Retained Risk-Taking Behaviors among Past Alcohol Dependent Trauma Patients

Abstract

Alcohol dependence has been associated with behavioral risk factors and risk-taking tendencies. We investigated whether past alcohol dependent trauma center patients (n=42) retain the characteristics of current alcohol dependent patients (n=67) or whether they resemble patients without history of alcohol dependence (n=262). We found that past alcohol dependence patients retain some of the risk-taking tendencies (impulsivity and sensation seeking) and risk-taking behaviors (drinking and driving, riding with a drunk driver, binge drinking, speeding for the thrill) common to current alcohol dependent patients and they remain at a higher injury risk than the non alcohol dependent population.

As shown by a significant prevalence of negative toxicology findings among acutely injured patients with alcohol disorders and the over-representation of alcohol disorders among the non intoxicated trauma patients (Rivara Arch Surg 1993, Soderstrom JAMA 1997, Maio 1997, Ryb 2002) there may be other risk factors for injury among these patients that are independent of alcohol intoxication.

The medical literature has documented an increase in both health related risk factors (Fillmore 1998, Leino 1998, Green 2001) and mortality (Mertens 1996, Dawson 2001) among ex-drinkers. However, no definitive association has been shown between past alcohol dependence and injury related risk factors among trauma patients.

The purpose of this study is to establish whether patients injured in a motor vehicle collision (MVC) who have a past history of alcohol dependence (PAD) retain injury related behavioral risk patterns of current alcohol dependent patients (CAD) or whether they behave more like patients with no alcohol disorder (NAD) diagnosis.

METHODS

STUDY SITE/POPULATION

The study was conducted at the R Adams Cowley Shock Trauma Center of the University of Maryland Medical Center in Baltimore. The center is a regional adult Level I trauma center that serves the most-populated counties of central Maryland. The center also serves the urban communities surrounding the medical center. Approximately 85% of patients treated at the trauma center are admitted from the scene of injury. Those injured in rural/suburban settings are usually transported by Medevac helicopters, and those injured in the city are transported by ambulance. In terms of mechanism of injury, age, and sex, our patient profile is similar to the aggregate of patients treated in trauma centers throughout the United States (Miller 1998). For patients admitted from the injury scene, time from injury to admission averages about 1 hour. For this study we included all MVC patients from the total 1118 Trauma Center patients interviewed. The population interviewed is representative of the entire trauma population at our trauma center.

ELIGIBILITY CRITERIA

Patients were eligible for recruitment if they were 18 years of age or older, were admitted from the scene of injury, had intact cognition, and had a length of stay 2 or more days. A length of stay of 2 or more days was chosen to identify patients with serious injuries. Patients were not eligible for interview while in intensive care units. Patients initially in intensive care units or who were cognitively impaired were followed until they became eligible or were discharged. Finally, a patient was not eligible for study if his or her attending surgeon thought that a patient interview would have a negative impact on the clinical course. The study design was approved by both the Institutional Review Board of the University of Maryland School of Medicine and the center’s research committee.

DATA COLLECTION

Demographic data, injury type (unintentional [vehicular crashes, falls, etc.] or intentional [shootings, stabbings, etc.]), injury severity score and results of the admission BAC tests were obtained from the center’s toxicology database (Soderstrom JT 1997[41]). The remaining data were obtained through the interview (see below).

ALCOHOL USE DIAGNOSES

Alcohol disorder diagnoses (CAD, alcohol abuse, PAD or NAD)) were made by using the Psychoactive Substance Use Disorders section of the Structured Clinical Interview for the DSM-III-R (SCID)(Spitzer 1987, APA 1987). The SCID is a widely accepted instrument that provides in-depth alcohol and other drug use diagnoses according to standardized criteria (Kitchens 1994, NIAAA 1991). SCID assessments allow categorization of diagnoses as either current or lifetime. Alcohol-dependent individuals who have not been in full or partial remission for at least 6 months before injury are classified as currently dependent (Soderstrom JAMA 1997). Individuals with lifetime but not current dependence where classified as PAD. Alcohol abuse diagnosis was given to individuals without alcohol dependence but with continued use of alcohol despite knowledge of persistent or recurrent occupational, psychological or physical problems related to alcohol use, or recurrent hazardous drinking (i.e. driving while intoxicated).

PATIENT INTERVIEWS

Eligible subjects were approached for study consent by the interviewers, who were trained in administration of the SCID instrument and three interview screening tests for alcoholism. Patients were considered cognitively competent if they had good memory of recent and remote events. The interviewers had no knowledge of admission BAC and other drug test results. The results of SCID assessments for alcohol and other drug diagnoses and the accuracy of the alcoholism screening tests were published previously (Soderstrom JAMA 1997, Soderstrom JT 1997[42]). Injury history, socioeconomic status, demographics, injury prone behaviors and risk-taking dispositions were assessed during the interview.

Injury prone behaviors were explored with questions evaluating the frequency or likelihood of the patient engaging in certain injury prone behaviors (IPB) (seatbelt use, drinking and driving, riding with a drunk driver, binge drinking, speeding for the thrill). Similar questions have been used by several authors and by the Behavioral Risk Factor Surveillance System (Cherpitel 1993, Cherpitel 1999, Soderstrom 2001, Field 2004, Hunt 2001). Risk-taking dispositions (impulsivity [IMP], risk perception [RP] and sensation seeking [SS]) were evaluated using questions utilized in the national alcohol survey in 1990. RP evaluation included 6 questions with answers graded from 1 (very unlikely) to 5 (very likely). IMP and SS evaluation included 5 and 4 questions respectively with answers graded from 1 (not at all) to 4 (quite a lot). Actual format of questions is found elsewhere (Cherpitel 1993, Cherpitel 1999, Soderstrom 2001, Field 2004).

ANALYSIS OF RESULTS

Comparisons were performed between 3 diagnostic groups (NAD, PAD, and CAD)for injury related behavioral risk factors (seatbelt use, drinking and driving, riding with a drunk driver, binge drinking, speeding for the thrill) and measures of risk perception, impulsivity and sensation seeking. Alcohol abuse patients (n=22) were excluded because the small number would have precluded statistical analysis. “Low seatbelt use” was defined as less often than “nearly always”. “Drinking and driving” and “riding with a drunk driver” were defined as the self-reported occurrence of the event during the previous 30 days. “Speeding for the thrill” was considered positive when individuals reported the behavior more often than rarely. Other substance abuse diagnoses (current drug dependence), demographic (age, gender, ethnicity and marital status) and socioeconomic factors (education, income and unemployment) that could influence outcome variables were also explored.

Overall (i.e., global) differences between diagnostic groups expressed as proportions were made using the Mantel-Haenszel chi-square statistic. Comparisons between mean values of the risk-taking disposition scales across diagnostic groups were assessed by analysis of variance. To adjust for possible confounding, multivariate logistic regression models were constructed using stepwise selection methods, with each injury related behavioral risk factor as the outcome. Independent variables included the 3 diagnostic groups, other substance abuse diagnosis, and demographic and socioeconomic factors. For this exploratory analysis, a p-value <0.05 was considered suggestive of statistical significance.

RESULTS

Following exclusion of the alcohol abuse group, a total of 375 patients were involved in a MVC. The study group was composed of 262 NAD, 46 PAD and 67 CAD patients. Demographic and socioeconomic characteristics of the population are listed in table 1. While ethnicity, income and unemployment were not significantly different between groups, PAD patients were in an intermediate position in relation to age and male gender prevalence.

Table 1.

Demographic and socioeconomic characteristics (%)

Risk factor	NAD (N=262)	PAD (N=46)	CAD (N=67)	Global p
Age ≥ 45	40	28	18	0.003
Male gender	43	63	75	0.001
Black	22	15	19	0.29
Income < $15,000	37	41	48	0.12
Education < 12 years	14	28	29	0.002
Unemployment	10	11	18	0.08
Not Married	57	61	75	0.012
Uninsured	16	41	46	<0.001

PAD patients experienced an intermediate risk between NAD and CAD in relation to binge drinking, drinking and driving, low use of seatbelt, and smoking (Table 2).

Table 2.

Injury related behavioral risk factors (%).

Risk factor	NAD (N=262)	PAD (N=46)	CAD (N=67)	Global p
Binge drinking	23	41	94	<0.001
Binge drinking (monthly or more)	4	20	72	<0.001
Drink and drive	4	17	68	<0.001
Rides with drunk	10	13	58	<0.001
Seatbelt low-use	48	61	94	<0.001
Speeds for thrill	5	15	18	<0.001

Similarly, PAD was in an intermediate position in relation to smoking and presence of acute alcohol intoxication (Table 3), most of other risk markers and injury history (Table 4), and measures of risk perception, impulsivity and sensation seeking (Table 5). PAD resembled CAD in relation to education, health insurance and revoked license, and resembled NAD in relation to employment, marital status and drug dependence rates (Tables 1, 2, 3 and 4).

Table 3.

Alcohol intoxication and substance dependence (%).

Risk factor	NAD (N=262)	PAD (N=46)	CAD (N=67)	Global p
Smoking	28	54	67	<0.001
Drug dependent	4	4	21	<0.001
BAC+	10	28	71	<0.001
BAC>80	6	22	64	<0.001

Table 4.

Other risk markers and injury history (%).

Risk factor	NAD (N=262)	PAD (N=46)	CAD (N=67)	Global p
< average school performance	3	11	22	<0.001
School suspension	22	39	55	<0.001
MVA history	22	50	61	<0.001
Assault history	9	30	36	<0.001
Other injury hx	35	52	58	<0.002
Any injury hx	54	80	90	<0.001
Moving Traffic Violation	31	41	54	<0.024
Single vehicle accident	28	35	61	<0.032
Revoked license	2	7	7	<0.014

Table 5.

Risk-taking dispositions (mean values).

Risk factor	NAD (N=262)	PAD (N=46)	CAD (N=67)	Global p
Impulsivity (1–4)	1.57	1.97	2.24	<0.007
Risk perception (1–5)	4.21	3.97	3.89	<0.001
Sensation Seeking (1–4)	1.94	2.35	2.44	<0.001

When comparing PAD individually with NAD and CAD with regard to risk-taking dispositions, PAD was only significantly different (p<0.05) from NAD for impulsivity and sensation seeking. There was no pairwise difference between PAD and CAD.

Using NAD as the reference group, multivariate logistic regression models were constructed to discern the association of alcohol diagnosis with outcome following adjustment for possible confounders. A PAD diagnosis had a positive significant association with drinking and driving, riding with a drunk driver, binge drinking, binge drinking at least monthly, and driving fast for the thrill (Table 6).

Table 6.

Results of multivariate stepwise logistic regression models: Odds ratios for injury prone behaviors (ns= p>0.05).

Risk factor	Low use seat-belt	Drinks and Drive	Rides with Drunk	Binges	Binges Monthly or more	Drives fast for thrill
CAD	3.67	90.3	13.5	58.9	66.5	3.1
PAD	ns	6.62	3.32	2.15	6.2	3.7
Married	0.53	ns	Ns	ns	0.39
Drug dependence	ns	ns	Ns	ns	ns	3.6
Income <$15,000	ns	0.34	Ns	ns	ns	ns
Education <12 years	ns	0.26	Ns	ns	ns	ns
Male gender	ns	ns	Ns	2.19	ns	ns
Age	ns	0.96	0.94	0.94	0.97	0.97

DISCUSSION

Even though trauma patients with PAD have a lower behavioral risk profile than CAD patients, they retain a higher behavioral risk profile than the NAD trauma population. Our data shows a persistence of risk-taking behaviors among past alcohol dependent patients. This fact could be explained by SES/environmental factors, or by pre-existing risk-taking traits that did not disappear after alcohol use cessation as shown by a persistently high score in risk-taking disposition measures (i.e. high impulsivity and high sensation seeking).

PAD patients were not statistically different from CAD in relation to risk-taking dispositions. The influence of past dependence even after adjusting for SES and demographic variables points toward risk-taking dispositions being the more likely cause of the persistence of these behaviors.

This is illustrated also by the different influence of PAD in relation to risk-taking behaviors. Seat belt use, more likely related to risk perception, was not related to PAD in the logistic regression model. Speeding for the thrill, being more likely a reflection of sensation seeking, occurred almost four times as much among PAD than among the reference NAD population. This was higher than for the CAD group. The remaining risk-taking behaviors, more likely related to impulsivity showed strong influence (2–6 times as much) of PAD but a much higher effect (13–90 times as much) for CAD.

Over the years, several authors have expressed their concern with regard to the high levels of trauma recidivism among trauma patients (Lowenstein 1986, Kaufman 1998, Rivara JAMA 1993). This recidivism was felt to be strongly related to the presence of substance abuse problems among trauma patients. Dischinger et al (2001) showed an increased injury mortality after discharge on trauma patients with positive toxicology (Dischinger 2001).

Adult former drinkers, when compared with long term abstainers, have been found to have increased rates of heavy smoking, depression, unemployment and low SES (Fillmore 1998). With regard to health behaviors, former drinkers have been shown to be consistently less healthy (mentally, physically, and functionally) than light and moderate drinkers (Green 2001), however, IPB were not included in this study.

An increased mortality from external causes (i.e. injury) was suggested among former drinkers after adjustment for age, gender, marital status, education, smoking, and poor baseline health (Dawson 2001). Contrary to these findings, a study of a population (n=128,934) in North Carolina found no increased mortality from external causes (or any of its subsets) among ex-drinkers (Klatsky 1993). Similarly, Bullock documented a return to baseline mortality among ex-alcohol dependent veterans who achieved long term abstinence (Bullock 1992).

Over the past 15 years, several authors have proposed that certain behavioral traits and personality characteristics are linked to the development of alcohol disorders (Cloninger 1998, Kruegger 2002, Young 2000, Finn 2000, Sher 1991, Kendler 2003, Chassin 2004). These include antisocial behavior, social deviance proneness, novelty seeking, sensation seeking, excitement/pleasure seeking, thrill/adventure seeking, impulsivity, emotionality, extroversion, disinhibition, lack of constrain, behavioral under-control, psychopathology and externalizing disorders.

Dawson, while analyzing the risk of occupational injury (Dawson 1994), found that some of the excess risk of injury among heavy drinkers was attributable to a risk-taking propensity independent of the direct effect of alcohol (using current smoking as a proxy for risk-taking).

These theoretical constructs (i.e. behavioral undercontrol, novelty seeking, etc.), in interaction with social expectancy, seem to mediate the development of alcohol disorders (Finn 2000). Twin studies (Kruegger 2002, Young 2000, Kendler 2003) and case control studies (Finn 2000, Sher 1991, Chassin 2004) suggest a significant genetic contribution to the development of alcohol disorders during adolescence.

In theory, some of the behavioral traits that put individuals at risk of developing substance abuse disorders could make these individuals more injury prone even in the absence of or during remission from their alcohol disorder. The trauma literature documents a high level of pre-existing psychopathology among trauma center patients (Poole 1997, Whetsell 1989).

A biological substrate with significant hereditability (Jonah 1997, Zuckerman 1980) has been postulated for some of these traits and has been directly implicated in risky driving (Jonah 1997 and 2001). Developmental changes (Rosembloom 2002) and hormonal correlates (Rosenblitt 2001) have been implicated. Serotoninergic neurotransmission dysfunction has been implicated in decreased harm avoidance, extraversion and venturesomeness among abstinent alcohol-dependent men (Weijers 2001).

A particular neuronal network activation has been identified among patients with antisocial disorders using functional MRI. (Vollm 2004). In normal individuals, behavioral disinhibition (impulsivity) was negatively correlated with activation of the right dorsolateral prefrontal cortex in a functional MRI study (Asahi 2004). Another functional MRI study found that more impulsive individuals activated paralimbic areas, while less impulsive individuals activated higher order association areas when inhibiting a pre-potent response (Horn 2003).

These pre-existing neurobehavioral patterns typical of individuals predisposed to develop alcohol dependence (i.e. impulsivity, novelty seeking, etc.) could be further amplified by acute alcohol intake or by neurobehavioral changes caused by chronic dependence causing even a much higher propensity to be exposed to injuries as shown by the much higher odds ratios for some risk-taking behaviors for CAD in the logistic regression models.

Interestingly, PAD was closer to CAD in relation to constitutional and historical variables (i.e. gender, education level, school performance, education, and injury history) and closer to NAD in current variables (employment, income and marital status). A high rate of uninsured status however remained present among PAD patients. This could be explained also by a risk-taking tendency, or be a reflection of the type of employment.

Despite the general pessimism about screening and treating substance abuse (Danielsson 1999), several authors have documented successful use of brief interventions (Bien 1993) in reducing drinking among primary care (Whitlock 2004) and ER patients (Longabaugh 2002, Monti 1999) with alcohol problems. Further studies have applied similar methodology in the trauma population (Dunn 1997, Apodaca 2003) and also showed improvement in hazardous drinking (Sommers 2001, Gentilelo 1999). The experience in Seattle (Gentillelo 1999) has shown a significant decrease in alcohol use sustained through 12 months among trauma center patients receiving brief interventions. This difference in alcohol intake did not, however, translate into a statistically significant decrease in trauma recidivism.

To our knowledge, this is the first study documenting increased behavioral risks among PAD MVC trauma center patients. The implications of our data are that treatment of CAD, if effective, would not necessarily bring risk-taking behaviors among this population of trauma patients back to baseline levels. Interventions addressing these behaviors (independently of alcohol disorder treatment), could further increase the effectiveness of “brief interventions” in the population with current alcohol problems, and prevent recidivism within the PAD group.