©
Annals of Advances in Automotive Medicine
ADOLESCENT BALLOON ANALOG RISK TASK AND BEHAVIORS THAT INFLUENCE RISK OF MOTOR VEHICLE CRASH INJURY 1,2
Federico E. Vaca, MD, MPH; 2Jessica M. Walthall, BS; 3Sheryl Ryan, MD; 4 Alison Moriarty-Daley, APRN; 5Antonio Riera, MD; 2Michael J. Crowley, PhD*; 2 Linda C. Mayes, MD* 1
Yale School of Medicine, Department of Emergency Medicine, New Haven, Connecticut 2 Yale School of Medicine, Child Study Center, New Haven, Connecticut 3 Yale School of Medicine, Department of Pediatrics, New Haven, Connecticut 4 Yale School of Nursing, New Haven, Connecticut 5 Yale School of Medicine, Pediatric Emergency Medicine, New Haven, Connecticut * shared senior authorship __________________________________
ABSTRACT – Risk-taking propensity is a pivotal facet of motor vehicle crash involvement and subsequent traumatic injury in adolescents. Clinical encounters are important opportunities to identify teens with high risk-taking propensity who may later experience serious injury. Our objective was to compare self-reports of health risk behavior with performance on the Balloon Analog Risk Task (BART), a validated metric of risk-taking propensity, in adolescents during a clinical encounter. 100 adolescent patients from a hospital emergency department and adolescent health clinic completed a computer-based survey of self-reported risk behaviors including substance use behaviors and behaviors that influence crash involvement. They then completed the BART, a validated laboratorybased risk task in which participants earn points by pumping up a computer-generated balloon with greater pumps leading to increased chance of balloon explosion. 20 trials were undertaken. Mean number of pumps on the BART showed a correlation of .243 (p=.015) with self-reported driver/passenger behaviors and attitudes towards driving that influence risk of crash injury. Regression analyses showed that self-reports of substance use and mean number of pumps on the BART uniquely predict self-reports of behaviors influencing the risk of crash injury. The BART is a promising correlate of real-world risk-taking behavior related to traffic safety. It remains a valid predictor of behaviors influencing risk of crash injury when using just 10 trials, suggesting its utility as a quick and effective screening measure for use in busy clinical environments. This tool may be an important link to prevention interventions for those most at-risk for future motor vehicle crash involvement and injury. __________________________________
INTRODUCTION Adolescence has been described as a dual-paradox life stage in which young people encounter rapid increases in physical strength and capacity for decision-making yet they remain highly vulnerable to morbidity and mortality related to increased risktaking and novelty seeking behavior [Arnett, 1992; Dahl, 2004; DiClemente et al., 1996]. In the United States in 2010 alone, 42% of deaths among teens and young adults age 15-24 years were due to unintentional injuries; 57% of were a result of motor vehicle crashes [Centers for Disease Control and Prevention, 2012]. Young drivers made up only 6.4% of licensed drivers in the US, yet in 2010 they accounted for 10% of fatal motor vehicle crashes and 14% of the more than 5.4 million police-reported crashes [NHTSA, 2012]. In the same year, almost 282,000 teens were sent to emergency departments (ED) for treatment of crash-related injuries [Centers for Disease Control and Prevention, 2012].
National injury statistics continue to show that adolescents are a highly vulnerable age group for injury-related disability and death. Much of this vulnerability has been popularly attributed to the propensity of adolescents to take risks with limited appraisal of the context or due to psychosocial effects on self-regulation abilities [Institute of Medicine (U.S.) et al., 2011; Steinberg, 2004]. This is particularly true when it comes to teens operating motor vehicles or when riding as an occupant with a teen driver. While Graduated Driver Licensing (GDL) laws, enforcement, and national education initiatives have helped lower the fatalities involving teen drivers in the past decade [Williams, 2012], for some teens, this downward trend appears to have ended in 2011. A recent report showed that 2011 driver deaths among 16- and 17-year-olds increased for the first time since 2002 [Williams, 2012]. Further, more recent preliminary national crash data
CORRESPONDING AUTHOR: Jessica Walthall, Child Study Center, 230 S. Frontage Road, NIHB G-10, New Haven, CT Email: [email protected]
57th AAAM Annual Conference Annals of Advances in Automotive Medicine September 22-25, 2013
77
Vol 57 • September 2013 results from the first six months of 2012 show that the number of driver deaths in this same age group has increased, up 19% from 2011 [Williams, 2012]. The worsening of national fatal crash statistics for these young drivers should serve as a call for renewed and refocused efforts in crash injury prevention. The pursuit to understand the developmental, behavioral, and contextual factors that contribute to teen crashes should be even more urgent. Developing new and effective prevention methods with a focus on the dual paradox of the adolescent developmental period is important and may help in the identification and attenuation of risktaking propensity in some adolescents. Risk-taking behavior has been most often studied with self-report instruments [DiClemente et al., 1996; Gullone and Moore, 2000; Jessor and Jessor, 1977]. These approaches, though useful, may yield incomplete results when used as the sole assessment method, especially among adolescent populations. Self-report measures are subject to response biases such as social desirability and require a willingness to provide truthful responses, as well as an insight and ability to provide an accurate self-report that adolescents may lack (e.g., [Bernstein et al., 2002; Hyman and Loftus, 1998; Ladouceur et al., 2000; Tversky and Koehler, 1994]. Relying exclusively on a unimethod assessment approach has its own limitations and may not be sufficient to evaluate the multidimensional nature of risk-taking [Eysenck and Eysenck, 1977; Wills et al., 2002]. Therefore, tools that can further explore these multiple facets of risk taking may provide more accurate assessments or predictions of risk-taking propensity. Lejuez and colleagues approached these methodological limitations by developing a computer-based behavioral risk task, the Balloon Analog Risk Task (BART) [Lejuez et al., 2002], in which the participant makes button presses (pumps) to inflate individual balloons for incremented points or money with increasing risk that the balloon will explode. This would lead to the loss of the balloon’s present point or monetary value. In this context, each button depression or pump is a “risky decision” with a finite probability that the balloon will explode, and thus the chance that no money will be earned increases with each pump. Empirical research has shown that the number of pumps on the BART is related to self-reports of real-world risk-taking behavior in both adults and adolescents. In adolescents, BART performance has been shown to be a valid indicator of real-life risk behavior such as substance use (cigarette, alcohol, and illicit drug),
78
gambling, unprotected sexual intercourse, stealing, carrying a weapon, fighting, and helmet and safety restraint misuse [Aklin et al., 2005; Lejuez et al., 2007; Lejuez et al., 2003]. Moreover, the BART has been further validated within clinical populations, effectively distinguishing between typicallydeveloping adolescents and adolescents with conduct and substance use disorder symptoms [Crowley et al., 2006] and differentiating 3,4-methylenedioxy-Nmethylamphetamine (MDMA) users from non-users [Hopko et al., 2006]. Finally, several recent neuroimaging and physiological testing studies have successfully used the BART to identify the neural correlates of risky decision-making [Crowley et al., 2009; Fukunaga et al., 2012; Rao et al., 2008]. To our knowledge, the BART’s relationship with behaviors that could influence the risk of crash injury has not been explored. We believe that this relationship is a worthwhile investigation, particularly because risky driving is generally not an isolated behavior and risk-taking in the BART has been shown to have an association with other lifestyle behaviors that co-vary with risk factors for crashes. Adolescents who smoke cigarettes have a higher likelihood of crash involvement [Hutchens et al., 2008; Lang et al., 1996], and tobacco and marijuana users have a higher instance of traffic violations [Bina et al., 2006]. Teens who misuse alcohol are more likely to drink and drive, be involved in single-car crashes, and ride with impaired drivers [Copeland et al., 1996; Lang et al., 1996; Van Beurden et al., 2005]. Because risky performance on the BART has been shown to relate to these cooccurring behaviors, the BART could be used to identify adolescents who exhibit other behaviors that could influence their risk of crash injury. It could further be used as an interventional component for preventing future teen injury and fatal motor vehicle crash involvement. We believe that the BART could serve as a viable screening tool in an environment in which prevention screening routinely occurs for adolescents. For this reason, we chose to investigate the BART’s utility in the clinical setting. Clinician-patient interactions within the emergency department and primary care setting can become notable “teachable moments” if behaviors related to risk-taking are successfully identified [Lawson and Flocke, 2009]. Much like other screening activities that routinely take place in the clinical setting, the BART may be a useful tool for healthcare providers who have a captive audience of teens during a clinical encounter. In this setting, rapid assessment of risk-taking propensity could take place with the opportunity to intervene.
©
Screenings of adolescent patients within the ED setting have been successful in detecting teenagers who are more likely to engage in specific risky behaviors, including those that contribute to greater risk of pregnancy [Chernick et al., 2012] and injury [Johnston et al., 2002]. Computer-based screenings within primary care settings have also been shown to be effective in identifying adolescents who are more likely to engage in risky behaviors [Anand et al., 2012; Paperny et al., 1990]. Further research has shown that technology-based evaluations and interventions reduce time and personnel burden, are effective, acceptable to patients, and often preferred by adolescent patients [Chisolm et al., 2008; Cunningham et al., 2012; Fein et al., 2010; Harris et al., 2012; Olson et al., 2009; Ranney et al., 2012; Vaca et al., 2010; Walton et al., 2010] Given the very recent unfavorable upward trend in teen crash fatalities, we believe there is a need for technology-based rapid health screening tools that can effectively identify risk-taking propensity in clinical environments. Adolescent patient contact episodes in these settings are important opportunities for implementing adolescent-focused prevention efforts. The goal of our study was to integrate the BART, a validated measure of risk-taking propensity, into two different clinical settings and explore its relationship to well-known behaviors that influence the risk of crash injury. We hypothesized that the BART would account for unique variance in predicting driving risk behavior after accounting for substance use risk behaviors. METHODS We conducted a prospective observational study enrolling adolescent patients from a large urban university tertiary care hospital pediatric ED with approximately 35,000 annual visits as well as a primary care adolescent clinic (AC), part of a pediatric primary care center with nearly 26,000 annual visits. Both centers serve a demographically diverse population reflective of the overall city population which is 27% Latino, 35.4% Black, and 42.6% White. We enrolled adolescent patients aged 14-18 years from the pediatric ED and AC from March 2011 to May 2012. ED patients were primarily identified using the electronic patient tracking system. Adolescents who were categorized at higher triage levels, did not speak English, were unable to provide assent/consent or parent/legal guardian consent, in the custody of the state, under suspicion of abuse or
Annals of Advances in Automotive Medicine
threat to self, or receiving a psychiatric evaluation were excluded from enrollment. Adolescent patients were approached by the trained research associate to discuss the study. If they agreed to participate in the study, they were consented (minors - assent with parental consent, those 18 years of age - consent). Consented participants then completed a computer-based self-report health risk assessment survey and the BART. Upon completion of the study tasks, participants were provided a popular store gift card for their time and effort. Instruments Computer based survey: History of risk-taking behaviors was assessed using a selection of questions from the 2011 Youth Risk Behavior Survey (YRBS) [Centers for Disease Control and Prevention, 2011] which has demonstrated acceptable reliability and validity in the past [Brener et al., 2002]. An inventory of 50 questions pertaining to safety, violence, sexual behavior, and tobacco, alcohol and drug use were chosen. For the purposes of this study, given the association between substance use risk and driving risk, we relied on questions reflecting substance use risk. These items served as our substance use risk behavior index. Additional questions from the 2007 Motor Vehicle Occupant Safety Survey pertaining to traffic safety and attitudes towards traffic safety (safety restraint use, traffic safety laws, speeding, accidents) were included in order to further explore behaviors that could influence the adolescent’s risk of crash injury [NHTSA, 2007]. These items served as our crash injury risk behavior index. All questions were administered using E-Prime software (Psychology Software Tools, Inc.) and delivered as a laptop-based self-administered electronic survey. Balloon Analogue Risk Task (BART): A modified automatic version of the BART [Pleskac et al., 2008] using 20 balloon trials was administered to assess risk-taking propensity. Our version of the BART-AP mirrored Lejuez’s BART-AP but rewarded participants with points instead of money, as previously deemed appropriate for adolescent populations [Lejuez et al., 2007]. Balloon pumps produced points represented visually in a meter, as designed for Lejeuz’s youth-oriented version of the BART (BART-Y) [Lejuez et al., 2007]. Upon initiation of the task, a deflated balloon appeared on the computer screen accompanied by a dial of numbers from 0 to 9, a “clear” button to reset the input number of pumps, and three boxes indicating the total number of points at stake, the total number
79
Vol 57 • September 2013
Figure 1 - BART initial screen
of points earned, and the explosion point for the previously presented balloon (Figure 1). Participants were informed that they would be presented with 20 different balloons (20 trials), and that “the actual number of pumps for any particular balloon will vary.” Participants were instructed to attempt to earn as many points as possible. At the beginning of each trial, the participant decided how many pumps they thought the balloon would hold and input this number using the mouse and the on-screen dial. Each balloon inflated for 3 seconds and then either popped or stayed intact depending on whether the participant’s wager was above or below the predetermined explosion point for that balloon. If the balloon was pumped past its explosion point, it would pop and the participant earned no points for that balloon (Figure 2). If the balloon was not pumped past the explosion point, the subject got to keep the number of pumps as points. After each outcome, a new deflated balloon appeared on the screen and points earned were added to the total. Each balloon could earn a maximum of 128 points with an explosion point equally likely to occur on any given pump subject to the constraint that within each sequence of 10 balloons the average explosion point was on pump 64. In order to limit extraneous variability, balloons were presented in the same order for all participants. For this study we relied on mean numbers of pumps across balloons as the main measure for the BART. We used this measure in our regression models described below. We also evaluated mean number of pumps for the first ten balloons and the second ten balloons. Analysis We were interested in predicting behaviors that influence the risk of crash injury with substance use behavior and risk-taking on the BART. We
80
Figure 2 - BART balloon explosion
calculated a composite risk “score” for each of these domains using a selection of questions from each category (crash injury risk or substance use). Questions used for the crash injury risk score were related to safety restraint use while driving or as a passenger, attitudes towards safety restraint use and safety restraint laws, driving without a license, driving or riding with someone under the influence of alcohol, being stopped by police while driving, or talking on the phone while driving. Questions used for the substance use score included tobacco use and frequency of use, alcohol use and frequency of use, involvement in binge drinking, marijuana use and frequency of use, and inhalant use. “Risky” responses in each of these domains were encoded with a “1”, and response values were then summed to create the total scores, with crash injury risk having a maximum score of 10 and substance risk a maximum of 12. We then employed hierarchical regression analysis to predict driving risk behavior, first entering sex, age, sample source and substance use risk behavior to evaluate the predictive power of the BART. This study was reviewed and approved by the Yale University Human Investigations Committee. RESULTS We enrolled 100 adolescents (mean age 15.9) in total, 58 of which were from the ED (29 males) and 42 from the AC (20 males). 31% of our sample self-identified as Hispanic/Latino, 26% as White, and 41% as Black/African American. All 100 adolescents completed the survey and the BART, but our first participant was excluded from the analyses because after his participation we added additional survey questions to our measures that impacted risk scores.
©
Annals of Advances in Automotive Medicine
Table 1. Frequency, Means, Standard Deviations and Correlations Among Main Study Variables Variable
Sex (M/F) 51/49
Sample Source (ED/AC) 58/42
M (SD) Range Sex
Age
Crash Injury Risk Behavior
Substance Use Risk Behavior
BART 1-10
BART 11-20
BART Total
15.94 (1.44)
2.76 (1.98)
2.99 (3.35)
45.90 (18.95)
41.91 (18.39)
43.91 (17.62)
-
Sample Source
.02
-
Age
.14
.06
-
Crash Injury Risk Behavior
.04
.18
.05
-
Substance Use Risk Behavior
.26*
.13
.35**
.37**
-
BART 1-10
-.02
-.08
-.04
.27**
.03
-
BART 11-20
-.13
-.13
-.14
.19
-.02
.78**
-
BART total
-.08
-.11
-.09
.24*
.01
.95**
.94**
-
*p
Annals of Advances in Automotive Medicine
ADOLESCENT BALLOON ANALOG RISK TASK AND BEHAVIORS THAT INFLUENCE RISK OF MOTOR VEHICLE CRASH INJURY 1,2
Federico E. Vaca, MD, MPH; 2Jessica M. Walthall, BS; 3Sheryl Ryan, MD; 4 Alison Moriarty-Daley, APRN; 5Antonio Riera, MD; 2Michael J. Crowley, PhD*; 2 Linda C. Mayes, MD* 1
Yale School of Medicine, Department of Emergency Medicine, New Haven, Connecticut 2 Yale School of Medicine, Child Study Center, New Haven, Connecticut 3 Yale School of Medicine, Department of Pediatrics, New Haven, Connecticut 4 Yale School of Nursing, New Haven, Connecticut 5 Yale School of Medicine, Pediatric Emergency Medicine, New Haven, Connecticut * shared senior authorship __________________________________
ABSTRACT – Risk-taking propensity is a pivotal facet of motor vehicle crash involvement and subsequent traumatic injury in adolescents. Clinical encounters are important opportunities to identify teens with high risk-taking propensity who may later experience serious injury. Our objective was to compare self-reports of health risk behavior with performance on the Balloon Analog Risk Task (BART), a validated metric of risk-taking propensity, in adolescents during a clinical encounter. 100 adolescent patients from a hospital emergency department and adolescent health clinic completed a computer-based survey of self-reported risk behaviors including substance use behaviors and behaviors that influence crash involvement. They then completed the BART, a validated laboratorybased risk task in which participants earn points by pumping up a computer-generated balloon with greater pumps leading to increased chance of balloon explosion. 20 trials were undertaken. Mean number of pumps on the BART showed a correlation of .243 (p=.015) with self-reported driver/passenger behaviors and attitudes towards driving that influence risk of crash injury. Regression analyses showed that self-reports of substance use and mean number of pumps on the BART uniquely predict self-reports of behaviors influencing the risk of crash injury. The BART is a promising correlate of real-world risk-taking behavior related to traffic safety. It remains a valid predictor of behaviors influencing risk of crash injury when using just 10 trials, suggesting its utility as a quick and effective screening measure for use in busy clinical environments. This tool may be an important link to prevention interventions for those most at-risk for future motor vehicle crash involvement and injury. __________________________________
INTRODUCTION Adolescence has been described as a dual-paradox life stage in which young people encounter rapid increases in physical strength and capacity for decision-making yet they remain highly vulnerable to morbidity and mortality related to increased risktaking and novelty seeking behavior [Arnett, 1992; Dahl, 2004; DiClemente et al., 1996]. In the United States in 2010 alone, 42% of deaths among teens and young adults age 15-24 years were due to unintentional injuries; 57% of were a result of motor vehicle crashes [Centers for Disease Control and Prevention, 2012]. Young drivers made up only 6.4% of licensed drivers in the US, yet in 2010 they accounted for 10% of fatal motor vehicle crashes and 14% of the more than 5.4 million police-reported crashes [NHTSA, 2012]. In the same year, almost 282,000 teens were sent to emergency departments (ED) for treatment of crash-related injuries [Centers for Disease Control and Prevention, 2012].
National injury statistics continue to show that adolescents are a highly vulnerable age group for injury-related disability and death. Much of this vulnerability has been popularly attributed to the propensity of adolescents to take risks with limited appraisal of the context or due to psychosocial effects on self-regulation abilities [Institute of Medicine (U.S.) et al., 2011; Steinberg, 2004]. This is particularly true when it comes to teens operating motor vehicles or when riding as an occupant with a teen driver. While Graduated Driver Licensing (GDL) laws, enforcement, and national education initiatives have helped lower the fatalities involving teen drivers in the past decade [Williams, 2012], for some teens, this downward trend appears to have ended in 2011. A recent report showed that 2011 driver deaths among 16- and 17-year-olds increased for the first time since 2002 [Williams, 2012]. Further, more recent preliminary national crash data
CORRESPONDING AUTHOR: Jessica Walthall, Child Study Center, 230 S. Frontage Road, NIHB G-10, New Haven, CT Email: [email protected]
57th AAAM Annual Conference Annals of Advances in Automotive Medicine September 22-25, 2013
77
Vol 57 • September 2013 results from the first six months of 2012 show that the number of driver deaths in this same age group has increased, up 19% from 2011 [Williams, 2012]. The worsening of national fatal crash statistics for these young drivers should serve as a call for renewed and refocused efforts in crash injury prevention. The pursuit to understand the developmental, behavioral, and contextual factors that contribute to teen crashes should be even more urgent. Developing new and effective prevention methods with a focus on the dual paradox of the adolescent developmental period is important and may help in the identification and attenuation of risktaking propensity in some adolescents. Risk-taking behavior has been most often studied with self-report instruments [DiClemente et al., 1996; Gullone and Moore, 2000; Jessor and Jessor, 1977]. These approaches, though useful, may yield incomplete results when used as the sole assessment method, especially among adolescent populations. Self-report measures are subject to response biases such as social desirability and require a willingness to provide truthful responses, as well as an insight and ability to provide an accurate self-report that adolescents may lack (e.g., [Bernstein et al., 2002; Hyman and Loftus, 1998; Ladouceur et al., 2000; Tversky and Koehler, 1994]. Relying exclusively on a unimethod assessment approach has its own limitations and may not be sufficient to evaluate the multidimensional nature of risk-taking [Eysenck and Eysenck, 1977; Wills et al., 2002]. Therefore, tools that can further explore these multiple facets of risk taking may provide more accurate assessments or predictions of risk-taking propensity. Lejuez and colleagues approached these methodological limitations by developing a computer-based behavioral risk task, the Balloon Analog Risk Task (BART) [Lejuez et al., 2002], in which the participant makes button presses (pumps) to inflate individual balloons for incremented points or money with increasing risk that the balloon will explode. This would lead to the loss of the balloon’s present point or monetary value. In this context, each button depression or pump is a “risky decision” with a finite probability that the balloon will explode, and thus the chance that no money will be earned increases with each pump. Empirical research has shown that the number of pumps on the BART is related to self-reports of real-world risk-taking behavior in both adults and adolescents. In adolescents, BART performance has been shown to be a valid indicator of real-life risk behavior such as substance use (cigarette, alcohol, and illicit drug),
78
gambling, unprotected sexual intercourse, stealing, carrying a weapon, fighting, and helmet and safety restraint misuse [Aklin et al., 2005; Lejuez et al., 2007; Lejuez et al., 2003]. Moreover, the BART has been further validated within clinical populations, effectively distinguishing between typicallydeveloping adolescents and adolescents with conduct and substance use disorder symptoms [Crowley et al., 2006] and differentiating 3,4-methylenedioxy-Nmethylamphetamine (MDMA) users from non-users [Hopko et al., 2006]. Finally, several recent neuroimaging and physiological testing studies have successfully used the BART to identify the neural correlates of risky decision-making [Crowley et al., 2009; Fukunaga et al., 2012; Rao et al., 2008]. To our knowledge, the BART’s relationship with behaviors that could influence the risk of crash injury has not been explored. We believe that this relationship is a worthwhile investigation, particularly because risky driving is generally not an isolated behavior and risk-taking in the BART has been shown to have an association with other lifestyle behaviors that co-vary with risk factors for crashes. Adolescents who smoke cigarettes have a higher likelihood of crash involvement [Hutchens et al., 2008; Lang et al., 1996], and tobacco and marijuana users have a higher instance of traffic violations [Bina et al., 2006]. Teens who misuse alcohol are more likely to drink and drive, be involved in single-car crashes, and ride with impaired drivers [Copeland et al., 1996; Lang et al., 1996; Van Beurden et al., 2005]. Because risky performance on the BART has been shown to relate to these cooccurring behaviors, the BART could be used to identify adolescents who exhibit other behaviors that could influence their risk of crash injury. It could further be used as an interventional component for preventing future teen injury and fatal motor vehicle crash involvement. We believe that the BART could serve as a viable screening tool in an environment in which prevention screening routinely occurs for adolescents. For this reason, we chose to investigate the BART’s utility in the clinical setting. Clinician-patient interactions within the emergency department and primary care setting can become notable “teachable moments” if behaviors related to risk-taking are successfully identified [Lawson and Flocke, 2009]. Much like other screening activities that routinely take place in the clinical setting, the BART may be a useful tool for healthcare providers who have a captive audience of teens during a clinical encounter. In this setting, rapid assessment of risk-taking propensity could take place with the opportunity to intervene.
©
Screenings of adolescent patients within the ED setting have been successful in detecting teenagers who are more likely to engage in specific risky behaviors, including those that contribute to greater risk of pregnancy [Chernick et al., 2012] and injury [Johnston et al., 2002]. Computer-based screenings within primary care settings have also been shown to be effective in identifying adolescents who are more likely to engage in risky behaviors [Anand et al., 2012; Paperny et al., 1990]. Further research has shown that technology-based evaluations and interventions reduce time and personnel burden, are effective, acceptable to patients, and often preferred by adolescent patients [Chisolm et al., 2008; Cunningham et al., 2012; Fein et al., 2010; Harris et al., 2012; Olson et al., 2009; Ranney et al., 2012; Vaca et al., 2010; Walton et al., 2010] Given the very recent unfavorable upward trend in teen crash fatalities, we believe there is a need for technology-based rapid health screening tools that can effectively identify risk-taking propensity in clinical environments. Adolescent patient contact episodes in these settings are important opportunities for implementing adolescent-focused prevention efforts. The goal of our study was to integrate the BART, a validated measure of risk-taking propensity, into two different clinical settings and explore its relationship to well-known behaviors that influence the risk of crash injury. We hypothesized that the BART would account for unique variance in predicting driving risk behavior after accounting for substance use risk behaviors. METHODS We conducted a prospective observational study enrolling adolescent patients from a large urban university tertiary care hospital pediatric ED with approximately 35,000 annual visits as well as a primary care adolescent clinic (AC), part of a pediatric primary care center with nearly 26,000 annual visits. Both centers serve a demographically diverse population reflective of the overall city population which is 27% Latino, 35.4% Black, and 42.6% White. We enrolled adolescent patients aged 14-18 years from the pediatric ED and AC from March 2011 to May 2012. ED patients were primarily identified using the electronic patient tracking system. Adolescents who were categorized at higher triage levels, did not speak English, were unable to provide assent/consent or parent/legal guardian consent, in the custody of the state, under suspicion of abuse or
Annals of Advances in Automotive Medicine
threat to self, or receiving a psychiatric evaluation were excluded from enrollment. Adolescent patients were approached by the trained research associate to discuss the study. If they agreed to participate in the study, they were consented (minors - assent with parental consent, those 18 years of age - consent). Consented participants then completed a computer-based self-report health risk assessment survey and the BART. Upon completion of the study tasks, participants were provided a popular store gift card for their time and effort. Instruments Computer based survey: History of risk-taking behaviors was assessed using a selection of questions from the 2011 Youth Risk Behavior Survey (YRBS) [Centers for Disease Control and Prevention, 2011] which has demonstrated acceptable reliability and validity in the past [Brener et al., 2002]. An inventory of 50 questions pertaining to safety, violence, sexual behavior, and tobacco, alcohol and drug use were chosen. For the purposes of this study, given the association between substance use risk and driving risk, we relied on questions reflecting substance use risk. These items served as our substance use risk behavior index. Additional questions from the 2007 Motor Vehicle Occupant Safety Survey pertaining to traffic safety and attitudes towards traffic safety (safety restraint use, traffic safety laws, speeding, accidents) were included in order to further explore behaviors that could influence the adolescent’s risk of crash injury [NHTSA, 2007]. These items served as our crash injury risk behavior index. All questions were administered using E-Prime software (Psychology Software Tools, Inc.) and delivered as a laptop-based self-administered electronic survey. Balloon Analogue Risk Task (BART): A modified automatic version of the BART [Pleskac et al., 2008] using 20 balloon trials was administered to assess risk-taking propensity. Our version of the BART-AP mirrored Lejuez’s BART-AP but rewarded participants with points instead of money, as previously deemed appropriate for adolescent populations [Lejuez et al., 2007]. Balloon pumps produced points represented visually in a meter, as designed for Lejeuz’s youth-oriented version of the BART (BART-Y) [Lejuez et al., 2007]. Upon initiation of the task, a deflated balloon appeared on the computer screen accompanied by a dial of numbers from 0 to 9, a “clear” button to reset the input number of pumps, and three boxes indicating the total number of points at stake, the total number
79
Vol 57 • September 2013
Figure 1 - BART initial screen
of points earned, and the explosion point for the previously presented balloon (Figure 1). Participants were informed that they would be presented with 20 different balloons (20 trials), and that “the actual number of pumps for any particular balloon will vary.” Participants were instructed to attempt to earn as many points as possible. At the beginning of each trial, the participant decided how many pumps they thought the balloon would hold and input this number using the mouse and the on-screen dial. Each balloon inflated for 3 seconds and then either popped or stayed intact depending on whether the participant’s wager was above or below the predetermined explosion point for that balloon. If the balloon was pumped past its explosion point, it would pop and the participant earned no points for that balloon (Figure 2). If the balloon was not pumped past the explosion point, the subject got to keep the number of pumps as points. After each outcome, a new deflated balloon appeared on the screen and points earned were added to the total. Each balloon could earn a maximum of 128 points with an explosion point equally likely to occur on any given pump subject to the constraint that within each sequence of 10 balloons the average explosion point was on pump 64. In order to limit extraneous variability, balloons were presented in the same order for all participants. For this study we relied on mean numbers of pumps across balloons as the main measure for the BART. We used this measure in our regression models described below. We also evaluated mean number of pumps for the first ten balloons and the second ten balloons. Analysis We were interested in predicting behaviors that influence the risk of crash injury with substance use behavior and risk-taking on the BART. We
80
Figure 2 - BART balloon explosion
calculated a composite risk “score” for each of these domains using a selection of questions from each category (crash injury risk or substance use). Questions used for the crash injury risk score were related to safety restraint use while driving or as a passenger, attitudes towards safety restraint use and safety restraint laws, driving without a license, driving or riding with someone under the influence of alcohol, being stopped by police while driving, or talking on the phone while driving. Questions used for the substance use score included tobacco use and frequency of use, alcohol use and frequency of use, involvement in binge drinking, marijuana use and frequency of use, and inhalant use. “Risky” responses in each of these domains were encoded with a “1”, and response values were then summed to create the total scores, with crash injury risk having a maximum score of 10 and substance risk a maximum of 12. We then employed hierarchical regression analysis to predict driving risk behavior, first entering sex, age, sample source and substance use risk behavior to evaluate the predictive power of the BART. This study was reviewed and approved by the Yale University Human Investigations Committee. RESULTS We enrolled 100 adolescents (mean age 15.9) in total, 58 of which were from the ED (29 males) and 42 from the AC (20 males). 31% of our sample self-identified as Hispanic/Latino, 26% as White, and 41% as Black/African American. All 100 adolescents completed the survey and the BART, but our first participant was excluded from the analyses because after his participation we added additional survey questions to our measures that impacted risk scores.
©
Annals of Advances in Automotive Medicine
Table 1. Frequency, Means, Standard Deviations and Correlations Among Main Study Variables Variable
Sex (M/F) 51/49
Sample Source (ED/AC) 58/42
M (SD) Range Sex
Age
Crash Injury Risk Behavior
Substance Use Risk Behavior
BART 1-10
BART 11-20
BART Total
15.94 (1.44)
2.76 (1.98)
2.99 (3.35)
45.90 (18.95)
41.91 (18.39)
43.91 (17.62)
-
Sample Source
.02
-
Age
.14
.06
-
Crash Injury Risk Behavior
.04
.18
.05
-
Substance Use Risk Behavior
.26*
.13
.35**
.37**
-
BART 1-10
-.02
-.08
-.04
.27**
.03
-
BART 11-20
-.13
-.13
-.14
.19
-.02
.78**
-
BART total
-.08
-.11
-.09
.24*
.01
.95**
.94**
-
*p
