Anxiety Psychopathology and Alcohol Use among Adolescents: A Critical Review of the Empirical Literature and Recommendations for Future Research

Abstract

Adolescent alcohol use is a critical public health concern; accordingly, a considerable body of work exists identifying developmentally salient risk and protective factors. One area receiving increasing attention among adults is the linkage between specific constellations of anxiety psychopathology and alcohol use problems. Relatively less is known about such linkages among adolescents, despite the onset of both anxiety-type problems and alcohol use during this developmental period. The current review presents a detailed summary and analysis of the empirical literature focused on specific forms of anxiety psychopathology as they relate to alcohol use among adolescents, and provides a number of specific recommendations for future work with an emphasis on the utility of experimental psychopathology techniques for clarifying basic questions and forwarding this body of work.

Introduction

Co-morbidity across clinical disorders is common (Kessler, Chiu, Demler, & Walters, 2005; Mineka, Watson, & Clark, 1998), and the co-occurrence of anxiety and problematic alcohol use (e.g., excessive intake, abuse, coping-related use) is receiving increasing recognition as researchers begin to understand how each condition exacerbates the other (e.g., Merikangas et al., 1998). One area that has received empirical attention is the association between alcohol use behaviors and specific constellations of anxiety psychopathology (e.g., generalized anxiety, social anxiety; Kushner, Sher, & Beitman, 1990), although relatively little work has been conducted among adolescents. Three primary objectives guide the current manuscript. The first is to outline developmental considerations related to the association between specific clusters of anxiety symptomatology and alcohol use behaviors among adolescents. The second goal is to comprehensively review the relevant extant literature. Finally, drawing from these sections, a number of research suggestions are offered with a particular emphasis on the utility of experimental psychopathology techniques for clarifying basic questions and forwarding this body of work.

Adolescent Alcohol Use and Anxiety: Background and Developmental Considerations

Alcohol use among adolescents is a critical public health concern, with linkages to an array of negative consequences including poor academic performance, risky sexual behavior, increased delinquency (e.g., vandalism; Hingson & Kenkel, 2004), other substance use (e.g., marijuana use; Yamaguchi, & Kandel, 1984a), as well as an increased risk of unintentional (e.g., traffic accidents; National Highway Traffic Safety Administration, 2002) and intentional (e.g., suicide; Levy, Miller, & Lox, 1999) death. Despite concerted efforts aimed at reducing adolescent alcohol use (Stewart, Conrad, Marlatt, Comeau, Thrush, & Krank, 2005), initiation and experimentation with alcohol is relatively normative during this developmental phase. For example, large scale cross-sectional (Johnston, O'Malley, Bachman, & Schulenberg, 2008) and prospective work (Kandel & Logan, 1984; Wittchen et al., 2008) suggests that by age 18 years, 75–90% of adolescents have experimented with alcohol, and a significant proportion of adolescents report alcohol use initiation prior to age 14 years (38.5%; Wittchen et al., 2008). Indeed, alcohol is the most commonly used substance among adolescents (cf. nicotine, marijuana), and the initiation of alcohol use most often occurs prior to the initiation of other substances (e.g., Johnston et al., 2008; Kandel & Logan, 1984). Importantly, although early initiation of alcohol use is linked to an elevated risk for the development of alcohol use problems later in life (e.g., alcohol abuse; Nelson & Wittchen, 1998), there is still considerable variation in subsequent patterns and trajectories of use, and the majority of adolescents, including those who begin use early in adolescence, do not progress to problematic use behaviors (Kandel & Logan, 1984; Labouvie & White, 2002; Wittchen et al., 2008). Researchers are therefore increasingly concerned with distinguishing “normal” and “problem-prone” alcohol use behaviors among adolescents (Labouvie & White, 2002; Weber, Graham, Hansen, Flay, & Johnson, 1989), including the identification of subgroups of adolescents who not only evidence early initiation, but also progress toward elevated use (i.e., frequency, amount), alcohol-related problems (e.g., missing school/work, driving under the influence), and the development of alcohol abuse and dependence. Indeed, a considerable body of work exists examining developmentally salient risk and protective factors related to the onset and course of alcohol consumption among adolescents, including biological (e.g., genetic influences; Schuckit et al., 1999), cognitive (e.g., executive function; Deckel, 1999), social (e.g., familial alcohol use; Alati et al., 2005), and psychological (e.g., externalizing problems; Dobkin, Tremblay, Masse, & Vitaro, 1995) factors; the reader is directed to several excellent reviews of these literatures (e.g., Brown et al., 2008; Marsh & Dale, 2005; Saraceno, Munafó, Heron, Craddock, & van den Bree, 2009; Windle et al., 2008).

Within this broad framework, one area of increasing empirical attention is the relation between adolescent alcohol use behaviors and anxiety problems (e.g., Clark et al., 1995; Zimmerman et al., 2003). Indeed, there is a substantial literature focused on linkages between anxiety and alcohol use problems among adults (i.e., over 18 years of age; see Allan, 1995; Brady, & Lydiard, 1993; Kushner et al., 1990; Schuckit, & Hesselbrock, 1994; Wilson, 1988 for reviews), and, importantly, research suggests that anxiety-alcohol use disorder (AUD) comorbidity predicts poorer treatment outcomes and an increased risk of relapse as compared to outcomes among non-comorbid individuals (e.g., Bruce et al., 2005; Kushner et al., 2005). As with other substances characterized by anxiolytic properties (e.g., marijuana, heroin; Stewart & Conrod, 2008; Tull, Schulzinger, Schmidt, Zvolensky, & Lejuez, 2007), theoretical accounts of this association typically center on the use of alcohol in the context of affect regulation/anxiety reduction among individuals with co-morbid anxiety and alcohol use problems (e.g., self-medication; Carrigan & Randall, 2003; Robinson, Sareen, Cox, & Bolton, 2009; Sayette, 1999). Importantly, this body of work suggests variability in the association between anxiety psychopathology and alcohol use outcomes when specific anxiety diagnoses are considered (e.g., Allan, 1995; Brady, & Lydiard, 1993; Kushner et al., 1990). For example, contemporary research suggests variability in terms of rates of co-occurrence (Kessler et al., 2005), as well as the use of alcohol to “self-medicate” (Robinson et al., 2009) as a function of anxiety diagnosis. Post-traumatic stress disorder (please see Stewart, 1996 for a review), and social anxiety disorder (see Morris, Stewart, & Ham, 2005), have been the most widely studied as research consistently indicates positive relations between problematic alcohol use and these two diagnoses; however, a growing body of work suggests that generalized anxiety disorder (e.g., Smith & Book, 2010), obsessive compulsive disorder (e.g., Manacebo, Grant, Pinto, Eisen, & Rasmussen, 2009), and panic disorder (e.g., Bunaciu et al., 2010; Seguí et al., 2001) also are important clusters of anxiety psychopathology to consider in relation to alcohol use problems among adults, even after variance associated with other commonly comorbid conditions (e.g., depression, illicit substance use disorders) are taken into account.

Although theoretically promising, the empirical literature focused on the anxiety-alcohol use linkage among adolescents has produced mixed results (Alati et al., 2005; Fidalgo, da Silveira, & da Silveira, 2008; Marmorstein, White, Loeber, & Stouthamer-Loeber, 2010; Pardini, White, & Stouthamer-Loeber, 2007; Shedler & Block, 1990). For instance, in a sample of 400 urban adolescents in the United States, Schwinn, Schinke, and Trent (2010) found that elevated anxiety, as measured by the Brief Symptom Inventory (Derogatis, 1993), was associated with increased frequency of alcohol consumption over the past 30 days. Similarly, Fidalgo and colleagues (2008) found that adolescents who reported heavy alcohol consumption (i.e., at least once every day over the past 30 days) also presented with elevated anxiety as measured by the Beck Anxiety Inventory (Beck, Epstein, Brown, & Street, 1988). Indeed, the level of anxiety reported by heavy users was almost twice as high as those reported by mild and non-users. Conversely, in a sample of almost 500 boys participating in the Pittsburgh Youth Study (Loeber, Farrington, Stouthamer-Loeber, & Van Kammen, 1998), elevated anxiety, as indicated by the Achenbach rating system (Achenbach & Edelbrock, 1986, 1987; Achenbach, 1991) was unrelated to the development of alcohol abuse, and negatively associated with subsequent symptoms and diagnoses of alcohol dependence (Pardini et al., 2007). Finally, Lewinsohn, Shankman, Gau, and Klein (2004) examined relations among anxiety and AUDs in a sample drawn from the Oregon Adolescent Depression Project (OADP; Lewinsohn, Hops, Roberts, Seeley, & Andrews, 1993). Employing a structured clinical interview, the authors found that while subthreshold anxiety was positively associated with the subthreshhold and threshold AUDs, the diagnosis of a full-symptom anxiety disorder (i.e., all diagnoses combined into a single group, “anxiety”) was unrelated to the presence of an AUD. Taken together, findings such as these suggest that anxiety psychopathology may be an important variable to consider in terms of alcohol use problems among adolescents, although the precise nature and direction of the relation remains unclear.

One significant limitation of this body of work is the use of varied definitions of “anxiety” (e.g., trait anxiety, anxious/withdrawn behavior), as well as the treatment of anxiety as a unitary construct (e.g., the presence of any anxiety disorder) as opposed to examining specific constellations of symptoms. Such concerns have culminated in a call for more refined analyses of the relation between anxiety psychopathology and alcohol use behaviors among adolescents, including the differentiation of associations across the anxiety disorders (Cicchetti & Rogosch 1999; Clark, Smith, Neighbors, Skerlec, & Randall, 1994; Falk, Yi, & Hilton, 2008). Drawing upon work conducted with adults, there is reason to hypothesize that specific clusters of anxiety psychopathology may evidence distinct relations with alcohol use behaviors (Falk et al., 2008; Kessler et al., 2005; Robinson et al., 2009), and thus non-specific indicators of anxiety (e.g., trait anxiety) and related risk factors (e.g., behavioral inhibition; Fox, Henderson, Marshall, Nichols, & Ghera, 2005) may not capture important variability in terms of specific relations across the anxiety disorders (e.g., Buckner et al., 2008; Marmorstein et al., 2010). To illustrate, Buckner and colleagues (2008) examined data collected in the OADP discussed above, from which Lewinsohn and colleagues (2004) reported no association between the presence of an anxiety disorder and AUDs. However, when adolescent social anxiety disorder was isolated as the primary predictor, findings indicated a significant association with alcohol dependence at age 30 years (after controlling for gender, conduct disorder, as well as additional anxiety and mood disorders; Buckner et al., 2008). Importantly, in contrast to the rather well-developed adult literature that includes a number of integrative reviews (e.g., Allan, 1995; Brady, & Lydiard, 1993; Kushner et al., 1990), a synthesis of the extant literature focused on patterns of comorbidity among adolescents remains absent, limiting conclusions that may be drawn in this domain.

Systematic examination of associations among specific constellations of anxiety symptoms and the onset, maintenance, and progression of alcohol use behaviors (e.g., initiation, regular, hazardous, and symptomatic use) among adolescents is important for at least two reasons. First, efforts aimed at delineating temporal and potentially causal relations among alcohol use behaviors and anxiety psychopathology requires research examining incidence and progression as they occur (cf. retrospective report) before strong conclusions may be drawn with regard to how these relations play out in adolescence. Alcohol use typically begins during adolescence (Johnston et al., 2008; Newes-Adeyi, Chiung, Williams, & Faden, 2005; Wittchen et al., 2008), adolescents progress more rapidly from initial use to problematic outcomes as compared to adults (e.g., alcohol dependence; Deas, Riggs, Langenbucher, Goldman, & Brown, 2000), and the onset of many anxiety disorders/symptomatology (e.g., simple/specific phobias, panic attacks/disorder, social anxiety disorder; Last, Perrin, Hersen, & Kazdin, 1992) often occurs during adolescence, highlighting adolescence as a key period in which to better understand the etiology of anxiety, alcohol use behaviors, and their co-occurrence. Second, adolescence itself is a unique developmental phase, characterized by profound biopsychosocial development (e.g., puberty; Hayward, 2003; Rogol, Roemmich, & Clark, 2002) combined with distinct socioemotional and demographic characteristics (e.g., living with a parent/guardian; Arnett, 2000, 2007). Researchers and theorists therefore argue that adolescent anxiety psychopathology and alcohol use behaviors must be considered in the context of this developmental phase (Cicchetti & Rogosch, 1999; Deas et al., 2000). For example, research suggests that, as compared to adults, adolescents evidence attenuated sensitivity to many of the negative effects of alcohol (e.g., motor impairment, dysphoria), along with heightened sensitivity to some of the positive effects (e.g., social facilitation; please see Windle et al., 2008 for a review). Further, adolescent alcohol consumption also may have particular effects on neurological, physiological, cognitive, and social development that are not evidenced by adult alcohol users by virtue of the rapid development occurring during this period (please see Brown et al., 2008 for a review). For these reasons, “downward extension” of risk-factor models defined by work with adults without direct evaluation of the nature and extent to which these relations exist among adolescents can be misleading, particularly in terms of alcohol use (Deas et al., 2000; Dunn & Mezzich, 2007).

Collectively, there is a need to better understand the linkage between adolescent anxiety psychopathology and alcohol use behaviors. The goal of the next section is therefore to critically review and synthesize the relevant empirical literature. Here, key constructs are defined and a detailed description of the extant literature is provided (also, please see Table 1).

Anxiety and Alcohol Use among Adolescents: Extant Literature

Definitions of Key Constructs

Anxiety

Research suggests that anxiety disorder diagnoses (e.g., generalized anxiety, social anxiety) represent unique dimensions of psychopathology among adolescents, and despite some overlap in symptoms specific diagnoses should be treated as distinct diagnostic constructs (e.g., Ferdinand, Dieleman, Ormel, & Verhulst, 2007). Accordingly, all studies utilizing criteria defined in the Diagnostic and Statistical Manual for Mental Disorders-IV-Text Revision [DSM-IV-TR; American Psychiatric Association (APA), 2000] for specific constellations of anxiety symptoms or diagnostic status were included (i.e., agoraphobia, generalized anxiety/overanxious disorder, obsessive-compulsive disorder, panic attacks/panic disorder, separation anxiety disorder, social anxiety disorder/social phobia, and simple/specific phobias). Research on traumatic event exposure or posttraumatic stress disorder was excluded from the current manuscript as these data have been comprehensively reviewed elsewhere (Blumenthal et al., 2008); however, general findings from this review are included to facilitate comparison across the anxiety disorders.

Alcohol use and use-related problems

All stages of alcohol use were considered relevant to the current review, including alcohol use history (i.e., ever having used, frequency of use, standard quantity consumed), age of onset, hazardous use/binge drinking (i.e., ≥ 5 drinks in a single sitting), as well as alcohol abuse/dependence diagnostic status (APA, 2000). Finally, consistent with extant work underscoring the importance of patterns of use (e.g., irregular, recurrent) as well as progressive trajectories (Kandel & Logan, 1984; Windle et al., 2008; Yamaguchi & Kandel, 1984a; 1984b), research focused on these dimensions of alcohol use also was evaluated for inclusion.

Age range

As outlined above, the current review targets research with youth (i.e., under 18 years of age) in order to uniquely extend the extant literature on alcohol use and anxiety psychopathology. All studies focused on children (below age 10) and adolescents (ages 10 to 17) were included; studies were excluded if a wide study age range primarily reflected adulthood (e.g., National Comorbidity Survey; Kessler, 1994), if they focused solely on participants age 18 years and older, or if they relied solely on adult reports of childhood experiences, which may be more susceptible to reporting and memory biases associated with retrospective reporting across long periods of time (Nisbett & Ross, 1980). Importantly, no studies were found examining youth below age 10 years, and thus the current manuscript focuses solely on adolescents.

Literature search method

First, a search of electronic databases (PsycINFO, PUBMed, Science Direct) was conducted. All combinations of the following key words were used: anxiety, separation anxiety, social anxiety, phobia, agoraphobia, social phobia, specific phobia, generalized anxiety disorder, GAD, worry, obsessive compulsive disorder, OCD, obsession, compulsion, panic, panic attack, and panic disorder; adolescents, adolescence, children, childhood, juvenile, and youth; and substance, alcohol, ethanol, beer, wine, liquor, and drinking. Second, all articles cited in these initially identified articles that were potentially relevant to the current review were collected (backwards literature search). Empirical studies written in English were included if they met the inclusion criteria outlined above. This search yielded 16 relevant empirical articles. Please see Table 1 for a synopsis of the study findings (i.e., analytic approach, sample, operational definitions, and relevant findings). Studies are organized first by the specific disorder/symptomatology examined. Within each category/symptom cluster studies are presented as a function of level of alcohol use examined (e.g., ever used, hazardous use, AUDs).

Agoraphobia

Two studies to date have focused on adolescent agoraphobia and alcohol use behaviors, using markedly different methodological approaches. First, Zimmerman and colleagues (2003) investigated relations among several different anxiety disorders, including agoraphobia, and problematic drinking across four years in the Munich Early Developmental Stages of Psychopathology study (EDSP). Specific drinking behaviors examined were: regular use (i.e., drinking at least three times per week at peak use), hazardous use [drinking at least 40 g/day (male participants) or 20 g/day (female participants) during peak use], any AUD diagnosis, alcohol abuse without dependence, and alcohol dependence. At baseline, a diagnosis of agoraphobia was significantly related to increased regular use, hazardous use, the diagnosis of any AUD, and alcohol dependence. Prospective analyses, controlling for several demographic and psychological variables (e.g., age, other anxiety disorders, depression, illegal drug use disorders; please see Table 1) demonstrated no further associations between baseline agoraphobia and the onset or persistence of evaluated alcohol use behaviors at follow-up. Second, Clark and colleagues (1995) examined the prevalence of anxiety disorders, including agoraphobia, among adolescents in an inpatient psychiatric unit; these authors found rates of the disorder to be much higher among adolescents presenting with an AUD (6.9%) than those found in the general community (e.g., 1–2%; Kessler at al., 2005).

The scarcity of studies currently precludes definitive conclusions regarding the association between agoraphobia and alcohol use behaviors among adolescents; concurrent relations with problematic use (e.g., regular use; AUD) were observed in both a clinical and a community-based sample, although there was no evidence that agoraphobia predicts subsequent alcohol use behaviors. Future research will be required to determine whether comorbidity is consistently evident across multiple studies, how symptoms and diagnosis relate to different stages of alcohol use (e.g., onset, hazardous use, AUD), and the specific role of agoraphobia as compared to co-occurring panic-like symptoms or panic disorder.

Overanxious/ Generalized Anxiety Disorder

Overanxious disorder no longer appears in the most recent addition of the DSM (i.e., DSM-IV-TR; APA, 2000); the associated cluster of symptoms (i.e., excessive worry accompanied by physiological response) has been re-conceptualized as generalized anxiety disorder (GAD) in this most recent addition. Nonetheless, it warrants mention that three independent studies suggest overanxious disorder does not predict consumption history (e.g., ever drank, binge drinking; King, Iacono, & McGue, 2004), or the presence of an AUD (Neighbors, Kempton, & Forehand, 1992), nor is it more frequent among adolescents diagnosed with an AUD (Clark et al., 1995).

With regard to GAD, two prospective examinations suggest that this condition may be related to particular alcohol use behaviors among adolescents. First, Kaplow, Curran, Angold, and Costello (2001) found a positive relation between symptoms of generalized anxiety among currently abstaining adolescents and the initiation of alcohol use across a four year period. This association was significant after controlling for age, gender, race, and depressive symptoms. Further, in the study discussed above, Zimmerman and colleagues (2003) found that the diagnosis of GAD at baseline was related to hazardous alcohol use; however, baseline GAD was not significantly associated with the onset or persistence of identified drinking behaviors across the four year period (e.g., regular use, hazardous use, AUD).

Taken together, the small extant literature suggests GAD symptoms and diagnosis may be related to increased experimentation, although the diagnosis does not uniquely predict problematic use over time. These are necessarily tentative conclusions; additional work is needed to more firmly establish the nature and direction of the GAD-alcohol use behavior association.

Obsessive Compulsive Disorder

Only one study to date has examined the relation between OCD and alcohol use behaviors among adolescents. Specifically, Clark and colleagues (1995) reported a diagnosis of OCD among 2.3% of their sample of inpatient adolescents with an AUD. Although comparable to rates found in the general community (i.e., 1–2%; Clark et al., 1994), additional research will be necessary to clarify potential associations among different stages of alcohol use (e.g., initiation, regular) and symptoms/diagnosis of OCD. Further, OCD is a heterogeneous diagnosis, encompassing symptom patterns related to a number of different target obsessions and/or compulsions (e.g., contamination, religiosity); thus, it also will be important to differentiate consumption patterns across varied manifestations of adolescent OCD.

Panic Attacks/Panic Disorder

Extending qualitative work suggesting a linkage between panic spectrum problems and problematic alcohol use among adolescents (i.e., binge drinking; Gardner & Kutcher, 1993), seven quantitative studies have employed a variety of techniques to further clarify the nature of this relation. First, Hayward and colleagues conducted a series of studies examining relations among panic and alcohol use history. These authors found no relation between a positive history of panic attacks and ever having used alcohol among 95 ninth-grade girls and boys (Hayward, Killen, & Taylor, 1989), or current alcohol consumption (i.e., at least 2–4 full beverages or “being drunk” over the past 30 days) within a group of 160 young adolescent girls (M_age = 11.8 years; Hayward et al., 1995). Conversely, among more than 1,000 adolescent girls (Hayward et al., 1997), participants reporting a history of both cued and uncued panic attacks evidenced significantly higher rates of current alcohol consumption (as described above) as compared to girls who had never had a panic attack.

Zimmerman and colleagues (2003) found that a positive history of panic attacks was associated with an increased likelihood of regular alcohol use, hazardous use, the diagnosis of any AUD, alcohol abuse, and alcohol dependence at baseline. Further, adolescents who had experienced a panic attack were more likely to report the onset of hazardous use and alcohol abuse, as well as the persistence of any AUD during the follow-up period. Finally, the diagnosis of panic disorder was associated with the diagnosis of any AUD, and alcohol dependence specifically, at the baseline assessment, as well as the persistence of any AUD during the follow-up period.

Similarly, Domalanta, Risser, Roberts, and Risser (2003) found that among incarcerated adolescents diagnosed with panic disorder, 61.4% presented with comorbid alcohol abuse symptoms; this rate of alcohol-related symptoms is much higher than typically observed in community-based samples (e.g., 0.4–9.7%; Chung, Martin, Armstrong, & Labouvie, 2002), although the ability to generalize from the results of this study are limited by the unique nature of the sample. Further, Goodwin and colleagues (2004) also analyzed data collected in the Munich EDSP discussed earlier, finding that AUDs were more prevalent among adolescents with a positive history of panic attacks as compared to those who had never experienced a panic attack at baseline, and the presence of panic attacks at baseline was significantly associated with the onset of an AUD during the four-year follow-up period (after controlling for age, gender, and all other diagnoses). Finally, Clark and colleagues (1995) approached this question from a different angle, finding that panic disorder was diagnosed among 2.3% of their sample of inpatient adolescents with a current AUD, a rate that is comparable to that observed in community-based samples (e.g., 2.1%; Kessler et al., 2009).

Taken together, this growing body of work suggests that among older adolescents, particularly after controlling for comorbid psychopathology and additional related variables (e.g., gender, ethnicity), both panic attacks and panic disorder may confer specific risk for problematic alcohol consumption (e.g., regular use, hazardous use; AUD). Indeed, these data suggest that the diagnosis of panic disorder evidences a robust relation with problematic alcohol consumption among adolescents (i.e., odds ratios ≥ 3.0; Haddock, Rindskopf, & Shadish, 1998), and there is initial evidence to suggest that the onset of panic may predate the onset of problematic consumption (e.g, Goodwin et al., 2004; Zimmerman et al., 2003). A next step in this line of research will be to examine the temporal relations among panic (i.e., symptoms, full-symptom attacks, panic disorder diagnosis) and specific alcohol use behaviors among adolescents, including more refined analyses aimed at addressing potential moderators, mediators, and/or shared factors in these relations.

Separation Anxiety

Four studies extended initial qualitative data that suggested a potential linkage between separation anxiety and episodic binge drinking (Gardner & Kutcher, 1993). First, in the study noted above, Kaplow and colleagues (2001) found that adolescents reporting elevated separation anxiety symptoms were significantly less likely to have initiated alcohol during the four-year follow-up period. Further, King and colleagues (2004) found no significant differences in alcohol use history (e.g., ever drank, binge drinking) by age 14 years as a function of SAD status at age 11 years.

Next, in their sample of inpatient adolescents diagnosed with an AUD, Clark and colleagues (1995) found that 11.6% also presented with comorbid separation anxiety disorder (SAD); a rate that falls within the range of lifetime estimates found in other community-based work (i.e., up to 18.6%; Clark et al., 1994). Finally, prospective analyses conducted by Brückl and colleagues (2007) indicated that SAD was unrelated to alcohol abuse, although it evidenced an association with alcohol dependence in the EDSP sample. Specifically, adolescents with SAD evidenced significantly higher rates of lifetime alcohol dependence (27.9%) as compared to those without the diagnosis (6.4%). Further, SAD at baseline predicted the subsequent onset of alcohol dependence at a four-year follow-up assessment, as well as demonstrating a significant association with lifetime alcohol dependence after controlling for additional DSM-IV psychopathology.

Collectively, available work suggests that SAD is generally unrelated to alcohol use behaviors among adolescents, with the important exception of one prospective study that reported a robust association with the presence, onset, and persistence of alcohol dependence. The specificity of this relation is striking, although it needs to be replicated, and future research is needed to better understand the underlying mechanisms and progression of these disorders as they co-occur.

Social Anxiety/Social Phobia

Five studies focused on the linkage between social anxiety and alcohol use behaviors. First, Blumenthal, Leen-Feldner, Frala, Badour, and Ham (2010) found that level of social anxiety was unrelated to frequency of alcohol consumption in a community-based sample of 50 adolescents, although rates of alcohol use in the sample were relatively low. Conversely, Zimmerman and colleagues (2003) found that social phobia was associated with the diagnosis of alcohol dependence, and prospective analyses demonstrated that baseline social phobia was related to the onset of regular and hazardous use, as well as the persistence of alcohol dependence across the four-year follow-up period.

Similarly, in a large sample of adolescent twin girls, Nelson and colleagues (2000) reported a two-fold increase in the risk for a concurrent AUD diagnosis among those diagnosed with social anxiety disorder as compared to those without the diagnosis (controlling for chronological age). Further, among those with comorbid social anxiety disorder-AUD, the age of first social fear significantly predated the age of first alcohol intoxication. Essau, Conradt, and Petermann (1999) found that the presence of an AUD was evidenced by 9.6% of adolescents presenting with significant social fear (without meeting criteria for social phobia), and 23.5% of adolescents diagnosed with social phobia/social anxiety disorder (cf., 0.8–9.2% in community-based samples; Chung et al., 2002). Finally, among the inpatient adolescents with an AUD diagnosis assessed by Clark and colleagues (1995), 20.9% also presented with comorbid social phobia; a rate that also is higher than those observed in community-based samples (e.g., 9.2%; Kessler et al., 2009).

Collectively, this growing body of work is suggestive of a positive association between social anxiety and problematic alcohol use behaviors among adolescents. Further, large-scale retrospective (Nelson et al., 2000) and prospective (Zimmerman et al., 2003) work suggests that social anxiety is often the primary, or antecedent, disorder. Future research will be required to further delineate the specificity of social anxiety as a risk factor for problematic alcohol use among adolescents, the role of alcohol use behaviors in the maintenance and progression of social anxiety, as well as additional variables that may influence these relations.

Simple/Specific Phobias

Three studies have examined the link between simple/specific phobias and alcohol use behaviors. First, Zimmerman and colleagues (2003) reported significant baseline associations between the diagnosis of a specific phobia and hazardous drinking, the diagnosis of any AUD, alcohol abuse, and alcohol dependence; however, prospective analyses indicated that these relations did not persist over time. Next, Clark and colleagues (1995) reported that 16.2% of inpatient adolescents with an AUD also were diagnosed with a simple phobia; a rate which was not significantly different from that found in a matched control group drawn from the general community. Finally, Ilomäka and colleagues (2004) examined the relation between phobic disorders broadly (i.e., simple phobia, agoraphobia, social phobia) and alcohol dependence among adolescents in a psychiatric inpatient unit. After controlling for age, gender, and additional diagnoses, phobic patients were approximately 4.5 times more likely to present with alcohol dependence (without another substance use disorder) than were non-phobic patients. Finally, the onset of the phobic disorder preceded or occurred at the same time as the onset of comorbid substance dependence (including alcohol) in 90.5% of comorbid cases. Importantly, these findings are conflated with the inclusion of social phobia and agoraphobia, thus must be viewed with caution in terms of assessing relations with specific phobias exclusively.

The scant research on the specific phobia-alcohol use association suggests the link is tentative at best. Importantly, the diagnosis of a specific phobia can include a wide variety of focal fears (e.g., water, injections, specific animals) and thus, as in the case of OCD, it may be important to parse apart relations among specific manifestations of the disorder before definitive conclusions can be drawn.

Traumatic Event Exposure/PTSD

Finally, a growing body of work suggests an association between problematic alcohol use, traumatic event exposure, and PTSD, although findings are not uniform (Blumenthal et al., 2008). Specifically, with few exceptions (Lipschitz et al., 2003; Stein et al., 2002), extant work indicates a relation between traumatic event exposure and elevated (e.g., increased frequency; Schiff, Benbenishty, & McKay, 2006) and/or problematic alcohol use among adolescents (e.g., AUD symptoms/diagnosis; Clark, Lesnick, & Hegedus, 1997; Giaconia, Reinherz, & Silverman, 1995; Perkonigg, Kessler, Storz, & Wittchen, 2000). Further, preliminary analyses of temporal patterning suggests a bi-directional relation (Clark, Lesnick, et al., 1997; Perkonigg et al., 2000), with some work finding the most frequent temporal pattern to be an AUD preceding traumatization (Clark, Lesnick, et al., 1997), and others reporting traumatization predicting the onset of a subsequent AUD (Perkonigg et al., 2000). Research regarding the association between problematic alcohol use behaviors and PTSD has been less consistent, with a number of large-scale studies indicating a positive relation between PTSD and AUD diagnoses (e.g., Clark, Pollock, et al., 1997; Giaconia et al., 1995; Perkonigg et al., 2000), and other work finding no association above and beyond the effects of traumatic event exposure alone (Lipschitz, Rasmusson, Anyan, Cromwell, & Southwick, 2000; Schiff et al., 2006). Please see Blumenthal and colleagues (2008) for a comprehensive review of this literature, including discussion of potential moderators and mediators (e.g., cultural influences, motivational models) as well as future directions in terms of research and treatment design.

Summary

The literature focused on specific constellations of DSM-defined anxiety and alcohol use behaviors among adolescents is still in its infancy although some general patterns are discernable. Specifically, a final analysis of the extant literature in which the more rigorous studies available are given particular emphasis suggests that OCD and overanxious disorder are unrelated to adolescent alcohol use behaviors; generalized anxiety, SAD, specific phobias, and agoraphobia evidence tentative associations with specific alcohol use behaviors, and panic attacks, panic disorder, and social anxiety demonstrate relatively consistent associations with a wide variety of problematic alcohol use indicators. For example, the research to date suggests that certain constellations of symptoms are unrelated to frequency of alcohol use (e.g., SAD, social anxiety), but positively associated with the development of AUDs (e.g, Blumenthal et al., 2010; Brückl et al., 2007; Nelson et al., 2000; Zimmerman et al,, 2003), and others appear related to initiation of use, but not problematic use behaviors (e.g., GAD; Kaplow et al., 2001; Zimmerman et al, 2003). Further, initial evidence suggests that certain symptom clusters tend to onset prior to the development of problematic use behaviors (i.e., panic, social anxiety), whereas others evidence concurrent, but not prospective, associations (e.g., agoraphobia; Zimmerman et al., 2003). Finally, prior work suggests that traumatic event exposure and PTSD also evidence associations with elevated alcohol use (e.g., frequency) as well as the presence of an alcohol use disorder (please see Blumenthal et al., 2008). Collectively, these data suggest variation in the anxiety-alcohol use linkage among adolescents when specific diagnostic categories are considered; importantly, these data suggest unique relations with certain constellations of symptoms (e.g., panic disorder, social anxiety) even after variance associated with important demographic (e.g., gender), psychological (e.g., depression), and behavioral (e.g., other substance use) variables are accounted for (e.g., Zimmerman et al., 2003). However, the defining limitation of this body of work is its size; independent replication and extension of extant work is indicated for every anxiety disorder if we are to have a sophisticated and comprehensive perspective on alcohol use –anxiety comorbidity among adolescents. Further, no studies have utilized experimental or laboratory-based designs to better understand this association. The following discussion summarizes the methodological strengths and challenges of the research base and utilizes an experimental psychopathology framework to provide specific suggestions for forwarding this important area of work.

Future Directions

Although considerable work is still required to draw definitive conclusions regarding the relative role of specific constellations of anxiety psychopathology in relation to adolescent alcohol use behaviors, there is some evidence to suggest that panic and social anxiety demonstrate relatively consistent associations and may hold predictive value in terms of the onset and progression of problematic drinking behaviors among adolescents. Accordingly, these two forms of psychopathology serve as primary illustrative examples in the discussion of specific ways in which this literature can best be advanced. In this section, methodological considerations and suggestions for future research are first articulated. Potential moderators, mediators, and shared factors of these relations are then highlighted and discussed as important factors to consider in forwarding developmentally sensitive models of anxiety-alcohol use comorbidity among adolescents.

General Methodological Considerations

First, it is important to note that the literature is characterized by single study data-collections (e.g., Clark et al., 1995; Zimmerman et al., 2003). Thus, independent replication of the current findings is a necessary first step in this line of research. Further, almost no work has sought to delineate relations among the onset, frequency, and progression of alcohol use as it relates to prodromal, subclinical, and diagnostic status across the anxiety disorders. A convergence of methodological approaches will be necessary to comprehensively address these questions. However, research to date has exclusively utilized self-report indices to measure anxiety and alcohol use behaviors. Efforts to reduce reporting biases (e.g., memory distortions), including the employment of multi-informant, “real time” (e.g., momentary random sampling), and experimental psychopathology procedures will enhance confidence in conclusions drawn from future work, particularly in terms of etiological and maintaining processes (e.g., self-medication; Robinson et al., 2009; Sayette, 1999). For example, preliminary work conducted with nicotine dependent adults indicates that individuals who report greater difficulty maintaining quit attempts (e.g., longer than 24 hours) also evidence increased negative affect (e.g., anxiety, irritation) in response to laboratory-based stressors, providing direct evidence for a link between emotional responding and smoking relapse indicated by prior self-report work (Brown, Lejuez, Kahler, & Strong, 2002; Zvolensky, Feldner, Eifert, & Brown, 2001). Because experimental psychopathology techniques represent an essential extension to the literature, the remainder of this section will focus specifically on this approach as an important avenue for better understanding the relation between adolescent anxiety and alcohol use outcomes.

The Role of Experimental Psychopathology

There is a rich history of using safe and effective techniques to model various affective and clinically-relevant states among children and adolescents (e.g., Brenner, 2000; Pine et al., 2005; Tsao, Lu, Kim, & Zeltzer, 2006). Careful application of these types of approaches, with particular attention to unique ethical issues that pertain to working with vulnerable populations, holds particular promise in terms of advancing our understanding of the anxiety-alcohol use linkage among adolescents. Indeed, experimental psychopathology techniques represent a uniquely powerful, yet thus far untapped, mechanism through which a more precise understanding of anxiety-problematic alcohol use comorbidity can be obtained. Work conducted with adults has provided important data regarding the linkage between anxiety and problematic alcohol use (e.g., Abrams, Kushner, Medina, & Voight, 2002; Coffey et al., 2002; Coffey, Stasiewiez, Hughes, & Brimo, 2006; de Wit, Söderpalm, Nikolayev, & Young, 2003; Mulligan & McKay, 2001). Ascertainment of similar or distinct findings among adolescents, above and beyond other related variables (e.g., depression, illicit substance use), would enhance our understanding of developmental factors contributing to the anxiety-alcohol association.

There are at least three other benefits of experimental psychopathology methods that could significantly advance current models. First, tightly controlled experimental designs permit causally-oriented hypothesis testing and control for a number of threats to internal validity that characterize naturalistic work in this area. Second, experimental psychopathology techniques can incorporate real-time multi-modal assessment, allowing for more sophisticated analyses of complex behavioral, physiological, and cognitive correlates and potential mechanisms underlying comorbid presentation. Finally, laboratory analogues can provide important information regarding the etiology, maintenance, and treatment of comorbid conditions, particularly in the context of longitudinal designs. Each of these features is discussed in more detail below.

Experimental Design

The ability of researchers to induce anxious responding in the laboratory provides fertile ground for the assessment of alcohol-related variables in the context of anxiety, and there are multiple studies utilizing analogues of both panic (e.g., voluntary hyperventilation; Leen-Feldner, Feldner, Bernstein, McCormick, & Zvolensky, 2005; Unnewehr, Schneider, Margraf, Jenkins, & Florin, 1996) and social anxiety (e.g., the Trier Social Stress Test; Buske-Kirschbaum et al., 1997) among adolescents. Further, although ethical considerations clearly exclude the use of conventional alcohol-related manipulations among adolescents (e.g., alcohol administration; c.f., Behar et al., 1983), there are a number of available methods for assessing alcohol-relevant outcomes, including self-reported desire to use alcohol (e.g., Love, James, & Willner, 1998) and alcohol use expectancies (e.g., Fromme & D'Amico, 2000), as well as experimental tasks such as dichotic listening, dot probe, and implicit memory tasks designed to assess for selective attention to alcohol-related cues. These types of methods lend themselves to addressing causally-oriented hypotheses and allow for other important experimental features (e.g., random assignment; control groups) which can reduce concerns related to the specificity of associations identified in naturalistic designs (e.g., comorbid psychopathology, additional substance use). For example, attentional bias toward alcohol-relevant stimuli among adolescents randomly assigned to a social anxiety induction, as compared to a control condition, would provide evidence relevant to whether social anxiety elevates risk for increased alcohol consumption. Indeed, related work conducted with adult samples provides evidence for elevated alcohol consumption (e.g., number of beverages, strength of beverage) following a social stressor (e.g., Abrams et al., 2002; deWit et al., 2003); however, to date no work has examined this relation among adolescents. Importantly, such designs could be used among healthy adolescents in efforts to clarify questions related to etiology (Zvolensky, Lejuez, Stuart, & Curtin, 2001) as well as among adolescents with elevated social anxiety/social anxiety disorder addressing questions related to maintenance-type processes.

Multimodal, Real-Time Assessment

Experimental psychopathology techniques also allow for real time, multi-modal assessment, such that psychological (e.g., disorder-specific symptom level), cognitive (e.g., desire to drink), physiological (e.g., heart rate), neurological (e.g., prefrontal activation; Tapert et al., 2003) and behavioral (e.g., approach/avoidance behaviors during a speech task; Roelofs, Elzinga, & Rotteveel, 2005) responding to a challenge procedure can be concurrently assessed. These types of indicators would allow for the kinds of sophisticated analyses of the anxiety-problematic alcohol use association that are currently missing from the literature. For example, drawing upon adult work in the area (e.g., Coffey et al., 2002; Mulligan & McKay, 2001), self-reported anxiety, desire to drink, and psychophysiological responding to alcohol-relevant cues (e.g., words, pictures; Lowman, Hunt, Litten, & Drummond, 2000; Tapert et al., 2003) following anxiety induction would provide a much richer picture of the alcohol-anxiety linkage among adolescents than is permitted by exclusive reliance on self-reported anxiety and alcohol use behaviors.

Longitudinal Designs

Prospective research designs are required to draw meaningful conclusions regarding the temporal, causal, and functional relations among anxiety symptomatology, disorder, and problematic alcohol use among adolescents. Indeed, questions related to temporal ordering are at the heart of this body of research. Although the majority of the extant literature has focused on the role of anxiety in promoting or exacerbating problematic drinking (i.e., anxiety temporally precedes and/or causally influences consumption; Carrigan & Randall, 2003; Robinson et al., 2009; Sayette, 1999), it is also plausible that 1) problematic alcohol use, via biological, cognitive, and/or social mechanisms precipitates or exacerbates anxiety psychopathology (e,g, Brady, & Lydiard, 1993; Haynes et al., 2005), or 2) both anxiety and problematic alcohol use may be related to an underlying, shared risk factor (e.g., genetic influences; Saraceno et al., 2009). Experimental psychopathology techniques can be integrated into longitudinal studies aimed at clarifying questions related to temporal order, providing a sophisticated and novel complement to traditional self-report methods and enhanced sensitivity to prodromal symptoms. Importantly, repeated assessment that begins prior to the onset of anxiety psychopathology and alcohol exposure will allow for a better understanding of the progression of comorbid problems. For example, if currently abstaining adolescents who evidence significant anxiety and/or attenuated dishabituation to a social stress task go on to evidence a problematic alcohol use trajectory (e.g., earlier initiation, symptoms of abuse or dependence, elevated coping-related consumption), the conclusion can be drawn that a proclivity to social anxiety is related to later alcohol use problems. Finally, experimental psychopathology methods can be used to assess the maintaining role of comorbidity in the context of treatment. Indeed, laboratory-based assessments of anxiety and alcohol craving in conjunction with clinical sessions among adults with these problems have provided promising functional and theoretical data which will be important to consider in the refinement of existing treatment models (e.g., Coffey et al., 2006).

Summary

At present, the available research on adolescent anxiety and problematic alcohol use poses more questions than are addressed. Further, while there is some evidence that anxiety psychopathology, particularly panic and social anxiety disorder, may confer risk for problematic alcohol use among adolescents, the literature is characterized by a number of limitations including a pressing need for additional studies. Going forward, sophisticated and creative research designs, including the use of experimental psychopathology techniques, will be needed to fully understand the nature and direction of the association. Finally, as articulated in the next section, the research base will be most usefully advanced by testing multivariate models in which key developmentally relevant mediators, moderators, and shared factors are examined.

Developmentally Relevant Moderators, Mediators, and Shared Factors

Comprehensive study of the nature of the anxiety-alcohol association among adolescents is needed, including systematic focus on variables that may account for (mediate) or influence (moderate) this linkage, as well as factors that drive the development and maintenance of anxiety and alcohol use problems (shared factors). Indeed, researchers have identified a broad range of factors which may contribute to problematic alcohol consumption among adolescents (e.g., genetic factors, executive function, depression, externalizing problems; Brown et al., 2008; Saraceno et al., 2009; Windle et al., 2008). Further, regular alcohol consumption often occurs in the context of additional substance use (e.g., nicotine, marijuana; Yamaguchi & Kandel, 1984a); accordingly, research examining the role of polysubstance use and related processes (e.g., distress tolerance; Zvolensky, Feldner, et al., 2001) will be important in forwarding the larger literature, as well as the specificity of the anxiety-problematic alcohol use linkage, among adolescents. Gender, parental psychopathology, alcohol expectancies and motives, anxiety sensitivity, and pubertal timing are briefly illuminated here as specific variables in need of additional study. These particular factors were selected because they: 1) reflect important developmental facets of adolescence (e.g., increased gender typing, cognitive development; Galambos, Almeida, & Petersen, 1990; Kuhn & Franklin, 2006; Luna, Garver, Urban, Lazar, & Sweeney, 2004), 2) have been linked to both anxiety and alcohol use, as well as additional forms of psychopathology associated with alcohol use problems (e.g., depression; Brown et al., 2008) and problematic substance use broadly (e.g., Marsh & Dale, 2005; Wallace et al., 2003), and 3) can be readily incorporated into ongoing and future research designs (e.g., via self-report measures and/or brief clinical interviews). Finally, the phenomenology of adolescent alcohol use behaviors (e.g., polysubstance use) is briefly discussed prior to conclusionary remarks.

Gender

Although research suggests that overall, adolescent boys tend to be at greater risk for the onset of problematic drinking behaviors (e.g., Wallace et al., 2003), numerous large-scale studies indicate that female adolescents evidence elevated panic and social anxiety symptoms (Chorpita, Yim, Moffitt, Umemto, & Francis, 2000), as well as increased rates of panic attacks, panic disorder (Reed & Wittchen, 1998) and social anxiety disorder (Essau et al., 1999) as compared to male adolescents. To date, the role of gender in this line of work has been addressed, if at all, via methodological (e.g., Hayward et al., 1997; Nelson et al., 2000) or statistical (e.g., Goodwin et al., 2004; Zimmerman et al., 2003) control, and no work has directly examined the relation between adolescent panic or social anxiety and alcohol use behaviors across genders. A number of theoretically-relevant questions need to be addressed, including the specificity of associations across genders (e.g., Buckner & Turner, 2009), whether temporal patterning is similar across genders, as well as the role of gender-specific biological (e.g., sensitivity to alcohol) and contextual (e.g., cultural expectations) factors that may influence trajectories in terms of alcohol-anxiety comorbidity.

Parental psychopathology

The extant literature also suggests a high degree of co-occurrence across parent-child anxiety psychopathology (Beidel & Turner, 1997; McClure, Brennan, Hammen, & LeBrocque, 2001) and problematic alcohol use (Alati et al., 2005; Russell, 1990). Given high rates of comorbid alcohol use disorders, panic, and social anxiety among adults (Kushner et al., 1990), future research will benefit from further assessing whether and how comorbid anxiety-AUDs among parents relate to panic, social anxiety, and alcohol use behaviors among adolescents. A number of methodological techniques can be utilized to examine questions such as the presence of comorbidity among offspring of parents of varying diagnostic status as well as the direct and indirect role of genetic transmission, parental modeling, and information transmission in precipitating familial co-occurrence. For example, parental psychophysiological and behavioral response to an anxiety provocation procedure and voluntary alcohol administration task could be examined as prospective predictors of offspring anxiety psychopathology and alcohol use behaviors; such designs would begin to clarify the unique and additive/interactive effects of parental alcohol use behaviors and anxious reactivity on comorbid anxiety-alcohol use among their children. Further, assessment tools such as the Learning History Questionnaire (Ehlers, 1993; Stewart et al., 2001), modified to examine parental alcohol use in the context of anxiety, could then be tested as a partial mediator of parent and offspring drinking behaviors (e.g., actual and/or desired) following a laboratory-based anxiety induction.

Alcohol use expectancies and motives

Both positive outcome expectancies and coping-related motives for alcohol consumption evidence strong associations with increased alcohol consumption (i.e., amount and frequency) and alcohol-related problems (e.g., missing school) among adolescents (Christiansen, Smith, Roehling, & Goldman, 1989; Comeau, Stewart, & Loba, 2001; Kuntsche, Knibbe, Gmel, & Engels, 2005). Drawing upon the extant literature conducted with adults, there is reason to believe that both panic and social anxiety may be related to increased coping-related alcohol consumption (Ham, Zamboanga, Bacon, & Garcia, 2009; Stewart, Morris, Mellings, & Komar, 2006; Thomas, Randall, & Book, 2008) as well as positive outcome expectancies (e.g., tension reduction, social facilitation; Lehman, Brown, Palfai, & Barlow, 2002; Mulligan & McKay, 2001; O'Hare, 1990), and preliminary evidence exists indicating increased coping-related motives for alcohol use among socially anxious (Blumenthal et al., 2010) and trauma-exposed (Dixon, Leen-Feldner, Ham, Feldner, & Lewis, 2009) adolescents. In addition to continued efforts to clarify the patterning of alcohol motives/expectancies and anxiety symptomatology among adolescents (e.g., Comeau et al., 2001), work in this area may be most usefully advanced by the use of experimental designs. For example, examination of the potential mediating or moderating role of alcohol use expectancies and motives in the relation between anxious responding to a laboratory-challenge and subsequent desire to consume alcohol would provide important information regarding potential mechanisms underlying these commonly comorbid conditions.

Anxiety sensitivity

Anxiety sensitivity (AS), or fear of anxiety-relevant sensations (e.g., racing heart; Reiss, 1991), is arguably the most well-established cognitive vulberability factor identified in research pertaining to both anxiety (Olatunji, & Wolitzky-Taylor, 2009; Taylor, 1999) and problematic alcohol use (e.g., Comeau et al., 2001; Schmidt, Buckner, & Keough, 2007). For example, research conducted with adolescents indicates that elevated AS is associated with fearful responding to a panic-relevant voluntary hypervenitlation challenge (Leen-Feldner et al., 2005) as well as problematic alcohol use behaviors (e.g., coping-related consumption; Comeau et al., 2001), and prospectively predicts the development of panic attacks (Hayward, Killen, Kraemer, & Taylor, 2000) as well as AUD diagnoses (sample age 16–24 years; Schmidt et al., 2007). Further, preliminary data suggest that selective interventions targeting AS, shown to increase abstinence and reduce alcohol-related problems (e.g., missing school; Conrod, Stewart, Comeau, & Maclean, 2006), also reduce the prevelance of panic attacks among adolescents (Castellanos & Conrod, 2006). Importantly, no work has directly tested the role of AS in terms of comorbid anxiety-problematic alcohol use among adolescents. The utilization of experimental psychopathology techniques as this literature advances could provide important data in terms of both prevention and treatment efforts, including the specific role of AS in this linkage. For example, the efficacy of prevention efforts targeting AS would be bolstered by data addressing real time responding to anxiety and alcohol-relevant provocations. Further, research indicating that comorbid, high-AS adolecents continue to evidence fearful responding to biological challenge procedures following successful AUD treatment would provide more compelling evidence for the role of AS as a shared vulnerability factor in terms of the panic-AUD linkage, and thus forward the development and research of intervention efforts targeting AS in precluding adolescent (and adult) panic-AUD comorbidity.

Pubertal timing

Finally, as previously discussed, the hallmark feature of this developmental phase, puberty (Hayward & Sanborn, 2002), evidences unique incremental validity in the prediction of both anxiety psychopathology and alcohol use (Hayward, 2003). Further, there is compelling evidence to suggest that early maturation in particular may confer specific risk for the development of panic-spectrum problems (please see Reardon, Leen-Feldner, & Hayward, 2009 for a review), social anxiety (Deardorff et al., 2007), as well as problematic alcohol use (Hayward, 2003), suggesting this may be a potential shared risk factor. A number of proposed mechanisms have been implicated in these relations, including preparation for the psychosocial changes that accompany puberty (Deardorff et al., 2007; Hayward, 2003), increases in somatic perturbation (Reardon et al., 2009), real or perceived isolation from the larger peer group (Blumenthal et al., 2009; Greene, Kremar, Walters, Rubin, & Hale, 2000), as well as association with deviant peers (Hayward, 2003). However, prior to testing of these more sophisticated models of co-occurring anxiety psychopathology and alcohol use in the context of maturational timing, research must first establish whether pubertal development, and the timing of development specifically, is related to an increased risk for comorbid alcohol problems and social anxiety or panic, or whether it simply marks the onset of processes related to the independent development of each.

Phenomenology of alcohol use among adolescents

In seeking to better understand etiologic, intervening, and maintenance factors related to the anxiety-problematic alcohol use association among adolescents, careful attention must be paid to the broader context in which this linkage unfolds over time. In particular, it is noteworthy that polysubstance use is common (e.g., Kessler et al., 2005; Yamaguchi, & Kandel, 1984a) so for instance, clarifying the role of nicotine and/or marijuana in this association is warranted. Better integration of extant work linking anxiety to other types of substances will inform these efforts. From a methodological perspective, this could mean selecting samples on the basis of particular constellations of substances used (e.g., comparing anxious responding to a laboratory challenge as a function of alcohol versus co-morbid alcohol and marijuana use). For example, Buckner, Silgado, & Schmidt (2011) examined the role of gender and diagnostic status in predicting craving for marijuana throughout a speech task (as compared to a reading task) among marijuana using adults. Results indicated that increased craving was evidenced by women and participants with social anxiety disorder, particularly during the speech task; importantly, the speech task did not elevate craving among men (as compared to women) or those without social anxiety disorder (as compared to those with the diagnosis). Utilization of a similar paradigm to examine craving among adolescent alcohol, marijuana, and co-morbid users will provide important information regarding the role of gender and social anxiety in predicting adolescent substance use behaviors. Finally, efforts must be directed at disentangling anxiety psychopathology per se from correlated risk processes that also elevate risk for alcohol use, including distress tolerance (e.g., Zvolensky, Feldner, et al., 2001), impulsivity (e.g., Askénazy et al., 2003), and attentional biases (e.g., Field, Christainsen, Cole, & Goudie, 2007). This kind of fine-grained analysis is now needed if we are to have a sophisticated and comprehensive understanding of the complex relation between anxiety and alcohol use behaviors among adolescents.

Conclusion

Collectively, an emerging body of work suggests reliable associations among panic-spectrum problems, social anxiety, and problematic alcohol use behaviors among adolescents. Overall, the literature base is small; needed are studies aimed at replicating and uniquely extending available work as well as a more comprehensive evaluation of why and how these factors are related. Further, rigorous and creative methodologies are necessary to parse apart these proposed associations. Specifically, longitudinal and experimental designs aimed at specifying the relations among anxiety psychopathology and specific stages of alcohol use will be necessary for the development of sophisticated prevention and treatment efforts for indicated adolescents. The use of experimental psychopathology techniques specifically will significantly advance our understanding of the boundaries and underlying mechanisms related to anxiety-problematic alcohol use comorbidity among adolescents, and thus represent a critical next step in this line of research.

Appendix A: Overview of research (listed alphabetically) on the relation between alcohol use and DSM-defined anxiety disorders among adolescents

Table 1.

Author(s)/Design/Analytic Approach	Sample	Primary Variables	Measures	Results
Blumenthal et al., 2010 Cross-sectional Descriptives; linear regression controlling for age, frequency of alcohol use, and affective problems	N = 50 adolescents (12–17 years; Mage = 16.35, SD = 1.1 years; 52% girls; 10% Hispanic, 78% Caucasian, 10% African American, 2% Native American, 10% “other”) with a positive history of alcohol consumption drawn from the general community	DSM Anxiety Level of social anxiety Alcohol Use Current frequency of use Motives for use (i.e., coping, enhancement, social, conformity)	RCADS-SP AADIS DMQ-R	Across entire sample: Social anxiety M = 8.10, SD = 5.08; 30% “tried” alcohol, 36% used several times per year, 18% several times per month, 16% on weekends Social anxiety unrelated to frequency of alcohol use (r = .10), enhancement (r = .20), social (r = .23), or conformity motives (r =.10), and significantly associated with coping motives (r = .47) Primary analyses After controlling for age, frequency of alcohol use, and affective problems, level of social anxiety significantly predicted coping-related motives (t = 2.69, β = .41, sr2 = .13), but not enhancement (sr2 = .00), social (sr2 = .07), or conformity motives (sr2 = .00)
Brückl, et al., 2007 Prospective (4-year) Frequency counts; Cox regressions with time-dependent covariates adjusted for gender and age	N = 1,090 youth (14–24 years) who completed baseline (ages 14–17 years) and two follow-up assessments (approximately 2 years apart), drawn from a larger random population sample of residents in Munich	DSM Anxiety SAD Symptomatic: at least one SAD symptom Subthreshold: falls short of diagnosis by one criterion Threshold: meets DSM-IV criteria Alcohol Use Lifetime alcohol abuse and dependence	M-CIDI	Alcohol abuse (AA) Present in 20.9% of no SAD group, 17.5% of Symptomatic group (OR: 1.0, 95% CI: 0.5–2.0), 15.3% Subthreshold (OR: 0.8, 95% CI: 0.4–1.6), and 7.3% Threshold (OR: 0.3, 95% CI: 0.0–2.7) AA was not significantly related to subthreshold (HR = 0.9, 95% CI: 0.5–1.6) or threshold (HR = 0.5, 95% CI: 0.0–2.8) SAD Alcohol dependence (AD) Present in 6.4% of no SAD group, 8.8% of Symptomatic group (OR: 1.9, 95% CI: 0.7–4.6), 11.3% Subthreshold (OR: 2.3, 95% CI: 1.0–5.2, p < .05), and 27.9% Threshold (OR: 17.7, 95% CI: 1.8–32.7, p < .05) AD significantly associated with subthreshold (HR = 2.1, 95% CI: 1.1–4.1) and threshold (HR = 4.7, 95% CI: 1.7–12.4) SAD Strictly prospective analyses indicated that SAD predicted the subsequent onset of AD (OR: 3.3, 95% CI: 1.06–10.2) Relation between subthreshold SAD and AD lost statistical significance when controlling for DSM-IV disorders occurring prior to SAD (HR = 1.8, 95% CI: 0.9–3.3)Relation between threshold SAD and AD remained stable after controlling for prior DSM-IV disorders as well as lifetime panic disorder with agoraphobia (HR = 3.5, 95% CI: 1.02–12.1)
Clark et al., 1995 Cross-sectional (descriptive) Frequency counts; McNemer test for dependent samples	N = 132 adolescents n = 43 youth (ages 12–18 years, Mage = 15.8, SD = 1.3 years; 51% girls; 79% Caucasian, 21% African American) diagnosed with an AUD from an inpatient psychiatric unit n = 89 adolescents from the community n = 30 pairs selected from the above samples matched on age, gender, race, and mother's education level	DSM Anxiety Social phobia Simple phobia Overanxious disorder SAD Agoraphobia OCD Panic Disorder GAD Alcohol use AUD (inpatient participants only)	K-SADS (1981) LHA Interview	Among all inpatient youth: 39.5% were met criteria for any anxiety disorder, 27.9% multiple anxiety disorders, 20.9% social phobia, 16.2% PTSD, 16.2% simple phobia, 13.9% overanxious disorder, 11.6% SAD, 6.9% agoraphobia, 2.3% OCD, 2.3% PD, and none for GAD Inpatient youth with an anxiety disorder: 88.2% comorbid affective disorder, 64.7% disruptive behavior disorder Matched pairs: anxiety disorders diagnosed in significantly more inpatient (36.6%) than community (10%) participants Simple phobias did not differ across groups 23.5% of patients who had met criteria for an anxiety disorder at intake were noted as having an anxiety disorder at discharge
Domalanta et al., 2003 Cross-sectional (descriptive) Frequency counts	N = 1,024 (19% Caucasian, 39% African American, 35% Hispanic, 7% “other") adolescents (ages 11–18 years; 26% girls) held in the Harris County Juvenile Detention Center for more than 48 hours	DSM Anxiety Panic Disorder Alcohol Use Symptoms of alcohol abuse in the past 6 months	PHQ	9.7% of youth presented with MDD, 11.9% with symptoms of a depressive disorder NOS, 4.7% with symptoms of another mood disorder, 40.3% with symptoms of drug abuse, 14.8% with symptoms of a somatoform disorder, and 8.4% with symptoms of any anxiety disorder 5.4% met criteria for PD 26.3% presented with symptoms of alcohol abuse Of those with PD, 61.4% presented with comorbid alcohol abuse symptoms
Essau et al., 1999 Cross-sectional (descriptive) Frequency counts	N = 1035 adolescents (ages 12–17 years; Mage = 14.3, SD = 1.7 years; 59% girls) randomly selected from 36 schools in Bremen	DSM Anxiety Social Phobia Social fear: endorsed significant fear of at least one social situation without meeting full criteria for social phobia Alcohol Use Lifetime AUD	M-CIDI	47.2% endorsed social fears; 1.6% met criteria for social phobia Any social fear 24.3% presented with a depressive disorder, 25.4% an anxiety disorder, 17.8% a somatoform disorder, and 12.3% any SUD 9.6% presented a lifetime AUD Social phobia 29.4% presented with a comorbid depressive disorder, 64.7% at least one other anxiety disorder, 41.2% a somatoform disorder, and 23.5% any SUD 23.5% presented with an AUD
Gardner & Kutcher, 1993 Case study Case descriptions	N = 3 youth (one 17 year old boy, one 17 year old girl, and one 18 year old girl) presenting with dimenhydrinate abuse	Case 1: DSM Anxiety Panic disorder Alcohol use Lifetime alcohol abuse Case 2: DSM Anxiety Separation anxiety Panic attacks Alcohol use History of use Case 3: DSM Anxiety Panic attacks Alcohol use History of use	K-SADS Urine analysis K-SADS Not identified Self-reported use Not identified Self-reported use	Case 1 17 year old boy, met criteria for MDD, dysthymia and PD Denied alcohol abuse and urine analysis was negative Case 2 18 year old girl diagnosed with depression, separation anxiety which spontaneously remitted by age 14 years, and repeated panic attacks Reported occasional alcohol use and binge drinking Case 3 17 year old girl, diagnosed with bipolar disorder type II and occasional panic attacks Reported occasional alcohol use and binge drinking
Goodwin et al., 2004 Prospective (4-year) Frequency counts; logistic regression adjusted for age and gender	N = 3,021 youth (14–24 years) drawn from a government registry of residents in Munich n = 1,228 (ages 14–17 years at baseline) completed the 2-year follow-up assessment n = 2548 (ages 14–24 at baseline) completed the 4-year follow-up assessment	DSM Anxiety Panic attacks Alcohol use Lifetime AUD	M-CIDI	Baseline prevalence 4.3% presented with a lifetime history of panic attacks 15.9% presented with an AUD AUD more frequent among those with panic attack history (34.9%) than those without (15.1%; OR: 4.3, 95% CI: 2.6–7.2, p < .05) Cumulative lifetime incidence (total incident cases at baseline and both follow-up assessments) 7.2% presented with a lifetime history of panic attacks 28.5% presented with an AUD Participants with panic attack history more likely to present with an AUD at any assessment than those without (OR: 2.7, 95% CI:1.9–4.0, p < .05) Incidence at follow-up Controlling for age, gender, and all other diagnoses, onset of AUD during follow-up more frequent among those with panic attacks at baseline (21.2%) than those without (14.8%; OR: 2.4, 95% CI: 1.2–5.1, p < .05)
Hayward et al., 1997 Cross-sectional Logistic regression	N = 1,013 female adolescents (ages 11.6–16.2 years; Mage = 13.5, SD = 0.7 years; 34.5% Caucasian, 29.3% Hispanic, 14.8% Asian, 8.1% Pacific Islander, 13.3% “other”) from four suburban public middle schools near San Jose, CA	DSM Anxiety Panic attacks Panic disorder Alcohol Use Consumption over past 30 days	SCID Selected questions from the Project SMART Questionnair e (Graham et al., 1984)	1.7% met criteria for PD 3.8% met criteria for panic attacks but not PD; 50% (1.9% entire sample) reported attacks uncued After controlling for ethnicity, alcohol consumption (i.e., at least 2–4 full drinks or “being drunk”) was more frequently reported by girls with PD (24.4%) and panic attacks (31.6%) than those without a history of panic (14.1%; OR: 3.1, 95% CI: 1.5–6.6, p < .05) Alcohol use did not differ significantly between girls with cued versus uncued (including those with PD) panic attacks
Hayward et al., 1989 Cross-Sectional Frequency counts; Yates-corrected chi-square analysis	N = 95 ninth-grade students (age 14–16 years, Mage = 14.5; 48% girls; 14.7% Hispanic, 48.4% Caucasian, 20% Asian, 6.3% African American, 3.2% Native American, 5.4% “other”) in one of three physical education classes at a suburban high school in California	DSM Anxiety Panic Attacks Alcohol use Consumption history	SCID Ever used alcohol	11% experienced at least one four-symptom panic attack; 3.2% at least one limited-symptom panic attack Youth who had experienced panic attacks evidenced significantly higher symptoms of depression and were significantly more likely to have tried smoking cigarettes Prior alcohol use reported by 77% of those with panic, and 83% of those without Ever having used alcohol did not differ significantly across groups
Hayward et al., 1995 Cross-sectional Frequency counts; matched-pair t-tests	N = 160 female adolescents (Mage = 11.8, SD = 0.7; 42.5% Caucasian, 12.5% Asian-American, 45% Hispanic) drawn from a sample of girls who participated in a health curriculum intervention study in two northern California school districts	DSM Anxiety Panic attacks Alcohol use Consumption over past 30 days	SCID Selected questions from the Project SMART Questionnair e (Graham et al., 1984)	40 girls had experienced at least one lifetime panic attack 120 girls were selected for the matched comparison group; for each panic participant, three girls who had never had a panic attack were matched for age, pubertal status, and ethnicity Alcohol use did not differ across groups 18% of each group reported alcohol consumption (i.e., at least 2–4 full drinks or “being drunk”)
Ilomäka et al., 2004 Cross-sectional Frequency counts; logistic regression adjusted for age, gender, and additional diagnoses	N = 238 adolescents (ages 12–17 years; 56% girls) admitted to the psychiatric inpatient facility at Oulu University Hospital between April 2001 and July 2003	DSM Anxiety Phobic disorders (simple phobia, agoraphobia, and social phobia) Alcohol use Alcohol dependence	K-SADS	43.3% presented with a depressive disorder, 3.4% with bipolar disorder, 43.3% with conduct and oppositional defiant disorder, 29.8% with an anxiety disorder other than a phobia, 11.3% psychotic, 8.8% “other” 18.1% presented with a phobic disorder Phobic disorders more prevalent among patients with alcohol dependence (28.9%) than those without a SUD (13.1%) After controlling for age, gender, and additional diagnoses, phobic patients were significantly more likely to present with alcohol dependence (without another SUD) than non-phobic patients (OR: 4.47, 95% CI: 1.63–12.29, p < .01) The phobic disorder preceded or began at the same time as substance dependence (including alcohol) in 90.5% of comorbid cases
Kaplow et al., 2001 Prospective (4-year) Descriptives; t-tests; logistic regression	N = 936 adolescents (Mage = 10.77, SD = 1.59; 45.4% girls; 70.3% Caucasian, 6% African American, 23.7% Native American) initially interviewed at age 9, 11, or 13, and then followed for four years	DSM Anxiety Generalized anxiety symptoms Separation anxiety symptoms Alcohol use Initiation of alcohol use	CAPA Parent or child reported child drinking at least one alcoholic beverage during the follow-up period	17% of participants initiated alcohol use within the follow-up Initiation of alcohol use not significantly related to generalized anxiety [t(934) = 0.88] or separation anxiety [t(934) = −1.13] symptoms Logistic regression controlling for age, gender, race, and depressive symptoms: Elevated symptoms of generalized anxiety at baseline predicted initiation of alcohol use during follow-up (OR: 1.14, 95% CI: 1.03–1.25, p < .05) Youth with elevated separation anxiety at baseline were less likely to initiate alcohol use during follow-up (OR: 0.71, 95% CI: 0.51–0.94, p < .05)
King et al., 2004 Prospective (3-year) Frequency counts; log-linear analyses adjusted for non-independence of observations on twin pairs	N = 708 twin girls (age 11 years at baseline and 14 years at follow up) taking part in the larger Minnesota Twin Family Study	DSM Anxiety SAD Overanxious disorder Alcohol use General consumption history	DICA-R Ever used Regularly use Ever drunk Ever binge drink	By age 14 years, 29% ever drank, 9.6% regularly drank (≥ 1 drink/month over the past year), 10.7% had been drunk, 8.8% at least one binge drinking episode (≥ 6 beers, bottle of wine, or 4–5 liquor drinks in one sitting) SAD at age 11 years By age 14 years 35.6% ever drank, 12.9% regularly drank, 13.7% had been drunk, and 10.9% binge drinking Not significantly related to any drinking outcome at 14 years: Ever drank OR: 1.38, 95% CI: 0.92–2.07; Regularly drank OR: 1.32, 95% CI: 0.71–2.47; Ever drunk OR: 1.57, 95% CI: 0.96–2.58; Binge drinking OR: 1.36, 95% CI: 0.72–2.57 Overanxious disorder at age 11 years By age 14 years 30.2% ever drank, 11.1% regularly drank, 9.3% had been drunk, and 9.3% binge drinking Not significantly related to any drinking outcome at 14 years: Ever drank OR: 0.82, 95% CI: 0.45–1.49; Regularly drank OR: 0.81, 95% CI: 0.81–3.60; Ever drunk OR: 0.99, 95% CI: 0.64–1.54; Binge drinking OR: 1.00, 95% CI: 0.42–2.41
Neighbors et al., 1992 Cross-Sectional Frequency counts; chi-square analysis; descriptives; ANOVA	N = 111 youth (ages 12.7–18.7 years, Mage = 15.11 years; 47.7% Caucasian, 52.3% African American) currently incarcerated in detention facilities in the southeast US and flagged at intake for potential mild-severe mental health problems	DSM Anxiety Overanxious disorder Alcohol use Current AUD	DISC-2	n = 59 youth with no AUD/SUD n = 23 youth with an AUD (alone or with comorbid marijuana abuse/dependence) n = 29 youth with an illicit SUD (polysubstance users; most with a comorbid AUD) Among youth diagnosed with overanxious disorder, 17% presented with no AUD/SUD, 22% with an AUD, and 38% polysubstance use/SUD Chi-square analysis evidenced no significant differences across groups (p < .09) Significantly more AUD (96%) and polysubstance (93%) youth diagnosed with conduct disorder as compared to no SUD group (62%) As compared to no SUD, AUD youth evidenced significantly higher symptoms of conduct disorder, and polysubstance youth significantly higher symptoms of both conduct disorder and depression Symptoms of overanxious disorder increased (M = 1.83, 2.14, and 2.84) but did not vary significantly across groups
Nelson et al., 2000 Cross-Sectional Frequency counts; t-tests; logistic regression adjusted for non-independence of observations on twin pairs	N = 1344 female twins (ages 16–19.5 years; Mage = 18.2 years) drawn from the Missouri Adolescent Female Twin Study	DSM Anxiety Social phobia Alcohol use Lifetime AUD	DICA	16.3% met criteria for lifetime social phobia; first social fear Mage of onset = 10.8 years Frequency of AUD higher among those with (17.8%) than without (9.3%) social phobia Controlling for age, social phobia (with or without comorbid depression) significantly predicted an AUD (OR: 2.07, 95% CI: 1.38–3.12) Controlling for age, social phobia (without comorbid depression) significantly predicted any AUD (OR: 1.73, 95% CI: 1.03–2.91) Among those with comorbid social phobia and an AUD, age of first social fear (Mage = 10.6, SD = 3.49) predated first alcohol intoxication (Mage = 15.4, SD = 1.85; t = 8.21, p < .01)
Zimmerman et al., 2003 Prospective (4-year) Frequency counts; logistic regression controlling for age and gender; prospective logistic regression controlling for age, gender, other anxiety disorders, depression, dysthymia, bipolar disorders, illegal drug use disorders, and antisocial behavior	N = 3021 youth (14–24 years) drawn from a government registry of residents in Munich n = 2548 completed the 4-year follow-up assessment	DSM Anxiety Panic attacks Panic disorder Agoraphobia w/o panic disorder Social phobia Specific phobia Phobia NOS GAD Alcohol use Regular use Hazardous use Alcohol dependence and abuse	M-CIDI	Baseline prevalence Lifetime: 46.4% reported occasional alcohol use, 10% regular use (3 times/week during peak use), 8.4% hazardous use [40 g/day (men) or 20 g/day (women) during peak use], 15.9% any AUD, 9.7% alcohol abuse (AA), and 6.2% alcohol dependence (AD) 12-month: 45.3% occasional use, 9.2% regular use, 3.9% hazardous use, 10.1% any AUD, 5.3% AA, and 4.7% AD Lifetime: 26.9% any anxiety disorder, 4.3% panic attacks, 1.6% PD, 2.3% agoraphobia, 7.3% social phobia, 16.2% specific phobia, 5.2% phobia NOS, and 3.4% GAD 12-month: 18.2% any anxiety disorder, 2.7% panic attacks, 1.2% PD, 1.4% agoraphobia, 5.2% social phobia, 10.9% specific phobia, 2.7% phobia NOS, and 1.7% GAD Significant baseline associations Regular use: Panic attacks (OR: 4.2, 95% CI: 2.2–7.7), and agoraphobia (OR: 2.2, 95% CI: 1.0–4.8) Hazardous use: Panic attacks (OR: 3.5, 95% CI: 1.7–6.9), agoraphobia (OR: 5.0, 95% CI: 2.3–10.9), specific phobia (OR: 1.9, 95% CI: 1.3–2.9), and GAD (OR: 2.2, 95% CI: 1.1–4.4) Any AUD: Panic attacks (OR: 4.4, 95% CI: 2.4–8.0), PD (OR: 4.1, 95% CI: 1.6–10.1), agoraphobia (OR: 3.3, 95% CI: 1.6–6.7), and specific phobia (OR: 2.0, 95% CI: 1.4–2.8) AA: Panic attacks (OR: 2.6, 95% CI: 1.2–5.5), and specific phobia (OR: 1.6, 95% CI: 1.1–2.4) AD: Panic attacks (OR: 6.2, 95% CI: 3.0–12.6), PD (OR: 7.1, 95% CI: 2.4–20.5), agoraphobia (OR: 5.0, 95% CI: 1.9–12.8), social phobia (OR: 2.0, 95% CI: 1.1–3.5) and specific phobia (OR: 2.5, 95% CI: 1.6–3.8) Significant baseline anxiety and onset/progression of use at follow-up Regular use: Social phobia (OR: 1.9, 95% CI: 1.0–3.4) Hazardous use: Panic attacks (OR: 2.5, 95% CI: 1.1–5.8) and social phobia (OR: 2.1, 95% CI: 1.2–3.8) AA: Panic attacks (OR: 2.7, 95% CI: 1.1–6.1) Significant baseline anxiety and persistence of use at follow-up Any AUD: Panic attacks (OR: 18.7, 95% CI: 4.7–73.6) and PD (OR: 13.2, 95% CI: 1.4–124.0) AD: Social phobia (OR: 19.7, 95% CI: 1.1–352.2)

AADIS: Adolescent Alcohol and Drug Involvement Scale (Moberg, 2000). ANOVA: Analysis of variance. AUD: Alcohol use disorder (includes both abuse and dependence). CAPA: Child and Adolescent Psychiatric Assessment (Angold & Costello, 1995). CI: Confidence interval. DICA: Diagnostic Interview for Children and Adolescents (Reich, 1996). DICA-R: Diagnostic Interview for Children and Adolescents-Revised (Reich & Welner, 1988). DISC-2: Diagnostic Interview for Children (Shaffer, Fisher, Piacentini, Schwab-Stone, & Wicks, 1989). DMQ-R: Drinking Motives Questionnaire-Revised (Cooper, 1994). GAD: Generalized anxiety disorder. HR: Hazard ratio. K-SADS: Schedule for Affective Disorders and Schizophrenia for School Age Children (unpublished manuscript, Puig-Antich and associates, 1981). LHA Interview: Lifetime History of Alcohol and Drug Use Interview (Skinner, 1982). M-CIDI: Munich Composite International Diagnostic Interview (Wittchen, Lachner, Wunderlich , & Pfister, 1998). MDD: Major Depressive Disorder. NOS: Not otherwise specified. OCD: Obsessive-Compulsive Disorder. OR: Odds ratio. PD: Panic Disorder. PHQ: Patient Health Questionnaire (Spitzer, Kroenke, & Williams, 1999). PTSD: Posttraumatic Stress Disorder. RCADS-SP: Revised Child Anxiety and Depression Scale-Social phobia subscale (Chorpita et al., 2000). SAD: Separation anxiety disorder. SCID: Structured Clinical Interview for DSM-III-R (Spitzer, Williams, & Gibbon, 1987). SUD: Substance use disorder (includes both abuse and dependence).