Young Unlicensed Drivers and Passenger Safety Restraint Use in U.S. Fatal Crashes: Concern for Risk Spillover Effect?

Abstract

Young unlicensed drivers are more likely to be in fatal crashes and to engage in high-risk driving behaviors like impaired driving, speeding, and driving unrestrained. In a crash context, the influence of these high-risk behaviors may spillover to adversely affect passenger safety restraint use. We conducted an analysis of the Fatality Analysis Reporting System from years 1996–2008. Fatal crashes involving a driver aged 15–24 years and at least one passenger aged 15–24 years were included. Logistic regression with generalized estimating equations was undertaken to assess the effect of unlicensed driving on passenger restraint use. We controlled for established predictors of passenger restraint use including driver restraint use, gender, number of occupants, driver drinking, number of occupants, crash year, and crash location (rural vs. urban). 102,092 passengers were involved in fatal crashes during the time period with 64,803 unique drivers. 6,732 (10.51%) were never licensed drivers and 5,603 (8.8%) were drivers with suspended, revoked, or expired licenses. Rates of unlicensed driving ranged from 17.7% to 22.1% and increased over time. While passengers in fatal crashes averaged a mere 40.9% restraint use, passengers of never and invalidly licensed drivers had a further decreased odds of wearing a safety restraint (OR 0.73, 95% CI 0.69–0.77, p<0.001) and (OR 0.84, 95% CI 0.79–0.90, p<0.001). Unlicensed driving is involved in a disproportionate and increasing number of preventable crash fatalities and plays a detrimental role in the lifesaving safety behaviors of their passengers. Our findings highlight an alarming peer influence between unlicensed drivers and passengers, placing increased emphasis on the need to better understand and characterize this present and growing threat.

INTRODUCTION

Making up over one-third of all teen deaths, motor vehicle collisions (MVCs) continue to be a leading cause of death for adolescents and young adults in the United States [NCSA, 2011; CDC, 2011; McCartt and Northrup, 2004; Vaca, 2007]. The National Highway Traffic Safety Administration (NHTSA) reports that in 2009, while young drivers 15–20 years of age made up 6.4% of drivers in the US, they represented 11% of drivers involved in fatal crashes. In 2009 alone, 2,336 15–20 year-old drivers were killed [NCSA, 2011]. Furthermore, recent reports suggest that the prevention gains in this age group have slowed, as early estimates show that 2011 saw the first increase (11%) in 16–17 year old driver crash deaths in over 8 years [Williams, 2012].

Graduated driver licensing (GDL) programs, which include a stepwise decrease in licensing restrictions such as number of passengers and amount of night time driving allowed, has proven effective in reducing young driver fatal crashes [Dee, Grabowski, and Morrisey, 2005; Hartling, Wiebe, Russell, et al., 2004]. While GDL has proven useful, drivers who circumvent the system and drive unlicensed may not reap the benefits of such safety programs. Several reports have highlighted worrisome estimates of unlicensed driving prevalence in teens and young adults. One self-report study estimated that 4.2% of 9^th through 11^th-graders in the US drove at least one hour/week without a license [Elliott, Ginsburg, and Winston, 2008]. Another self-report study in New Zealand of newly licensed drivers found that as high as 65% of urban and 83% of rural new drivers had driven unlicensed [McDowell, 2009].

Exacerbating inexperience, unlicensed teen drivers have been found to be more likely to engage in risky driving behaviors like seat belt non-use, consuming alcohol, and speeding [Elliott, et al., 2008; Huber Jr, Carozza, and Gorman, 2006; Lam, 2003]. This combination of lack of experience and high risk behaviors leads to an estimated 11-fold increased odds of car crash injury [Blows, Ivers, and Connor, et al., 2005]. The risk of driving unlicensed appears to persist even after obtaining a license. New drivers who had driven unlicensed were twice as likely to get into a MVC during the first 12 months of licensure [Stevenson and Palamara, 2001]. Crashes involving young unlicensed drivers tend to occur on weekends and evenings and more frequently involve young passengers [Hanna, Taylor, Sheppard, et al., 2006; Huber Jr, et al., 2006].

While it is well documented that unlicensed driving endangers occupants because of increased crash risk, little is known about its impact on the passenger risk behaviors. In a crash context, the high-risk behavior of unlicensed driving may have a risk spillover effect, leading to increased passenger risk behaviors. Safety restraint use is unequivocally known to be the single most important behavior to preventing both morbidity and mortality of MVCs. Some studies have found that driver’s safety restraint use is a predictor of passenger safety belt use [Agran, Anderson, and Winn, 1998; Nambisan and Vasudevan, 2007]. Given the low seat belt use and high-risk behaviors of unlicensed drivers, passengers of unlicensed drivers may also have low seat belt use. Such an influence would place passengers of unlicensed drivers at much increased risk for crash injury and death. The objective of our study was to determine the association between unlicensed driving and passenger safety restraint use in teens and young adults involved in fatal crashes.

METHODS:

We explored US fatal MVCs in the years 1996–2008. Data were obtained from the National Highway Traffic Safety Administration’s Fatality Analysis Reporting System (NHTSA-FARS) database. The FARS database is a compilation of all police-reported fatal traffic accidents on public roadways in the United States. It includes accidents that resulted in the death of the driver, an occupant, or a non-occupant within 30 days of the accident.

Case Selection

We included fatal MVCs involving a driver aged 15–24 and at least one passenger also aged 15–24. Age ranges were limited to examine peer to peer influences between drivers and passengers. Cases involving buses, farming equipment, bicyclist, pedestrians, and other non-passenger vehicles or vans were excluded to focus on passenger vehicles.

Variable Definitions

Driver license status was categorized as licensed and unlicensed. Unlicensed driving was further divided into invalidly licensed (revoked, expired, suspended) and never licensed. Provisional license status was inconsistently collected and made up only 1% of cases. As such, they were excluded from the analysis. Passenger status was determined based on seating position. Passengers in row one were coded as front seat, and passengers in rows two or three were coded as rear seat. Passengers with incorrect or no restraint use were coded as unrestrained.

Data Analysis

Variables of interest were license status, driver restraint use, passenger gender, driver alcohol use, number of occupants, crash year, and crash location (rural vs. urban). Race was a variable of interest, but was not collected consistently during this time period, so it was not included in the modeling. Chi-square analysis and univariate logistic regression were used to assess variables with putative associations with passenger restraint use. These variables were controlled for in the multivariate model. To account for passengers in the same vehicle having the same driver, logistic regression with generalized estimating equations (GEE) analysis, clustering on the vehicle, was carried out. Analysis was done with Stata 11.0, using xtgee and the vce(bootstrap) option to better estimate standard errors and confidence intervals [Hanley, 2003]. Results were expressed as odds ratios (ORs) with 95% confidence intervals. This study was registered with the Yale University Human Investigations Committee and exempted from review.

RESULTS

From 1996–2008, there were 64,803 unique 15–24 year-old drivers and 102,092 15–24 year-old passengers involved in fatal crashes. 10.5% of these drivers had never been licensed, and 8.8% had a suspended, revoked, or expired license. 80.7% of drivers were driving with a valid license. Rates of invalid or never licensed driving ranged from 17.7% to 22.1% and increased over time.

Males were more likely than females to be driving with a suspended, revoked, or expired license (10.1% vs. 4.5%) (Table 1). Males were also more likely to drive if they had never had a driver’s license (10.5% vs. 8.0%). 15 year-old drivers involved in fatal crashes were more likely than young drivers of other ages to drive without a license. Rates of invalidly licensed driving increased from 1.9% in 15 year olds to 14.8% in 24 year olds. As drivers increased in age, they had more time to have their license suspended, revoked, or expired. 30.5% of drivers were driving under the influence of alcohol.

Table 1.

Driver Demographics by Licensing Status

Driver Demographic		Licensed	Invalid License	Never Licensed	p
Gender	Female	13,443 (87.5%)	685 (4.5%)	1,229 (8%)
	Male	38,259 (78.6%)	4,918 (10.1%)	6,731 (10.5%)	<0.001
Age	15	313 (35.8%)	17 (1.9%)	545 (62.3%)
	16	5,240 (85.1%)	97 (1.6%)	820 (13.3%)
	17	6,989 (84.8%)	313 (8.8%)	937 (11.4%)
	18	8,573 (84.9%)	595 (5.9%)	935 (9.3%)
	19	7,473 (82.8%)	765 (8.5%)	787 (8.7%
	20	6,198 (80.1%)	836 (10.8%)	704 (9.1%)
	21	5,704 (79.1%)	882 (12.2%)	628 (8.7%)
	22	4,546 (76.4%)	825 (13.8%)	578 (9.7%)
	23	3,715 (76.6%)	698 (14.4%)	439 (9.1%)
	24	2,952 (76.0%)	575 (14.8%)	359 (9.2%)	<0.001

Characterizing passengers, 64,752 (63.5%) were males. 60,679 (60.9%) of these passengers were sitting in the front seat compared to 38,898 (39.1%) in the rear seats (Table 2). 51.4% of crash vehicles had one passenger, 25.0% had two passengers, 14.3% had three passengers, and 9.3% had four or more passengers. 58.0% of crashes occurred on roadways classified as rural.

Table 2.

Passenger Restraint Use in Fatal Crashes by Driver, Passenger, and Crash Characteristics

Characteristic		Restraint Use	n	Unadjusted OR	95% CI
Driver License Status	Licensed	43.5%	74,088	0.60* 0.62*	[0.57, 0.63] [0.59, 0.65]
	Invalid License	30.3%	7,805
	Never Licensed	31.1%	10,087
Passenger Gender	Female	46.7%	34,352	0.69*	[0.67, 0.71]
	Male	37.6%	58,810
Passenger Seat Position	Front Seat	50.6%	55,989	2.73*	[2.65, 2.81]
	Rear Seat	27.3%	35,522
Number of Occupants	2	52.4%	30,742	0.71*	[0.70, 0.71]
	3	40.1%	24,743
	4	37.4%	19,822
	5+	26.3%	17,865
Crash Location	Urban	47.1%	37,375	0.65*	[0.64, 0.67]
	Rural	36.8%	55,475
Driver Alcohol Use	No	47.1%	63,683	0.43*	[0.41, 0.44]
	Yes	27.5%	29,521
Driver Restraint Use	Unrestrained	13.9%	42,869	12.0*	[11.6, 12.4]
	Restrained	66.0%	46,861

^*significant with p<0.001

Overall restraint use among passengers averaged 40.9% compared to 52.8% among drivers. Passengers of never licensed drivers wore their seat belt 31.1% of the time versus 30.3% in passengers of drivers with invalid licenses. Passengers of licensed drivers had the highest restraint use of 43.5% (p<0.001). From 1996–2008, overall passenger restraint use increased from 31.9% to 49.8%. Restraint use in passengers of licensed drivers increased from 34.1% to 52.2%, compared to restraint use in passengers of invalidly licensed drivers (23.4% to 39.2%) and drivers who had never been licensed (18.8% to 41.0%) (figure 1).

Figure 1.

Passenger Restraint Use by Driver License Status and Year

Restraint use among young passengers in fatal crashes varied by gender and seat position. On average, males wore safety restraints 37.6% of the time compared to 46.7% in females. Front seat passengers were much more likely than rear seat passengers to wear a safety restraint (50.6% vs. 27.3%). Restraint non-use increased with increasing number of occupants in the vehicle. 52.4% of passengers in vehicles with two occupants wore their safety restraint. This percentage trended downwards to 26.3% when there were five or more occupants in the vehicle. Restraint use also varied with driver drinking and crash location. Passengers in vehicles with driver alcohol use wore safety restraints 27.5% compared to passengers of drivers who were not drinking alcohol 47.1%. Passengers in rural crashes wore safety restraints 36.8% compared to passengers in urban crashes 47.1%.

In the final model, controlling for passenger gender, passenger seat position, number of occupants, crash location, driver drinking, driver restraint use, and crash year, unlicensed driving was a statistically significant risk factor for passenger restraint non-use. Compared to passengers of licensed drivers, passengers of never licensed and invalidly licensed drivers had a decreased odds of restraint use (OR 0.73, 95% CI 0.69–0.77) and (OR 0.84, 95% CI 0.79, 0.90).

Driver restraint use was associated with the largest increase in odds of restraint use (OR 15.40, 95% CI 14.71–16.11). Being a passenger in the front seat compared to in the rear seats was also associated with an increased odds of restraint use (OR 3.61, 95% CI 3.47–3.74). Crash year was associated with a 5% increased odds of restraint use per year from the 1996 baseline.

Driver drinking, crashes in rural locations, being a male passenger, and crashes involving increased number of occupants were associated with a decreased odds of passenger restraint use. Each additional occupant in the vehicle was associated with an additional decreased odds (OR 0.84, 95% CI 0.83–0.86).

Restraint use was missing for 8.7% of passengers. Missingness was similar across calendar years. There were slightly more missing data points in passengers of never and invalidly licensed compared to validly licensed drivers (10.5% vs. 10.5% vs. 8.2%). There were also more missing points in urban vs. rural crashes (12.4% vs. 6.0%) and for males vs. females (9.2% vs. 7.9%).

DISCUSSION

Our study found that unlicensed driving was associated with passenger restraint non-use in unadjusted analyses. In addition, we found a strong inverse relationship between teen/young adult unlicensed driving and passenger restraint use when controlling for other predictors.

The influence unlicensed drivers have on their passengers may be explained by several developmental characteristics of risk, including: friend influence and optimistic bias [Arnett, 2002]. Contextual features, coined the “friend influence,” suggest that for adolescents and young adults, perception and judgment by peers may be most important. One study of fatal crashes found that having others in the car increased crash risk for drivers under 30 but decreased crash risk for those over 30 [Preusser, Ferguson, and Williams, 1998]. Optimistic bias is the mindset that despite high-risk behaviors, poor outcomes are more likely to befall other people. A young adult getting into the car with an unlicensed driver exhibits this type of bias and will most likely be less likely to wear a safety restraint. One dynamic that our study did not look at was gender or race interplay. One study suggests that the presence of a young female passenger led to safer driving practices for young male drivers, whereas a young male passenger was associated with more dangerous driving for both male and female drivers [Mckenna, 1998]. Another study found that female and male drivers risk behaviors were affected differently by their passengers, especially regarding aggressiveness and distraction [Curry, 2012]. It is also possible that these effects could vary across cultures.

Despite reports of persistently high rates of unlicensed driving [AAA foundation, 2008], passenger restraint use continued to increase from 1996–2008, suggesting that education campaigns and enforcement programs like “click it or ticket” have been effective [Solomon, Preusser, Tison, et al., 2009]. In multivariate models, each year from baseline 1996 was associated with positive odds of restraint use. While this is indeed good news, the overall restraint use in fatal crashes is far from promising.

Sitting in rear seats was associated with a much decreased odds of restraint use. Rear seat passengers in this population wore safety restraints a mere 27.3%, leaving significant room for improvement. Such low restraint use suggests that programs are missing rear seat passengers. Being in the rear seat may come with a false sense of security that can lead passengers to wear their restraints less. This misconception may be perpetuated by typically less stringent rear seat restraint legislation and enforcement. For both front and rear passengers, however, safety restraint use is associated with a significant decreased risk of ejection and death. Future education campaigns and enforcement programs may prove more successful if they emphasize rear seat restraint use.

Our findings support prior work that showed an association between driver restraint use, passenger gender, crash location, crash year, alcohol involvement, increased number of occupants, and passenger restraint use [McCartt and Northrup, 2004]. Alcohol has been associated with restraint non-use [Allen, Zhu, Sauter, et al., 2006; Gross, Axberg and Mathieson, 2007], and we also found that alcohol involvement had a significant effect on passenger restraint use in our study population. Rural crash location was also found to be a predictor in a study involving adult emergency department patients in Wisconsin [Allen, et al., 2006]. Another study of Swedish unlicensed drivers found that unlicensed drivers in rural areas had a much greater risk for crashes [Hanna, 2010].

This study included only crashes found in the FARS database, limiting its conclusions to fatal motor vehicle crashes on public motorways. Fatal crashes may involve more unlicensed driving and high-risk behaviors, so data may be skewed and less easily generalized to the population of all crashes. The database consistency, completeness, and accuracy are dependent on the data collection of many different law enforcement officers. Under high stress situations, law enforcement officers may make data collection and even their own safety less of a priority as seen by sometimes conflicting and missing data points [Fu and Vaca, 2011]. While the National Highway Traffic Safety Administration, manages FARS with well validated consistency checks and statistical control charts to optimize validity, missing data points are an inherent limitation to this database.

CONCLUSION

Teens and young adults continue to have the lowest restraint use and the highest crash fatality rates of any age group. Our study found that a large portion of these deaths involve young unlicensed driving. Passengers of unlicensed drivers had a decreased odds of wearing safety restraints compared to the passengers of licensed drivers, placing them at much increased risk of crash injury. Young passengers are especially susceptible to the risk influences of their peers, creating a unique opportunity for targeted intervention. Our findings highlight a risk spillover effect that has significant implications for highway safety and injury prevention programs. Further in-depth study of driver-passenger peer interactions can guide future countermeasures and traffic safety programs.

Table 3.

Multivariate Logistic Regression Model for Passenger Restraint Use

Variable	OR	Bootstrap SE	95% CI	P value
Never Licensed Driver	0.73	0.02	[0.69, 0.77]	<0.001
Invalidly Licensed Driver	0.84	0.03	[0.79, 0.90]	<0.001
Male Passenger	0.81	0.02	[0.78, 0.85]	<0.001
Front Seated Passenger	3.61	0.07	[3.47, 3.74]	<0.001
Number of Occupants	0.84	0.01	[0.83, 0.86]	<0.001
Rural Crash	0.71	0.01	[0.68, 0.74]	<0.001
Driver Alcohol Use	0.74	0.02	[0.70, 0.77]	<0.001
Driver Restraint Use	15.40	0.36	[14.71, 16.11]	<0.001
Crash Year	1.05	0.01	[1.05, 1.06]	<0.001