer these questions, and the outcomes will be validated by appropriate statistical assessments. The study is based on Safe System principles for identifying the problems and suggesting countermeasures.

Methods

We categorize the pedestrian crashes in Tennessee spanning 11 years from 2009 to 2019 into two groups: urban; and rural. This division is relevant because of the nature of the crashes and countermeasures required. We do not include pedestrian crashes on interstates because they are mostly linked with disabled vehicles and are only a minuscule part of a big problem. We begin the trend analyses with descriptive statistics and visualizations, followed by an appropriate modeling tool to confirm the findings. These analyses are based on distinct crash variables, which in this study fall into four groups: pedestrian characteristics; driver characteristics; the built environment and situational factors; and vehicle characteristics. We employ three critical metrics for the descriptive statistics and modeling approach: pedestrian fatality count (PFC); pedestrian crash count (PCC); and pedestrian fatality rate (PFR). PFC is the annual number of pedestrian deaths, and PCC is the annual number of pedestrians involved in crashes, whatever the injury outcome. PFR is the ratio of PFC to PCC.

Data

This study uses TITAN, a database of all traffic crashes reported to police in Tennessee. For uniformity, crash reports in the TITAN data set are based on the MMUCC Guideline: Model Minimum Uniform Crash Criteria, produced by the National Highway Traffic Safety Administration. ( 41 ). TITAN uses the KABCO scale for injury outcomes in which K stands for a fatal crash, A for an incapacitating (serious) injury, B for a nonincapacitating apparent injury, C for possible injury, and O for no injury ( 42 ). TITAN contains three data sets: person; vehicle; and crash. We obtain complete information about the crash by connecting these data sets with a unique master record number associated with each crash. We present all variables as factors, so each variable has at least two values. From this point, we will use the term “variable categories” to refer to the values of variables. TITAN also records personal information, for example, the home addresses of drivers and pedestrians. We geocode the home addresses of pedestrians using the Google application program interface to convert them into coordinate values. Then, using the geodesic distance formula, we calculate the distance between the pedestrian’s home coordinates and the crash coordinates, which are also recorded in TITAN. Using this information, we explore the time trends for the distance between the pedestrian’s home and the crash location after determining the home coordinates by geocoding the pedestrian’s home addresses. Then, we calculate the distance between the pedestrian’s home and the crash location.

The 11-year study period incorporates the time during which electronic records became mainstream. Thus, there is a significant disparity between the information recorded during the initial phase of the study and that recorded during its final phase, with a considerable number of missing/unknown entries. The use of PFR overcomes this limitation to some extent. If missing/unknown values in TITAN are documented as random events uniformly affecting the police reporting in all types of outcomes, dividing PFR by PCC will normalize the underreporting created by the missing/unknown values. Pedestrians were also involved in multivehicle traffic crashes, although this occurred rarely. We simplify the case of multivehicle crashes by retaining the largest vehicle unless the police officer recorded the smaller vehicle as being at fault.

Descriptive Statistics and Trend Visualizations

First, we explain why we decided to look only at urban pedestrian crashes, in particular those that yielded fatal outcomes. Then, we prepare a frequency table for the crash variables and total PFC/PCC. This enables us to identify which attributes are causing fatal outcomes to an excessive degree. Although this analysis might help distinguish between the critical variables and variable categories, and those that are not important, it does not provide any insights into how they affect the pedestrian crash severity over time. To do so, we visualize the PFC, PCC, and PFR (expressed per 100 pedestrians) trends for each variable category consolidated in a single graph for a variable. Past studies only visualized PFC and failed to incorporate the effect of pedestrian exposure and, thus, were unable to identify the severity trends. Finally, we present a summary table for the trend analyses. We also perform one-way analysis of variance (ANOVA) tests to determine whether the variation within each variable (among the variable categories) is statistically significant for PFC and PFR. Ferenchak and Abadi ( 38 ) performed ANOVA tests for variables with three or more variable categories and t-tests for variables with only two variable categories ( 38 ). We relied on one-way ANOVA tests for all variable categories.

Modeling Approach

Hu and Cicchino ( 35 ) utilized Poisson regression models to determine the average annual increase in pedestrian fatalities. The current study fitted separate Poisson models for each variable category with the PFC associated with the variable category as the dependent variable and years as the independent variable ( 35 ). Similar to Hu and Cicchino ( 35 ), this study fits individual Poisson models for the PFC of each variable category. However, the current study extends their work by fitting another set of Poisson models to integrate the exposure variables and estimate the average annual increase in PFR. The functional form for estimating PFR using Poisson regression for a variable category “i” is given by the equation $\log (PF R_{i}) ~ a_{r} + b_{ri} \cdot year$ , where PFR_i = PFC_i/PCC_i. The equation can also be expressed as $\log (PF C_{i}) ~ a_{r} + b_{ri} \cdot year + \log (PE C_{i})$ .

We fit the above equation in Stata; PCC_i is the exposure variable.

For both PFC and PFR models, we determine the average annual change (AAC), expressed as a percentage, associated with the variable category “i” using the expression ${(AAC)}_{i} = (e^{b_{i}} ‒ 1) \times 100 %$ ( 35 ). Finally, we also test for the statistical significance of these estimates.

Results and Discussion

There were 20,445 pedestrians involved in traffic crashes reported between 2009 and 2019. This study excludes the 504 pedestrians who were struck by a vehicle on controlled access roads such as expressways and interstate highways. Of the remaining 19,941 pedestrians involved in crashes, 1,030 died. Figure 2 illustrates the profile for the injury outcome of these crashes from 2009 to 2020. We see a large growth in the proportion of fatal injuries, but other injury outcomes do not show a sizable increase.

Figure 2.

Pedestrian crash outcome profile.

Urban and Rural Pedestrian Crashes

Our study uses the Tennessee Department of Transportation definition for urban areas, which includes cities, suburbanized areas, and suburban fringe areas. Pedestrians involved in traffic crashes in rural areas account for only 8% of the total crashes in Tennessee, but still contribute to about 20% of all pedestrian fatalities. We did not include rural crashes in our study for two reasons. First, these crashes are scattered over a large area, unlike the urban ones, which took place on only a small part of Tennessee’s land but accounted for 80% of all fatalities. Second, although the overall severity of crash outcomes is relatively high in rural crashes (Figure 3), PFR has consistently increased and has in fact doubled over time for urban crashes; in rural areas it has increased more modestly. Therefore, we will only focus on urban crashes for the remaining part of this paper.

Figure 3.

Pedestrian trends in rural and urban areas.

Pedestrian Characteristics

This group of variables includes pedestrian attributes such as age, sex, race, walking under the influence of alcohol or drugs, the distance of the crash location from home, and a pedestrian’s location during the crash. Male pedestrians, pedestrians older than 36, intoxicated pedestrians, and pedestrians not using a crosswalk are more likely to be involved in fatal crashes than the other groups (see Table 1). The ANOVA test reveals that all variables within this group are significantly varied within themselves except the PFR associated with the distance from home to the crash location (Table 2). Fatality counts for pedestrian location and pedestrian distance from home were missing some data and, therefore, ANOVA tests could not be performed.

Table 1.

Frequency Table (Crash Attributes versus Fatal Outcome/Crash Involvement)

Variables	Fatal	(%)	Crash	(%)	Variables	Fatal	(%)	Crash	(%)
Pedestrian characteristics					Built environment and situational factors contd.
Age					Intersection
15 years and under	38	4.7	2,849	15.2	No	650	80.1	13,807	73.6
16–35	180	22.2	6,582	35.1	Yes	161	19.9	4,959	26.4
36–50	197	24.3	3,891	20.7	Number of lanes
51–65	278	34.3	3,749	20.0	Three or less	449	55.4	11,259	60
66 years and over	118	14.5	1,695	9.0	Four or more	325	40.1	3,919	20.9
Sex					Unknown	37	4.6	3,588	19.1
Female	228	28.1	7,822	41.7	Weekdays/weekend
Male	583	71.9	10,944	58.3	Weekdays	574	70.8	14,236	75.9
Race					Weekend	237	29.2	4,530	24.1
Black	244	30.1	5,691	30.3	Time of day
White	449	55.4	8,646	46.1	Midnight–6:00 a.m.	133	16.4	1,825	9.7
Other	26	3.2	377	2.0	6:00 a.m.–noon	100	12.3	3,710	19.8
Unknown	92	11.3	4,052	21.6	Noon–6:00 p.m.	133	16.4	7,143	38.1
Alcohol/drugs					6:00 p.m.–midnight	445	54.9	6,088	32.4
Not present	632	77.9	17,670	94.2	Lighting
Present	179	22.1	1,096	5.8	Daylight	173	21.3	10,753	57.3
Distance from home					Dark—lighted	460	56.7	5,490	29.3
Less than 2 mi	352	45.2	7,957	43.6	Dark—unlighted	144	17.8	1,576	8.4
2 mi or more	332	42.6	7,836	43.0	Others	34	4.2	947	5
Unknown	95	12.2	2,443	13.4	Parking lot or private property
Location					Neither	764	94.2	13,805	73.6
Road—not in crosswalk	324	41.6	3,929	21.5	Parking lot	30	3.7	3,615	19.3
Road—crosswalk not available	162	20.8	1,850	10.1	Private property	17	2.1	1,346	7.2
In the crosswalk	29	3.7	1,472	8.1	Land use
Not in roadway	86	11	4,631	25.4	Residential	190	23.4	5,190	27.7
Unknown	178	22.8	6,354	34.8	Nonresidential	621	76.6	13,576	72.3
Driver characteristics					Vehicle characteristics
Age					Maneuver
14–25	145	17.9	3,081	16.4	Straight (intersection)	123	15.2	1,966	10.5
26–55	399	49.2	7,351	39.2	Straight (midblock)	512	63.1	6,962	37.1
56 years and over	157	19.4	3,748	20	Backing and parking	30	3.7	2,848	15.2
Others/unknown	110	13.6	4,586	24.4	Turning	35	4.3	2,996	16
Sex					Other/unknown	111	13.7	3,994	21.3
Female	201	24.8	6,511	34.7	Vehicle type
Male	505	62.3	8581	45.7	Cars (sedans and coupes)	336	41.4	9,202	49
Unknown	105	12.9	3,674	19.6	SUVs	156	19.2	3,356	17.9
Race					Pickups and minivans	189	23.3	3,580	19.1
Black	260	32.1	5,063	27	Heavy vehicle	66	8.1	792	4.2
White	432	53.3	9,205	49.1	Others/unknown	64	7.9	1,836	9.8
Other	13	1.6	481	2.6	Model year
Unknown	106	13.1	4,017	21.4	1999 and older	157	19.4	3,199	17
Alcohol/drugs					2000–2009	389	48	7,976	42.5
Not present/unknown	726	89.5	18,266	97.3	2010–2019	164	20.2	3,714	19.8
Present	85	10.5	500	2.7	Unknown	101	12.5	3,877	20.7
Driving license					Hit and run
Valid	515	63.5	10,032	53.5	Yes	187	23.1	4,405	23.5
Invalid/not licensed	162	20	3,324	17.7	No	624	76.9	14,361	76.5
Unknown	134	16.5	5,410	28.8	First impact
					Front end	513	63.3	7,732	41.2
Built environment and situational factors					Rear end	23	2.8	2,048	10.9
Posted speed limit					Right side	93	11.5	3,400	18.1
15 mph or less	56	6.9	5,772	30.8	Left side	73	9	2,350	12.5
16–34 mph	128	15.8	5,262	28.0	Other	109	13.4	3,236	17.24
35 mph and over	627	77.3	7,732	10.5

Note: SUV = sport utility vehicle.

Table 2.

Summary of Crash Trends with Estimated Average Annual Change Using Poisson Regressions

Variables	Pedestrian fatality count							Pedestrian fatality rate
	PFC			% Growth	ANOVA		AAC (%)	PFR (per 100 exposed)			% Growth	ANOVA		AAC (%)
	2009	2019	Mean	% Growth	F	p	AAC (%)	2009	2019	Mean	% Growth	F	p	AAC (%)
Pedestrian characteristics
Age
15 and under	4	2	3.5	-50.0	23.62	<0.001	-6.93	1.29	0.95	1.30	-26.8	46.02	<0.001	-3.80
16–35	10	27	16.4	170.0			8.44**	1.99	4.10	2.70	105.7			5.61*
36–50	18	33	17.9	83.3			7.11**	5.49	7.97	4.99	45.2			4.77*
51–65	13	36	25.3	176.9			9.45***	5.53	8.80	7.29	59.1			4.57*
66 and over	10	13	10.7	30.0			6.16*	8.26	6.16	6.99	-25.5			-0.14
Sex
Female	14	32	20.7	128.6	46.87	<0.001	6.53**	2.17	4.10	2.89	88.7	38.23	<0.001	4.46*
Male	41	79	53.0	92.7	46.87	<0.001	7.65***	4.82	7.03	5.28	45.9	38.23	<0.001	4.85***
Race
Black	2	42	22.2	2000	26.1	<0.001	19.74***	7.14	5.71	4.53	-20.1	10.95	<0.001	6.47*
White	11	68	40.8	518.2			13.34***	4.87	6.76	5.13	38.9			3.12
Other	1	1	2.4	0.0			-1.91	14.29	2.63	7.88	-81.6			-11.35
Alcohol/drugs
Not present	50	76	57.5	52.0	81.15	<0.001	4.67***	3.39	4.32	3.57	27.4	57.42	<0.001	2.76*
Present	5	35	16.3	600.0	81.15	<0.001	18.00***	26.32	24.48	16.78	-7.0	57.42	<0.001	6.37*
Distance from home^a
Less than 2 mi	22	29	32.0	-	-	-	2.65	4.23	4.97	4.44	17.6	0.62	0.440	0.17
2 mi or more	11	41	30.2	-	-	-	13.68***	2.11	5.91	4.09	180.4	0.62	0.440	9.35***
Location^a
Road—not in crosswalk	9	35	29.5	-	-	-	16.35*	14.06	9.64	8.75	-31.4	54.41	<0.001	2.67
Road—crosswalk not available	4	23	14.7	-			12.91***	8.51	14.29	8.81	67.9			7.03*
In the crosswalk	0	4	2.6	-			11.48***	0.00	2.63	1.80	na			2.41
Not in roadway	1	7	7.8	-			5.15	0.77	1.77	1.82	129.8			-3.49
Driver characteristics
Age
14–25	5	22	13.2	340.0	39.95	<0.001	6.01*	1.84	7.51	4.70	308.5	2.14	0.136	6.07*
26–55	32	55	36.3	71.9			6.03***	5.25	7.77	5.42	48.1			4.99**
56 and over	11	21	14.3	90.9			7.97**	4.40	5.26	4.16	19.6			3.05
Sex
Female	11	38	18.3	245.5	50.82	<0.001	10.81***	2.00	6.31	3.09	215.0	29.78	<0.001	9.80***
Male	37	61	45.9	64.9	50.82	<0.001	5.10***	5.69	7.00	5.87	22.9	29.78	<0.001	2.97*
Race
Black	15	41	23.6	173.3	74.57	<0.001	9.85***	3.96	7.96	5.10	101.2	10.56	<0.001	7.99***
White	31	57	39.3	83.9			5.50***	4.04	6.90	4.70	71.0			4.63**
Other	2	0	1.2	-100			-19.64*	3.13	0.00	2.40	-100.0			-11.86
Alcohol/drugs
Not present	50	103	66	106.0	113.3	<0.001	7.66***	3.43	5.46	3.93	59.1	20.68	<0.001	4.99***
Present	5	8	7.7	60.0	113.3	<0.001	4.59	13.16	47.06	19.04	257.6	20.68	<0.001	9.19*
Driving license
Valid	12	77	46.8	541.7	23.78	<0.001	12.76***	6.15	6.38	5.18	3.8	1.26	0.274	1.83
Invalid/not licensed	0	30	14.7	na	23.78	<0.001	23.36***	0.00	6.30	4.42	na	1.26	0.274	8.23*
Built environment and situational factors
Intersection
No	46	89	59.1	93.5	77.78	<0.001	7.00***	4.05	6.44	4.68	59.0	9.81	<0.003	4.89***
Yes	9	22	14.6	144.4	77.78	<0.001	8.67**	2.51	4.21	3.19	67.7	9.81	<0.003	5.09
Number of lanes
Three or less	33	56	40.8	69.7	5.10	<0.001	4.01**	3.66	5.25	3.99	43.4	74.59	<0.001	2.52
Four or more	18	53	29.5	194.4	5.10	<0.001	14.56***	7.29	10.39	7.94	42.6	74.59	<0.001	6.69***
Posted speed limit
15 mph or less	5	2	5.1	-60.0	65.71	<0.001	-6.79	1.01	0.37	0.98	-63.6	131.6	<0.001	-7.50
16 mph–34 mph	12	12	11.6	0.0			2.29	2.78	2.46	2.44	-11.5			0.91
35 mph and over	38	97	57.0	155.3			9.84***	6.70	11.13	7.96	66.1			5.14***
Weekdays/weekend
Weekdays	39	74	52.2	89.7	43.48	<0.001	5.55***	3.36	5.24	4.01	55.9	3.65	<0.001	3.33*
Weekend	16	37	21.5	131.3	43.48	<0.001	11.91***	4.78	7.52	5.13	57.5	3.65	<0.001	8.45***
Time of day
Midnight–6:00 a.m.	7	22	12.1	214.3	50.29	<0.001	12.37***	6.31	10.28	7.25	63.0	29.88	<0.001	5.50
6:00 a.m.–noon	7	16	9.1	128.6			6.78*	2.35	4.27	2.66	81.6			4.10
Noon–6:00 p.m.	8	11	12.1	37.5			4.63	1.29	1.60	1.88	24.5			3.92
6.00 p.m.–midnight	33	62	40.5	87.9			6.84***	7.11	9.87	7.25	38.8			3.66*
Lighting
Daylight	10	21	15.7	110.0	52.09	<0.001	3.84	1.09	2.02	1.61	85.9	34.69	<0.001	2.76
Dark—lighted	33	65	41.8	97.0			8.89***	8.13	11.30	8.26	39.1			5.09**
Dark—unlighted	11	24	13.1	118.2			7.96**	9.48	11.76	9.04	24.1			1.76
Others	1	1	3.1	0.0			2.38	1.89	1.16	3.66	-38.4			-2.44
Parking lot or private property
Neither	51	109	69.5	113.7	117.0	<0.001	8.25***	4.73	7.52	5.47	58.9	95.01	<0.001	4.83***
Parking lot	4	2	2.7	-50.0			-8.11	1.35	0.57	0.84	-57.6			-9.30
Private property	0	0	1.5	na			-2.90	0.00	0.00	1.21	na			-0.15
Land use
Residential	17	22	17.2	29.4	50.19	<0.001	3.00	3.84	4.95	3.67	28.9	5.37	<0.031	2.11
Nonresidential	38	89	56.5	134.2	50.19	<0.001	8.72***	3.61	6.10	4.51	69.0	5.37	<0.031	5.55***
Vehicle characteristics
Maneuver
Straight (intersection)	8	18	11.1	125.0	84.22	<0.001	10.56**	5.67	9.05	6.14	59.6	42.71	<0.001	7.59*
Straight (midblock)	38	72	46.5	89.5			7.73***	6.83	10.36	7.30	51.6			4.94**
Backing and parking	3	1	2.7	-66.7			-11.63*	1.23	0.40	1.05	-67.1			-12.32*
Turning	0	4	3.2	na			6.23	0.00	1.20	1.17	na			2.12
Vehicle type
Cars (sedans and coupes)	22	47	30.5	113.6	31.52	<0.001	7.78***	3.31	4.96	3.64	49.8	9.6	<0.001	5.06**
SUVs	12	25	14.2	108.3			8.10**	5.06	7.23	4.60	42.7			3.37
Pickups/minivans	12	21	17.2	75.0			4.12	4.08	6.31	5.30	54.5			3.61
Heavy vehicle	3	7	6.0	133.3			6.96	6.52	8.97	8.36	37.6			3.52
Others/unknown	6	11	5.8	83.3			13.59**	2.36	5.5	3.47	132.8			11.51**
Model year
1999 and older	19	8	14.3	-57.9	15.01	<0.001	-12.41***	3.87	8.51	5.08	119.9	2.48	0.101	1.38
2000–2009	29	42	35.4	44.8			4.65**	4.13	7.08	5.00	71.4			7.36***
2010–2019^b	2	47	14.9	na			39.60***	4.17	6.03	3.68	na			8.15*
Hit and run
Yes	3	28	17.0	833.3	69.49	<0.001	18.71***	7.50	4.68	4.41	-37.6	0.01	0.928	4.09
No	52	83	56.7	59.6	69.49	<0.001	4.35**	3.57	6.36	4.37	77.8	0.01	0.928	5.26***
First impact
Front end	43	64	46.6	48.8	116.3	<0.001	4.65**	6.97	8.24	6.63	18.2	33.67	<0.001	2.64
Rear end	2	0	2.1	-100.0			-10.12	1.08	0.00	1.10	-100.0			-8.99
Right side	5	18	8.5	260.0			14.03***	2.1	5.44	2.68	158.9			11.33**
Left side	0	14	6.6	na			18.08***	0.0	5.30	2.95	na			12.64**
Other/unknown	5	15	9.9	200.0			12.74***	1.63	4.25	3.28	160.9			8.62**

Note: PFC = pedestrian fatality counts; PFR = pedestrian fatality rate; ANOVA = analysis of variance; AAC = average annual change; SUV = sport utility vehicle.

Calculations skipped because of missing values.

Pedestrian data of 2010 instead of 2009 is used for calculation for vehicles with model year 2010-2019.

p-value < 0.05. **p-value < 0.01. ***p-value < 0.001.

We saw a substantial increase in PFC and PFR from 2009 to 2019 for age groups 16 to 35 and 51 to 65, corroborated by an approximately 5% estimated annual average change using Poisson regression with a 0.05 significance level. PFR associated with children and elderly pedestrians over 65 declined over the study period, although the trends were not statistically significant (Figure 4 and Table 2). Similarly, a higher increase in PFC was observed for males, with the average PFR and PFC double that of females. Nevertheless, both males and females exhibit an annual growth rate in PFR of more than 4% and in PFC by more than 6%, both significant with 95% confidence (see Figure 4 and Table 2).

Figure 4.

Fatality and severity trends: (a) age, and (b) sex of the pedestrians.

PFC trends are usually steeper than PFR trends because PFC does not account for pedestrian involvement; however, unknown values distort PFC trends even more, as seen in Figure 5a. PFR can manage a plausible trend even with a disproportionate number of missing/unknown values (refer to Figure 5a). PFR trends for pedestrian race suggest a significant (p < 0.05) average annual increase of 6.5 for black pedestrians (refer to Table 2, Figure 5a). Likewise, although the average annual increase in severity is not statistically significant for other variable categories, it has increased by 7% with a 0.05% increase every year for pedestrians who are struck by a vehicle where there is no crosswalk available (Table 2, Figure 5b). The fatality count in the case of walking under the influence increased by 600% from 2009 to 2019, suggesting a possible case of missing/unknown data (Table 2, Figure 5c). However, we can see a statistically significant (p < 0.05) rise in average annual PFR with a value of 6.4% for intoxicated individuals and 2.8% for others. Although the distance from home variable failed the ANOVA test for variability, the Poisson regression model estimated an annual increase of 9.4% with a 0.001 significance level for pedestrians living at least 2 mi away from the crash location. In contrast, the estimate was neither large nor significant for pedestrians who lived within 2 mi of the crash location (Table 2, Figure 5d).

Figure 5.

Fatality and severity trends: (a) race, (b) location during the crash, (c) alcohol or drug presence, and (d) distance between a pedestrian’s home and the crash location.

Driver Characteristics

TITAN also records drivers’ demographic information, including age, sex, race, and presence of alcohol or drugs. Table 1 suggests that female drivers are less likely to be involved in a fatal pedestrian crash than male drivers. It also illustrates that drivers under the influence of alcohol or drugs are five times more likely to strike a pedestrian with a fatal outcome than sober drivers. Although the fatality counts associated with driver characteristics are distinct, ANOVA, in Table 2, found similarities within the PFR in relation to the age of the driver and license status variables.

Pedestrian deaths and severity of injury in relation to young drivers aged 14 to 25 increased by a considerable margin, over 300%. The Poisson regression coefficients are positive and significant (p < 0.05) for both metrics, PFC, and PFR. The PFR trend for older drivers showed a weaker increase in severity with no statistical significance (Table 2, Figure 6a). It is surprising that although female drivers are less likely to cause a fatal pedestrian crash, the associated severity and fatality are highly significant (p < 0.001), with annual increases of 9.8% and 10.8%, respectively. For reference, the rate of increase in PFC and PFR for male drivers is less than for female drivers, although both metrics are statistically significant (Table 2, Figure 6b). Black drivers also contribute more to the increase in severity of pedestrian crashes than white drivers, because the average annual increase was 8% and 4.6%, respectively; both figures are statistically significant (Table 2, Figure 6c). We do not see a significant increase in PFC for drivers under the influence. However, the AAC associated with PFR was positive and significant (Table 2, Figure 6d). Nonlicensed drivers and drivers with an invalid license contribute to 20% of the total fatalities, and they are also becoming involved in more severe crashes, with a significant annual growth of 8.2%.

Figure 6.

Fatality and severity trends with regard to drivers: (a) age, (b) alcohol or drug presence, (c) sex, and (d) race.

Built Environment and Situational Factors

This group includes variables that are mostly concerned with land use, road design, the environment, and temporal factors. High-speed multilane roadways and nighttime crashes produce disproportionately high fatalities (Table 1). All of the values of the variables in this group are sufficiently varied within themselves, as shown by the ANOVA results from Table 2 in the built environment and situational factors section.

Further exploring Table 2, we can see where the increase in fatalities and PFR is taking place in relation to the road design attributes. Multilane roads with more than three lanes exhibit an enormous growth in their contribution to fatalities and severity of injury, because the Poisson regression average values for fatalities and the fatality rate are +14.6% per year and +6.7% per year, respectively, estimated with 99.9% confidence. High-speed roads, that is, those with speeds greater than 35 mph, also show a strong increase in their contribution to the severity of pedestrian crashes, with PFR increasing by 5.1% each year (p < 0.001). A nonintersection location is linked with an increase in fatal pedestrian crashes (AAC = +7 % per year, p < 0.001) and PFR (AAC = +4.9% per year, p < 0.001). Intersection locations, narrow roads, and roads with a posted speed limit of less than 35 mph are not significantly correlated with an increase in the severity of crashes. Thus, we maintain that wide, high-speed multilane roads with few intersections play a huge role in the increase in severity of crashes in urban Tennessee. Figure 7 illustrates the PCC, PFC, and PFR trends for the posted speed limit, number of lanes, and intersection/nonintersection locations, respectively.

Figure 7.

Fatality and severity trends: (a) posted speed limit, (b) number of lanes, (c) intersection and nonintersection locations, and (d) residential and nonresidential land uses.

Neither parking lots nor roads on private property are associated with an increase in fatalities. Although statistically insignificant, those variables have a negative slope contributing to a slight decrease in PFR. However, PFR in locations that are not a parking lot or private property increases by almost 60% (AAC estimate = 4.8% per year, p-value < 0.001) (see Table 2). The reason for this may be that underreporting rates change over time. Table 2 shows that fatal crashes occurring in residential areas (as defined by TITAN), have barely increased from 17 in 2009 to 22 in 2019. Although these accounted for 23% of total fatal crashes, as shown in Table 1, the proportion decreased from 31% to 20% in 11 years. Fatal crashes occurring in nonresidential areas saw an estimated 134% increase with an 8.7% increase per year (p < 0.001). The PFR trend for the latter is also statistically significant, with a 69% increase from 3.61 to 6.1 during the study period (AAC = 5.55% per year, p-value < 0.001). Figure 7 shows the respective PFR trends for pedestrian crashes occurring in residential and nonresidential areas.

Another set of variables under this category includes time of day, lighting, and weekdays or weekends. The 6:00 p.m. to midnight attribute of time of day variable and dark attribute of lighting variable are similar. Crashes from 6:00 p.m. to midnight have almost doubled from 33 to 62 over the 11 years of the study period, with an annual average of 40.5 deaths and an estimated 6.84% increase per year (p < 0.001). Only crashes occurring from 6:00 p.m. to midnight and dark—lighted conditions are significant for PFR (p-value < 0.05 and 0.01, respectively), and both are associated with an overall increase of 39%. Figure 8a shows the increasing trends for dark—lighted and unlighted conditions. Finally, weekend crashes are gradually becoming associated with more severe injury outcomes for pedestrians. Estimated PFR growth is statistically significant and has a value of 8.4% per year with 99.9% confidence. Weekday crashes are relatively less severe than weekend crashes but do have a significant upward trend (AAC = +3.3% per year, p < 0.05) as per the regression results (see Table 2 and Figure 8b).

Figure 8.

Fatality and severity trends: (a) lighting, and (b) weekends/weekdays.

Vehicle Characteristics

Variables such as vehicle type, hit and run, backing maneuver, and so on, are often the most speculative variables in the media and literature. According to Table 1, a front-end collision, straight midblock and straight intersection maneuvers, and heavy vehicles such as trucks are more represented in fatal pedestrian crashes. Although all variables pass the ANOVA test for detecting variations in relation to PFC, the ANOVA test for variation within the data finds that vehicle age PFR values have no significant variation among themselves. Additionally, PFR values of hit-and-run crashes cannot be distinguished from PFR values of non-hit-and-run crashes (see Table 2).

Surprisingly, no vehicle type has shown a significant increase in PFR except cars (sedan and coupes). We observe there is a higher chance of being injured if struck by a larger vehicle. However, with regard to an increase in the severity of crashes over time, estimates for SUVs, pickups and minivans, and heavy vehicles are not within the statistical significance threshold of 0.05. Counterintuitively, the Poisson regression estimate of the average change in PFR per year is the highest for cars, with a 5.1% increase every year on average, indicating that cars are still significant contributors to the increase in severity of crashes in Tennessee (Figure 9b). Vehicle age (model year minus crash year) in Table 2 and model year in Figure 9a do not have extreme differences over time. Neither older vehicles nor the advanced safety features in newer vehicles have particularly improved pedestrian safety. Such safety features, meant primarily to protect car occupants, although they can have an effect on certain external conditions, have not had the desired effect of improving pedestrian safety because they need substantial improvements before they can be effective at nighttime and in high-speed conditions. This is an area for future research, because the rollout of new vehicles correlates perfectly with exogenous growth in the severity of pedestrian crashes. The straight midblock maneuver is the one that most often causes fatal crashes, and it conforms with the overall increase in fatal pedestrian crashes; the PFR caused by straight midblock maneuvers was 6.83 in 2009, but rose to 10.36 in 2019, almost a 51.6% increase with significance (p < 0.01) (refer to Table 2 and Figure 9c, maneuver type). Table 2 and Figure 10a show that non-hit-and-run crashes have contributed significantly to the overall pedestrian fatality trend with significance. Finally, PFR is associated significantly with left- and right-side collisions instead of front-end collisions. Despite the significant increases in fatalities associated with right- and left-side collisions, a front-end collision remains more than 100% more lethal (see Table 2, Figure 10b).

Figure 9.

Fatality and severity trends: (a) vehicle model year, (b) vehicle type/size, and (c) type of maneuver.

Figure 10.

Fatality and severity trends: (a) hit and runs, and (b) place of first impact.

Conclusions and Recommendations

We produced descriptive statistics, performed various analyses on the chosen urban pedestrian crash data with respect to various crash attributes, and meticulously explored each of the latter with regard to PFC and PFR metrics. We identified the main causes of the current situation in relation to pedestrian safety in Tennessee and provided recommendations for mitigating and reducing pedestrian fatalities.

We found there has been an increase in severity of crashes leading to an increase in pedestrian fatalities in the urban areas of Tennessee, and that roadway design is a major contributory factor. We also discovered that the severity of crashes had increased more on straight roads, on roads with a speed of more than 35 mph, and on multilane roads. These characteristics are typical of urban arterials in Tennessee. Nighttime crashes accounted for 75% of total fatalities and significantly contributed to the increase in severity. In addition, severe crashes have occurred further from pedestrians’ homes. Because pedestrians get struck by a vehicle further from their homes on large urban arterials, we could speculate that a sprawling suburban scenario ( 43 ) has a role to play in this outcome. An analysis of National Household Travel Survey data from 2001, 2009, and 2017 revealed a disproportionate reliance on walking as a mode of transportation among low-income households in U.S. suburbs, with an unexpected upward trend for low-income car-owning households ( 44 ). This suggests the need for further investigation to explore the impact of suburbanization of poverty on pedestrian crash trends in Tennessee.

Our findings conform with the most recent U.S. pedestrian safety research, which associates the urban pedestrian safety crisis in the U.S.A. with the functional classification of the roadways (34 –36, 38). Other variables with increasing representation in fatal pedestrian crashes were male pedestrians, middle-aged older adults (51–65), female drivers, driving under the influence, unlicensed drivers and drivers with an invalid license, walking under the influence, weekend crashes, and newer vehicle models. In contrast to the widespread speculation about growing SUV and pickup truck sizes causing a reduction in the probability of survival after a crash, our study does not see any significant PFR trends that support this claim. However, the average value of PFR is greater for larger vehicles. Cars are still largely responsible for the increasing severity of pedestrian crashes in Tennessee. Our study found that although statistically insignificant, the severity of crashes associated with elderly pedestrians is declining. This finding challenges another speculation that maintains the aging population is responsible for the increase in crash severity ( 33 ).

Vehicle technologies such as pedestrian detection and automatic emergency braking do not appear to be of much help in increasing pedestrian safety because they cannot detect pedestrians during the night and in high-speed conditions in midblocks ( 27 ). Moreover, the trend of increasing severity is pointing primarily toward road infrastructure and design. This study recommends that cities and urban stakeholders pursue the following. First, safe speeds on the roads, preferably below 35 mph, because roads with speeds of over 35 mph have seen a notable increase in severity of crashes during the last decade. Second, transportation agencies should provide adequate accommodation for pedestrians with appropriate pedestrian infrastructures such as isolated sidewalks and crossing opportunities. Third, cities should provide more pedestrian-scale lighting because of the increased crash severity associated with nighttime crashes. There are some limitations of this study. The first is that we rely on police data. Police observations can be prone to subjective bias, and officers may not follow standards every time they record a crash. In addition, not every crash is documented. For example, the police do not get to hear about unreported crashes and near-miss incidents. The second limitation is that this study only examines the overall crash circumstances and probably misses other essential variables. Future studies should use additional data sources such as sociodemographic census data, traffic volumes, vehicle crashes, extensive road design features, and spatial data and explore individual variables that contribute significantly to the increase in severity of crashes. Nonetheless, this study provides the most precise breakdown of traffic crashes in Tennessee to date and offers a model for investigating other factors in more states ( 44 ).

Footnotes

Author Contributions

The authors confirm their contribution to the paper as follows: study conception and design: S. Parajuli, C. Cherry; data collection: C. Cherry; analysis and interpretation of results: S. Parajuli, C. Cherry; draft manuscript preparation: S. Parajuli, C. Cherry, E. Zavisca, W. Rogers III. All authors reviewed the results and approved the final version of the manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The work in this paper was funded in part by the Tennessee Department of Transportation (Grant RES 2021-11).

ORCID iDs

Saurav Parajuli

Christopher R. Cherry

The opinions, findings, and conclusions expressed in this publication are those of the authors and not necessarily those of the Tennessee Department of Transportation or the U.S. Department of Transportation.

References

Centers for Disease Control and Prevention. WISQARS – Web-Based Injury Statistics Query and Reporting System. https://www.cdc.gov/injury/wisqars/index.html. Accessed August 23, 2023.

National Center for Statistics and Analysis. Traffic Safety Facts: Pedestrians. Report No. DOT HS 813 079. National Highway Traffic Safety Administration, Washington, DC, 2021.

European Transport Safety Council. Urgent Action Needed to Tackle Deaths of Pedestrians and Cyclists. https://etsc.eu/urgent-action-needed-to-tackle-deaths-of-pedestrians-and-cyclists/#:~:text=The%20report%20recommends%20applying%20a,levels%20of%20walking%20and%20cycling. Accessed August 23, 2023.

Department for Transport. Reported Road Casualties in Great Britain: 2019. Annual Report. Department for Transport, London, 2020.

Australian Government. Road Deaths Database. Department of Infrastructure, Transport, Regional Development, Communications and the Arts, Canberra, 2020. https://www.bitre.gov.au/statistics/safety/fatal_road_crash_database. Accessed August 23, 2023.

National Center for Statistics and Analysis. Traffic Safety Facts 2009 Data: Pedestrians. Report No. DOT HS 811 394. National Highway Traffic Safety Administration, Washington, DC, 2011.

Tennessee Highway Safety Office. Tennessee Integrated Traffic Analysis Network. Tennessee Highway Safety Office, Nashville, TN, 2021.

Zajac

S.S.

Ivan

J. N.

Factors Influencing Injury Severity of Motor Vehicle–Crossing Pedestrian Crashes in Rural Connecticut. Accident Analysis & Prevention, Vol. 35, No. 3, 2003, pp. 369–379.

Dultz

L. A.

Frangos

S. G.

The Impact of Alcohol in Pedestrian Trauma. Trauma, Vol. 15, No. 1, 2013, pp. 64–75.

10.

Kröyer

H. R. G.

Is 30km/h a “Safe” Speed? Injury Severity of Pedestrians Struck by a Vehicle and the Relation to Travel Speed and Age. IATSS Research, Vol. 39, No. 1, 2015, pp. 42–50.

11.

Oxley

Fildes

Ihsen

Charlton

Day

Differences in Traffic Judgements Between Young and Old Adult Pedestrians. Accident Analysis & Prevention, Vol. 29, No. 6, 1997, pp. 839–847.

12.

Schwebel

D. C.

Davis

A. L.

O’Neal

E. E.

Child Pedestrian Injury: A Review of Behavioral Risks and Preventive Strategies. American Journal of Lifestyle Medicine, Vol. 6, No. 4, 2012, pp. 292–302.

13.

Pitcairn

Edlmann

Individual Differences in Road Crossing Ability in Young Children and Adults. British Journal of Psychology, Vol. 91, No. 3, 2000, pp. 391–410.

14.

Oxley

Lenné

Corben

The Effect of Alcohol Impairment on Road-Crossing Behaviour. Transportation Research Part F: Traffic Psychology and Behaviour, Vol. 9, No. 4, 2006, pp. 258–268.

15.

Roll

McNeil

Race and Income Disparities in Pedestrian Injuries: Factors Influencing Pedestrian Safety Inequity. Transportation Research Part D: Transport and Environment, Vol. 107, 2022, Article 103294.

16.

Noland

R. B.

Klein

N. J.

Tulach

N. K.

Do Lower Income Areas Have More Pedestrian Casualties?

Accident Analysis & Prevention, Vol. 59, 2013, pp. 337–345.

17.

Hickox

K. L.

Williams

Beck

L. F.

Coleman

Fudenberg

Robinson

Middaugh

Pedestrian Traffic Deaths among Residents, Visitors, and Homeless Persons—Clark County, Nevada, 2008–2011. Morbidity and Mortality Weekly Report, Vol. 63, No. 28, 2014, p. 597–602.

18.

Liu

Hainen

Nie

Nambisan

Pedestrian Injury Severity in Motor Vehicle Crashes: An Integrated Spatio-Temporal Modeling Approach. Accident Analysis Prevention, Vol. 132, 2019, Article 105272.

19.

Hezaveh

A.M.

Cherry

C.R.

, Walking under the Influence of Alcohol: A Case Study of Pedestrian Crashes in Tennessee. Accident Analysis & Prevention, Vol. 121, 2018, pp. 64-70.

20.

Khan

N. A.

Habib

M. A.

Exploring the Impacts of Built Environment on Pedestrian Injury Severity Involving Distracted Driving. Journal of Safety Research, Vol. 80, 2022, pp. 97–108.

21.

Jurewicz

Sobhani

Woolley

Dutschke

Corben

Exploration of Vehicle Impact Speed – Injury Severity Relationships for Application in Safer Road Design. Transportation Research Procedia, Vol. 14, 2016, pp. 4247–4256.

22.

Kröyer

H. R. G.

Jonsson

Várhelyi

Relative Fatality Risk Curve to Describe the Effect of Change in the Impact Speed on Fatality Risk of Pedestrians Struck by a Motor Vehicle. Accident Analysis & Prevention, Vol. 62, 2014, pp. 143–152.

23.

Salon

McIntyre

Determinants of Pedestrian and Bicyclist Crash Severity by Party at Fault in San Francisco, CA. Accident Analysis & Prevention, Vol. 110, 2018, pp. 149–160.

24.

Davis

G. A.

Relating Severity of Pedestrian Injury to Impact Speed in Vehicle–Pedestrian Crashes: Simple Threshold Model. Transportation Research Record: Journal of the Transportation Research Board, 2001. 1773: 108–113.

25.

Keller

C. G.

Dang

Fritz

Joos

Rabe

Gavrila

D. M.

Active Pedestrian Safety by Automatic Braking and Evasive Steering. IEEE Transactions on Intelligent Transportation Systems, Vol. 12, No. 4, 2011, pp. 1292–1304.

26.

Strandroth

Sternlund

Lie

Tingvall

Rizzi

Kulgren

Ohlin

Fredriksson

Correlation between Euro NCAP Pedestrian Test Results and Injury Severity in Injury Crashes with Pedestrians and Bicyclists in Sweden. Stapp Car Crash Journal, Vol. 58, 2014, pp. 213–231.

27.

Cicchino

J. B.

Effects of Automatic Emergency Braking Systems on Pedestrian Crash Risk. Accident Analysis & Prevention, Vol. 172, 2022, Article 106686.

28.

Stoker

Garfinkel-Castro

Khayesi

Odero

Mwangi

M. N.

Peden

Ewing

Pedestrian Safety and the Built Environment: A Review of the Risk Factors. Journal of Planning Literature, Vol. 30, No. 4, 2015, pp. 377–392.

29.

Tay

Choi

Kattan

Khan

A Multinomial Logit Model of Pedestrian–Vehicle Crash Severity. International Journal of Sustainable Transportation, Vol. 5, No. 4, 2011, pp. 233–249.

30.

Damsere-Derry

Ebel

B. E.

Mock

C. N.

Afukaar

Donkor

Kalowole

T. O.

Evaluation of the Effectiveness of Traffic Calming Measures on Vehicle Speeds and Pedestrian Injury Severity in Ghana. Traffic Injury Prevention, Vol. 20, No. 3, 2019, pp. 336–342.

31.

Ranjitkar

Zhao

Rashidi

Analyzing Pedestrian Crash Injury Severity under Different Weather Conditions. Traffic Injury Prevention, Vol. 18, No. 4, 2017, pp. 427–430.

32.

Eluru

Bhat

C. R.

Hensher

D. A.

A Mixed Generalized Ordered Response Model for Examining Pedestrian and Bicyclist Injury Severity Level in Traffic Crashes. Accident Analysis & Prevention, Vol. 40, No. 3, 2008, pp. 1033–1054.

33.

Schmitt

Right of Way: Race, Class, and the Silent Epidemic of Pedestrian Deaths in America. Island Press, Washington, D.C., 2020.

34.

Schneider

R. J.

United States Pedestrian Fatality Trends, 1977 to 2016. Transportation Research Record: Journal of the Transportation Research Board, 2020. 2674: 1069–1083.

35.

Cicchino

J. B.

An Examination of the Increases in Pedestrian Motor-Vehicle Crash Fatalities during 2009–2016. Journal of Safety Research, Vol. 67, 2018, pp. 37–44.

36.

Tefft

B. C.

Arnold

L. S.

Horrey

W. J.

Examining the Increase in Pedestrian Fatalities in the United States, 2009–2018. Research Brief. AAA Foundation for Traffic Safety, Washington, D.C., 2021.

37.

Tyndall

Pedestrian Deaths and Large Vehicles. Economics of Transportation, Vol. 26–27, 2021, Article 100219.

38.

Ferenchak

N. N.

Abadi

M. G.

Nighttime Pedestrian Fatalities: A Comprehensive Examination of Infrastructure, User, Vehicle, and Situational Factors. Journal of Safety Research, Vol. 79, 2021, pp. 14–25.

39.

Hezaveh

A. M.

Azad

Cherry

C. R.

Pedestrian Crashes in Tennessee: A Data Mining Approach. Transportation Research Board 97th Annual Meeting, Washington, D.C., 2018, pp. 7–11.

40.

Pannell

Z. H.

“A Framework to Predict High Risk Roadways for Pedestrians in Tennessee.” Master’s thesis, University of Tennessee, 2013.

41.

National Highway Traffic Safety Administration. MMUCC Guideline: Model Minimum Uniform Crash Criteria. NHTSA, Washington, D.C., 2017.

42.

Federal Highway Administration (FHWA). KABCO Injury Classification Scale and Definitions. https://safety.fhwa.dot.gov/hsip/spm/conversion_tbl/pdfs/kabco_ctable_by_state.pdf

43.

Ewing

Schieber

R. A.

Zegeer

C. V.

Urban Sprawl as a Risk Factor in Motor Vehicle Occupant and Pedestrian Fatalities. American Journal of Public Health, Vol. 93, No. 9, 2003, pp. 1541–1545.

44.

Ghimire

Bardaka

Active Travel among Carless and Car-Owning Low-Income Populations in the United States. Transportation Research Part D: Transport and Environment, Vol. 117, 2023, Article 103627.