Riding With an Impaired Driver and Driving While Impaired Among Adolescents: Longitudinal Trajectories and their Characteristics

Abstract

Purpose:

To identify and characterize trajectory classes of adolescents that ride with an impaired driver (RWI) and drive while impaired (DWI).

Methods:

We analyzed all seven annual assessments (Waves, W1-W7) of the NEXT Generation Health Study, a nationally representative longitudinal study starting with 10^th grade (2009-2010 school year). Using all seven waves, the latent class analysis was used to identify trajectory classes with dichotomized RWI (last 12 months) and DWI (last 30 days) (≥once=1 vs. none=0). Co-variates were race/ethnicity, sex, parent education, urbanicity, and family affluence.

Results:

Four RWI trajectories and four DWI trajectories were identified: Abstainer, Escalator, Decliner, and Persister. For RWI and DWI trajectories respectively, 45.0% (N=647) and 76.2% (N=1,657) were abstainers, 15.6% (N=226) and 14.2% (N=337) were escalators, 25.0% (N=352) and 5.4% (N=99) were decliners, 14.4% (N=197) and 3.8% (N=83) persisters. Race/ethnicity (χ²=23.93, p=.004) was significantly associated with the RWI trajectory classes. Race/ethnicity (χ²=20.55, p=.02), sex (χ²=13.89, p=.003), parent highest education (χ²=12.49, p=.05), urbanicity (χ²=9.66, p=.02), and family affluence (χ²=12.88, p=.05) were significantly associated with DWI trajectory classes.

Conclusions:

Among adolescents transitioning into emerging adulthood, race/ethnicity is a common factor associated with RWI and DWI longitudinal trajectories. Our results suggest that adolescent RWI and DWI are complex behaviors warranting further detailed investigation of the respective trajectory classes. Our study findings can inform the tailoring of prevention and intervention efforts aimed at preventing illness/injury and preserving future opportunities for adolescents to thrive in emerging adulthood.

INTRODUCTION

Riding with an impaired driver (RWI) is as high as 38% among U.S. high school students (Li, Simons-Morton, Brooks-Russell, Ehsani, & Hingson, 2014). No major decrease has occurred since 2013 (Kann et al., 2016; Li, Simons-Morton, Brooks-Russell, et al., 2014). Driving while impaired (DWI) among high school teens is as high as 14% (Li, Simons-Morton, Brooks-Russell, et al., 2014). Our previous research shows that earlier substance use and heavy episodic drinking predicts future RWI among emerging adults and factors predicting DWI among teens include heavy episodic drinking, perceived peer norms of alcohol/substance use, low parental monitoring knowledge, and exposure to RWI (Li, Simons-Morton, Brooks-Russell, et al., 2014; Li, Simons-Morton, Vaca, & Hingson, 2014; Vaca et al., 2020). Other studies also demonstrate a close relationship between RWI and more frequently studied DWI, revealing that exposure to RWI increases the likelihood of future DWI among teens after licensure (Harris et al., 2017; Li, Simons-Morton, Vaca, et al., 2014).

In the past decade, a substantial body of research has documented an increased trend in crashes among youth caused by binge drinking, drug use, or the combination of binge drinking and drug use (Elvik, 2013). In 2015, the U.S. encountered the largest year-to-year percent increase (i.e., 7.2%) in crash fatalities in 50 years (National Highway Traffic Safety Administration, 2016b). This included a significant increase in alcohol-impaired driving fatalities encompassing 1 in every 5 alcohol-related crash deaths occurring in passengers and 1 alcohol-related crash fatality occurring every 51 minutes of every day (National Highway Traffic Safety Administration, 2016a). Furthermore, 16-19 y/o young novice drivers who used drugs had the highest risks of crashes (Bates, Davey, Watson, King, & Armstrong, 2014). Complicating the current and immediate future landscape of impaired driving crashes is the number of states seeking legislation to legalize marijuana use. This will plausibly continue to impact the number of impaired driving crashes. While relatively less is known about marijuana-related property damage only and injury crashes, studies show that the number of marijuana-positive drivers involved in a fatal crash increases in states that legalize its use (Tefft & Arnold, 2020). The National Highway Transportation Safety Administration (NHTSA) notes that drivers that had previously driven while drug-impaired accounted for 7.3% of all traffic fatalities involving alcohol in 2003 (Subramanian, 2005) and increased fourfold (28%) in 2016 (National Center for Statistics and Analysis, 2017).

Personal transportation and driver licensure continue to be a major developmental milestone for the adolescents. Both are intimately tied to greater independence and facilitate a broad array of immediate and later social and formal opportunities (e.g., employment). However, with driving licensure comes the need for major individual and social responsibility. This is particularly true when considering that driving peer-passengers heightens the risk of a serious crash (Ouimet et al., 2010). For a young driver, being convicted or incarcerated as a result of a DWI, could threaten immediate health as a result of risk of physical injury to self and/or others, and limit opportunity for college matriculation or employment as well as compromise later-related opportunities to thrive in life. Further examination is needed to advance our understanding of individual and social-environmental determinants of RWI and DWI behaviors.

The main purpose of this study was to identify and characterize the trajectory classes of longitudinal RWI and DWI engagement among adolescence as they transition into early adulthood.

METHODS

Sampling

The data were from all seven annual waves (W1-W7; 2010-2016) of the NEXT Generation Health Study, a longitudinal nationally representative cohort study starting in10^th grade. The sampling strategy for NEXT has been previously reported. (Li, Simons-Morton, Brooks-Russell, et al., 2014) From W1-W7, 91% (16.27 years, se=0.03) (260 recruited students were unable to complete the study at W1 due to delayed approval from the school district), 88% (17.19 years, se=0.03), 86% (18.17 years, se=0.03), 78% (19.16 years, se=0.02), 79% (20.28 years, se=0.02), 84% (21.28 years, se=0.02), and 83% (22.64 years, se=0.03) of the full sample (N=2785) completed the survey during spring each year. African American participants were oversampled to allow for more accurate population estimates and representativeness. Parent consent and participant assent were obtained for those under age 18. After turning age 18, participants were consented as adults. The study protocol was approved by the Institutional Review Board of the Eunice Kennedy Shriver National Institute of Child Health and Human Development.

Trajectory variables

Riding with an alcohol or drug-impaired driver in the past 12 months (RWI, W1-W7)

At W1-W4, RWI was measured using one question derived from the Youth Risk Behavior Survey (YRBS) (Centers for Disease Control and Prevention, 2011) asking participants how many times, during the last 12 months, they rode in a vehicle driven by someone else who had been drinking alcohol or using illegal drugs. At W5-W7, RWI was measured by three questions. Participants were first asked: “During the last 12 months, how many times did you ride in a vehicle driven by someone who had been drinking alcohol?” The same question was repeated for “smoking marijuana” and “using illicit drugs other than alcohol or marijuana.” The questions at W1-W4 and W5-W7 were collapsed and dichotomized as: 1=RWI ≥once vs. 2=no RWI (last 12 months).

Driving while alcohol- or drug-impaired in the past 30 days (DWI, W1-W7)

At W1-W3, DWI was measured using one question derived from the YRBS questionnaire (Centers for Disease Control and Prevention, 2011) asking participants on how many days in the last 30 days they drove after drinking alcohol or using illegal drugs. At W4-W7, DWI was assessed with three items asking participants how many days they drove after drinking alcohol, smoking marijuana, or using illicit drugs. The questions at W1-W3 and W4-W7 were collapsed and dichotomized as: 1=DWI ≥1 day vs. 2=no DWI (past 30 days).

Characteristic variables

Participants reported sex (male vs. female), race/ethnicity (Whites, Hispanic/Latinos, African Americans, and Others), parental education, family affluence and urbanicity as demographic variables. Parental education was the highest level of education attainment of either parent (mother or father) as reported by the consenting parent. Family affluence was measured at W1 and W2 using the Family Affluence Scale (Currie et al., 2008), including number of cars and computers owned, if student had his/her own bedroom, and number of family vacations in the last 12 months. Then, students were categorized into a three-level category: low, moderate, or high affluence level primarily based on W1 data and supplemented by W2 data if W1 data were missing. Urbanicity was defined as participants’ school location at W1 using a seven-level scale ranging from metropolitan city to rural region. Those attending schools in an urban region were categorized as urban, and remaining categories were classified as non-urban (i.e., suburban/rural).

Statistical analysis

We used Latent Class Analysis (LCA) to identify a set of discrete mutually exclusive latent classes of individuals based on their responses to a set of observed categorical variables. The LCA model fit was assessed with log-likelihood, AIC, BIC, and Entropy for RWI/DWI trajectory classes. For absolute log-likelihood, AIC, BIC, CAIC, ABIC, and G2 lower values were preferred but for Entropy, higher values indicate better separation/interpretability of classes (Lanza, Dziak, Huang, Xu, & Collins, 2015).

Separately for RWI and DWI, we conducted Rao-Scott Chi-square test to examine the bivariate associations between the trajectory classes from the LCA and characteristic variables. We then conducted multivariate multinomial logistic regression models to examine the association between the RWI/DWI trajectory classes and characteristic variables using the abstainers group as the reference category in the outcome variable. The computations were performed in SAS software version 9.4 (SAS Institute, Cary, NC) and PROC LCA procedure (PROC LCA & PROC LTA (Version 1.3.2) [Software]) taking into consideration complex sampling features.

RESULTS

Riding with an impaired driver (RWI)

For RWI, 3-, 4- and 5-class models were estimated from the LCA. All model fit statistics of the 4-class model were better than the 3-class model. The absolute log-likelihood of the 5-class model (−5102.6) was slightly lower than the 4-class model (−5123.2). None of the other model fit statistics were better than the 4-class model. Therefore, the 4-class model (Figure 1) was selected for the analysis. The four classes included abstainer (consistently low probability over 7 waves), escalator (low to high probability over 7 waves), decliner (high to low probability over 7 waves), and persister (consistently high probability over 7 waves). Among participants, 647 (45.0%, weighted and hereafter) were categorized as abstainer, 226 (15.6%) escalator, 352 (25.0%) decliner, and 197 (14.4%) persister (Table 1).

Figure 1.

Trajectories of riding with an alcohol or drug-impaired driver (RWI) in the past 12 months using latent class analysis

Table 1.

Frequency of RWI, 4-group trajectories by demographic variables

	Abstainers		Escalators		Decliners		Persisters
	N	%#	N	%#	N	%#	N	%#	N	%#	Rao- Scott χ2	P
	647	45.0	226	15.6	352	25.0	197	14.4	1422	100	-	-
Race
Latino	197	45.2	73	10.7	124	34.6	57	9.4	451	100	23.93	0.004
AA	123	45.2	51	17.1	76	28.5	29	9.2	279	100
White	278	44.0	95	16.6	137	21.1	104	18.3	614	100
Others	49	59.2	7	15.5	14	21.0	7	4.3	77	100
Sex
Male	235	34.7	82	32.5	125	34.3	81	38.5	523	100	0.70	.87
Female	235	44.9	82	14.5	125	24.7	81	15.9	899	100
Parent Education
High Sch.	221	42.0	66	12.9	144	32.1	71	13.0	502	100	10.34	.11
Some Col.	230	49.4	84	13.4	120	23.1	69	14.1	503	100
Bachelor+	157	41.9	67	21.8	68	20.6	47	15.6	339	100
Urban/Rural
Urban	247	47.6	85	15.1	133	22.1	68	15.2	533	100	1.16	.76
Non-urban	400	44.6	141	15.6	219	25.6	129	14.2	889	100
Affluence
Low	183	43.2	71	18.7	119	24.0	48	14.1	421	100	5.06	.54
Moderate	321	46.2	95	13.4	172	27.0	89	13.4	677	100
High	143	44.3	60	17.2	61	22.2	60	16.3	324	100

Note:

^#weighted %. Others: including Asian, American Indian or Alaska Native, and Native Hawaiian or Other Pacific Islanders.

Race/ethnicity (χ²=23.93, p=.004) was significantly associated with the RWI trajectory classes. Proportions of abstainers were the highest among all racial/ethnic groups (Latinos (45.2%); African Americans (45.2%); and Whites (44.0%) compared to escalators, decliners, and persisters. Proportions of decliners were second highest among all racial/ethnic groups (Latinos (34.6%); African Americans (28.5%); and Whites (21.1%). Latinos (10.7%) had lower proportion of escalators compared to African Americans (17.1%) and Whites (16.6%). Whites (18.3%) had about a twofold higher proportion of persisters compared to Latinos (9.4%) and African Americans (9.2%).

Driving while alcohol- or drug-impaired (DWI)

For DWI, 3-, 4- and 5-class models were estimated from the LCA. All model fit statistics of the 4-class model were better than the 3-class model. The 5-class model did not converge. Therefore, the 4-class model (Figure 2) was selected for the analysis. As shown in Table 2, among DWI participants, 1657 (76.2%) were categorized as abstainer, 337 (14%) escalator, 99 (5%) decliner, and 83 (4%) persister.

Figure 2.

Trajectories driving while alcohol- or drug-impaired (DWI) in the past 30 days using latent class analysis

Table 2.

Frequency of DWI, 4-group trajectories by demographic variables

	Abstainers		Escalators		Decliners		Persisters		Total
	N	%#	N	%#	N	%#	N	%#	N	%#	Rao- Scott χ2	P
	1657	76.2	337	14.2	99	5.4	83	3.8	2176	100	-	-
Race
Latino	453	83.0	78	9.1	16	5.2	17	2.7	564	100	20.55	.02
AA	371	72.4	96	19.0	14	2.5	17	6.2	498	100
White	740	74.8	145	14.4	64	5.9	48	4.9	997	100
Others	85	65.2	16	23.2	5	11.4	1	0.2	107	100
Sex
Male	716	68.8	176	18.4	53	6.0	52	6.7	997	100	13.89	.003
Female	940	80.7	161	11.7	46	4.8	31	2.7	1178	100
Parent Education
High Sch.	527	77.8	99	14.3	22	4.5	21	3.4	669	100	12.49	<.05
Some Col.	600	78.2	105	12.7	40	6.2	29	2.9	774	100
Bachelor+	391	68.7	102	18.6	29	5.7	25	7.0	547	100
Urban/Rural
Urban	477	79.2	83	14.7	17	2.8	16	3.4	593	100	9.66	.02
Non-urban	1050	75.2	210	14.4	71	5.7	62	4.8	1393	100
Affluence
Low	477	82.0	83	11.6	15	2.0	16	4.4	591	100	12.82	<.05
Moderate	802	74.9	162	15.0	57	5.1	39	5.0	1060	100
High	376	71.7	92	16.4	27	8.1	28	3.8	523	100

Note:

^#weighted %. Others: including Asian, American Indian or Alaska Native, and Native Hawaiian or Other Pacific Islanders.

Race/ethnicity (χ²=20.55, p=.02), sex (χ²=13.89, p=.003), parent highest education (χ²=12.49, p<.05), urbanicity (χ²=9.66, p=.02), and family affluence (χ²=12.88, p<.05) were significantly associated with DWI trajectory classes. The proportions of abstainers were the highest in all demographic groups. Latinos had lower proportions of escalators (9.1% vs. 19.0% and 14.4%) and persisters (2.7% vs. 6.2% and 4.9%) compared to African Americans and Whites, respectively. Males had higher proportions of escalators (18.4 vs. 11.7%), decliners (6.0 vs. 4.8%), and persisters (6.7% vs. 2.7%) compared to females. Those whose parent had a bachelor or higher degree had higher proportions of escalators (18.6% vs. 14.3% and 12.7%) and persisters (7.0% vs. 3.4% and 2.9%) compared to those whose parent had high school or some college education, respectively. Non-urban participants had higher proportions of decliners (5.7% vs. 2.8%) and persisters (4.8% vs. 3.4%). High-affluence participants had higher proportions of escalators (16.4% vs. 15.0% vs. 11.6%) and decliners (8.1% vs. 5.1% vs. 2.0%), but lower proportion of persisters (3.8% vs. 5.0% vs. 4.4%) compared to moderate- and low-affluence participants, respectively.

DISCUSSION

Today, adolescents face major challenges to their safety, development, and health while living in a setting where national crash fatalities are high and, year-to-year, nearly 1/4^th of all young drivers deaths are among impaired drivers (National Conference of State Legislatures, 2016). Youth that engage in RWI and DWI could compromise their opportunities for optimal development and advancement in emerging adulthood. Our results show that longitudinal trajectories of RWI were associated with race/ethnicity (i.e., proportion of the race/ethnicity group that was correlated specifically with the RWI abstainer, escalator, decliner, persister trajectory class). Comparatively, DWI trajectories were associated with several more demographic characteristics including race/ethnicity, sex, parent education, urbanicity, and family affluence.

To the best of our knowledge, this is the first study to simultaneously identify RWI and DWI longitudinal trajectories and examine their possible determinants. The identification of RWI and DWI trajectory classes sheds an important light on the complexity of these two different, yet related, behaviors across adolescence in order to move our understanding in this area forward. Furthermmore, this study also hightlights and provides more evidence of how social-contextual factors may influence longitudinal RWI and DWI trajectories among adolescents into emerging adulthood; a time when DWI prevalence and the proportion of drivers involved in an alcohol-related fatal crash is at its highest (27% for the 21- to 24-year-old age group) (National Center for Statistics and Analysis, 2019, December).

Our study shows that race/ethnicity is associated with longitudinal trajectories (7-annual waves of assessment) of both RWI and DWI behaviors. Previous studies show that the prevalence of both RWI and DWI is higher for Latino than White high school students (Yellman, Bryan, Sauber-Schatz, & Brener, 2020). Other studies have found that RWI is more prevalent than DWI among Latino youth when compared to their White counterparts among adolescents aged about 17 (Li, Simons-Morton, & Hingson, 2013).

Though an earlier study using the NEXT data suggested that neither RWI or DWI had an association with family affluence and some levels of parent education (Li et al., 2013), we identified associations between DWI longitudinal trajectories and these factors as well as several other important demographic characteristics (i.e., sex, urbanicity). These demographic-related associations could provide instrumental insight and instruction for the intentional tailoring and bolstering of promising adolescent prevention education/intervention strategies, particularly for more vulenreable youth. This may also be of considerable relevance given that a recent study of alcohol/drug screening, brief intervention, and referral to specialized treatment (SBIRT) was shown to improve health, development, and well-being outcomes in later adolescence and early adulthood (Sterling et al., 2019).

We believe our study findings have important implications for development and education as well as health-related policy (Fell, Scherer, Thomas, & Voas, 2016). To begin with, adolescents that engage in RWI and/or DWI are a key public health concern given their vulnerability, physically in terms the risk for crash-injury (National Highway Traffic Safety Administration, 2017), and behaviorally as a result of neurotoxic effects on the developing brain due to alcohol/drug exposure (Jones, Lueras, & Nagel, 2018). As a result, one goal for primary intervention in this context should be directed at increasing attentiveness and prevention-action orientation toward the mitigation and control of underage drinking as well as impaired driving during high school and within communities that have large adolescents populations (Harding et al., 2016). Policy makers should revisit well-known effective state-level underage drinking laws (e.g., minimum legal drinking age, zero-tolerance) and reassess the quality and extent of state-level implementation that could save many more lives (Fell et al., 2016).

We acknowledge limitations of our study. First, in the NEXT study, the eligibility of driver licensure increased among participants over the time of the study and the DWI variable used was inclusive of licensure. This influenced variability in our sample size and related analyses. Second, we conducted the longitudinal trajectory analysis based on dichotomous outcomes of RWI and DWI due to skewed distributions. As such, the transition probabilities and their correlates are only for “No” RWI and DWI to “Yes” RWI and DWI. The magnitude (i.e., frequency) of those outcomes were left for future investigation. Third, RWI and DWI were measured based on different time periods (i.e., RWI: in the last 12 months; DWI: within the last month). This may have contributed to some of the under-reporting relationships between RWI and DWI longitudinal trajectory classes and their characteritics.

Among adolescents transitioning into emerging adulthood, race/ethnicity is a common factor associated with RWI and DWI longitudinal trajectory classes. A more broad array of demographic characteristics are associated with DWI trajectory classes pointing to the complexity of these behaviors over time warranting further detailed investigation of the repective trajectory classes. Our initial study findings can inform the tailoring of prevention and intervention efforts aimed at preventing illness/injury and preserving future opportunities for adolescents to thrive in emerging adulthood.