Clinical Psychology Review, Vol. 20, No. 2, pp. 149–171, 2000 Copyright © 2000 Elsevier Science Ltd. Printed in the USA. All rights reserved 0272-7358/00/$–see front matter PII S0272-7358(99)00027-6 THE RELATIONSHIP BETWEEN ANXIETY DISORDERS AND ALCOHOL USE DISORDERS: A REVIEW OF MAJOR PERSPECTIVES AND FINDINGS Matt G. Kushner, Kenneth Abrams, and Carrie Borchardt University of Minnesota ABSTRACT. It is generally agreed that problems related to alcohol use and anxiety tend to occur within the same individual (“comorbidity”); however, the cause of this association remains controversial. Three prominent perspectives are that anxiety disorder promotes pathological alcohol use, that pathological alcohol use promotes anxiety disorder and that a third factor promotes both conditions. We review laboratory, clinical, family, and prospective studies bearing on the validity of these explanatory models. Findings converge on the conclusion that anxiety disorder and alcohol disorder can both serve to initiate the other, especially in cases of alcohol dependence versus alcohol abuse alone. Further, evidence from clinical studies suggests that anxiety disorder can contribute to the maintenance of and relapse to pathological alcohol use. Relying heavily on pharmacological and behavioral laboratory findings, we tentatively propose that short-term anxiety reduction from alcohol use, in concert with longer-term anxiety induction from chronic drinking and withdrawal, can initiate a vicious feed-forward cycle of increasing anxiety symptoms and alcohol use that results in comorbidity. © 2000 Elsevier Science Ltd. KEY WORDS. Comorbidity, Dual diagnosis, Anxiety disorder, Alcohol use disorder. STUDIES INTO THE association between anxiety disorders and substance use disorders, especially those involving alcohol use, have been reviewed with surprising frequency (e.g., Allan, 1995; Brady & Lydiard, 1993; Cowley, 1992; Cox, Norton, Swinson, & Endler, 1990; Kushner, Sher, & Beitman, 1990; Marshall, 1997; Schuckit, 1996; Schuckit & Hesselbrock, 1994; Stockwell & Bolderston, 1987; Wilson, 1988). This decade-long string of nearly annual reviews could be reflective of a dynamic area in which new knowledge is accumulating rapidly. However, we suggest that it is more reflective of competing efforts at organizing much the same literature from a handful of Correspondence should be addressed to Matt G. Kushner, Department of Psychiatry, University of Minnesota, F282/2A West Building, 2450 Riverside Avenue, Minneapolis, MN 55454. 149 150 M. G. Kushner, K. Abrams, and C. Borchardt differing perspectives. In the end, no consensus has been achieved as to the picture painted by the totality of this literature (e.g., Kushner, 1996; Schuckit & Hesselbrock, 1996). Rather than writing yet another review from a particular perspective, it was our aim to explicate and evaluate the multiple perspectives/hypotheses that have been applied to the phenomenon of “comorbidity” (cases in which a single individual experiences both an anxiety disorder and a substance use disorder). In this regard, we emphasize theoretical and conceptual issues over those that are of primarily clinical interest. More specifically, we include studies using experimental or prospective methodologies that bear on the validity of ostensibly competing views of comorbidity. We also include family studies capable of addressing specific theoretical predictions. We focus predominantly on comorbidity involving alcohol (vs. other drug) use disorders. This reflects the fact that the vast majority of relevant studies focus on alcohol use (e.g., most of the aforementioned reviews concern the relationship between anxiety disorders and alcoholism). Furthermore, the many unique pharmacological, psychological, and sociological aspects of alcohol use/abuse on the one hand, versus cocaine, marijuana, or hallucinogen use/abuse on the other, could easily justify separate review efforts. Nonetheless, we do attempt to state general research strategies and problems in terms that apply to substance use disorder (SUD) more broadly and we also include studies involving drugs other than alcohol where applicable. Finally, we do not include literature focused primarily on comorbidity involving posttraumatic stress disorder (PTSD), an area we believe to be sufficiently unique as to warrant a separate review effort (e.g., see the review by Stewart, 1996). The first major section of this article briefly overviews studies estimating the prevalence of comorbidity. Next we outline the major hypotheses accounting for this association. Against this background, we review: (a) laboratory studies looking at the interaction between anxiety symptoms and alcohol use; (b) clinical studies looking at the effect of one comorbid disorder upon the course of the other; (c) family studies looking at psychopathology in the families of probands with one or both comorbid disorders; and, (d) prospective studies looking at the longitudinal relationship of comorbid disorders. The concluding section aims at integrating the findings reviewed into a coherent picture of the processes contributing to the etiology and maintenance of comorbidity. THE PREVALENCE OF COMORBIDITY Complexities in Estimating the Prevalence of Comorbidity Defining comorbidity. There exists no consensus as to the optimal way to operationalize comorbidity. The simplest and most often used approach counts as comorbid those cases in which both substance use and anxiety symptoms have been deemed clinically significant (typically defined as meeting criteria from the Diagnostic and Statistical Manual of Mental Disorders) at some time (but not necessarily at the same time) over the life of the individual (e.g., Kessler et al., 1997; Regier et al., 1990). This broad definitional approach makes few assumptions as to the etiology of comorbidity and, therefore, would appear to be an appropriate starting point for evaluating a clinical construct for which the etiology is not well-understood. On the other hand, embracing a specific theory as to the origins of comorbidity could suggest more The Relationship Between Anxiety Disorders and Alcohol Use Disorders 151 restrictive definitions for this phenomenon. For example, the assumption that the symptoms of one of the disorders directly causes those of the other might lead one to count as comorbid only those cases in which the putative causal disorder manifests first and is actively symptomatic at the onset of the second (caused) disorder. In this regard, the most specific and restrictive criteria for operationalizing comorbidity have been put forward by Schuckit and Hesselbrock (1994). They recommended that, beyond having significant problems with both anxiety and alcohol use, the following conditions must also hold: (a) alcohol problems must be formally diagnosed as alcohol dependence; (b) the onset of the anxiety disorder must precede that of the alcohol dependence; (c) a third diagnosis that might account for the presence of both disorders must not be present; and, (d) assortative mating (e.g., a history of anxiety disorder for either parent) must not be evident. However, because a clear understanding of the etiology of comorbidity is yet lacking (see this review), it strikes us as premature to implement theory-based restrictions and exclusions in defining this phenomenon. Diagnostic specificity. Comorbidity rates are also likely to be affected by the specific anxiety disorder and SUD under consideration. It is convenient and conventional to refer to comorbidity involving SUD and anxiety disorder in very general terms. However, there is no a priori reason to expect that various drugs and alcohol will share the same relationship to each anxiety disorder. In fact, the substantial differences (e.g., base-rate, treatment response, course, etc.) between the various anxiety disorders and between the various SUDs (including alcohol abuse vs. dependence), suggest the likely value of diagnostic specificity in considering comorbidity. In this regard, past reviews have documented significant differences in comorbidity rates among various anxiety disorders and substance use disorders (e.g., Cox et al., 1990; Kushner et al., 1990). Population specificity. Comorbidity rates are also likely to be affected by the population sampled. Individuals with multiple disorders are more likely than those with one disorder to be self- or other-referred to treatment (Berkson, 1949). Further, referral patterns may not result in a random distribution of comorbid cases across various treatment settings. For example, clinicians may be inclined to recommend alcoholism treatment, versus anxiety disorder treatment, as the priority for their comorbid patients. Consistent with these observations, comorbidity rates tend to be greater in clinical versus community samples and in alcoholism treatment versus psychiatric treatment samples (Cox et al., 1990; Kushner et al., 1990). Estimating the Prevalence of Comorbidity This discussion highlights advantages to estimating comorbidity prevalence based on representative community surveys that employ maximum diagnostic specificity and minimum exclusions in defining comorbidity.1 With this in mind, we examined find1The fact that comorbidity prevalence rates in clinical samples provide biased estimates of overall comorbidity rates does not detract from the potential clinical relevance of these estimates (e.g., to help project treatment and aftercare needs of a given clinical population). The reader interested in summaries of studies examining comorbidity rates in clinical samples is referred to any of the reviews on this topic cited above. 152 M. G. Kushner, K. Abrams, and C. Borchardt ings from the Epidemiological Catchment Area (ECA) survey (e.g., Regier et al., 1990) and the National Comorbidity Survey (NCS) (e.g., Kessler et al., 1997) in order to estimate the prevalence of comorbidity. The ECA survey. Regier et al. (1990) reported that in the ECA survey, those with “any anxiety disorder,” as compared to the rest of the sample, had a 50% increase in the odds of being diagnosed with an alcohol disorder (odds ratio [OR] 5 1.52).2 Similarly, the anxiety-disordered subsample had about a 70% increase in the odds of being diagnosed with any substance use disorder (OR 5 1.7). Further, because both anxiety disorders and substance use disorders are among the most common psychiatric conditions, these odds ratios indicate that the real number of comorbid cases is relatively high. Notably, panic disorder presented an even greater increase in the odds of being diagnosed with alcohol or drug dependence (OR 5 3.3 and 4.4, respectively) than was true for the broader diagnostic categories discussed above (Regier et al., 1990). The same pattern held for the ECA subsample with obsessive compulsive disorder (OCD) in which the odds ratios associated with alcohol and drug dependence were 2.5 and 3.6, respectively (Regier et al., 1990). Interestingly, panic and OCD were not significantly related to alcohol or drug abuse in the absence of dependence. Unlike Regier et al. (1990), Himle and Hill (1991) analyzed comorbidity prevalence in the ECA for the separate phobic disorders. They reported an approximate 2- to 2.5-fold increase in the likelihood of being diagnosed with alcohol abuse or dependence among those with agoraphobia and social phobia, respectively; however, they did not find an increased risk for alcohol abuse or dependence among those with simple phobia. The NCS survey. Some methodological differences not withstanding, the NCS (e.g., Kessler et al., 1997) produced findings regarding comorbidity that are reasonably consistent with those of the ECA survey. For example, the NCS also found that the odds of being diagnosed with alcohol dependence nearly tripled in women with panic disorder (OR 5 2.9) and more than doubled in men with panic disorder (OR 5 2.3); again, having these anxiety disorders did not change significantly one’s odds of being diagnosed with alcohol abuse in the absence of dependence. Risk for alcohol dependence in the NCS was also significantly elevated among both men and women with social phobia, simple phobia and agoraphobia (ORs range from 1.8 for agoraphobia in men to 3.1 for simple phobia in women). Generalized anxiety disorder (GAD) was also associated with a significant increase in the odds of being diagnosed with alcohol dependence for both men and women (3.9 and 3.0, respectively). Finally, as was the case in the ECA, significant comorbid associations were 2Odds ratio is the ratio of the odds of having the comorbid diagnosis when the index disorder is present to those of having the comorbid disorder when the index disorder is absent. For example, an odds ratio of 1 means that the odds of being diagnosed with the comorbid diagnosis is the same whether the index diagnosis is or is not present. An odds ratio of two means the odds of being diagnosed with the comorbid diagnosis is doubled when the index diagnosis is present, and so on. The reader should also note that when discussing odds ratios, no causal relationship is implied. Therefore, no implication as to causal priority of one comorbid disorder over the other should be construed based on the wording used to describe conditional probabilities/odds. The Relationship Between Anxiety Disorders and Alcohol Use Disorders 153 found mainly in cases involving alcohol dependence versus alcohol abuse alone. Exceptions to this include the significant positive relationship of simple and social phobia to alcohol abuse without dependence among females. Summary. Both the ECA and the NCS provide evidence of an approximate doubling to quadrupling of risk for alcohol or drug dependence given the presence of most anxiety disorders; however, simple phobia was not associated with alcohol disorders in the ECA survey but was in the NCS. It was particularly notable that alcohol and drug problems involving dependence, versus abuse alone, were the most strongly associated with anxiety problems. Exceptions to this were found among women of the NCS with alcohol abuse who also showed an increase in risk for simple and social phobia. These prevalence data suggest that all or most anxiety disorders have a significant relationship to alcohol disorder; however, they do not clarify the nature of these relationships. ETIOLOGICAL MODELS OF COMORBIDITY Causal Explanations Anxiety disorder promotes alcoholism. One causal explanation for the etiology of comorbidity is that anxiety disorder (or anxiety symptoms per se) serves to promote directly the pathological use of alcohol or drugs. The primary hypothesis fitting within this general view, sometimes referred to as the “self-medication hypothesis” (e.g., Quitkin, Rifkin, Kaplan, & Klein, 1972), suggests that the pharmacological and/or psychological effects of alcohol and some drugs serve to decrease aversive anxiety symptoms, thereby promoting persistent and escalating use via negative reinforcement. This view is most commonly espoused by anxiety researchers focusing on the comorbidity problem (e.g., Brady & Lydiard, 1993; Cox et al., 1990; Cowley, 1992; Kushner et al., 1990). Alcoholism promotes anxiety disorder. Another causal explanation for the etiology of comorbidity is that pathological alcohol or drug use serves to promote directly the development of anxiety disorder. This general view is most commonly associated with the hypothesis that significant anxiety symptoms are a bio-psycho-social consequence of chronic substance use and/or the withdrawal syndrome (e.g., George, Nutt, Dwyer, & Linnoila, 1990). This view is most commonly espoused by alcoholism researchers focusing on the comorbidity problem (e.g., Allan, 1995; George et al., 1990; Schuckit, 1996; Schuckit & Hesselbrock, 1994). Shared Etiologic Explanation Rather than a direct causal relationship, an alternative view suggests that comorbid conditions may both be promoted by a shared third factor. Stemming, in part, from the strong familial component observed for both anxiety disorder and alcoholism (e.g., Crowe, Noyes, Pauls, & Slymen, 1983; McGue, 1994; Merikangas & Gelernter, 1990; Noyes, Clancy, Crowe, Hoenk, & Slymen, 1978), some researchers have suggested that this causal third factor could be transmitted within the family (e.g., Merikangas, Stevens, & Fenton, 1996). Merikangas et al. (1996) suggested that such a factor might be “. . . genetic factors predisposing to both types of disorders, biological environmental risk factors, non-biological environmental factors (e.g., a disruptive family 154 M. G. Kushner, K. Abrams, and C. Borchardt environment or parental abuse or neglect), or exposure to prenatal environmental factors (e.g., maternal alcohol use)” (p. 103). As this quote suggests, the shared etiologic explanation of comorbidity is the most difficult view to fit with specific hypotheses because the critical (causal) third variable and, hence, any parameters as to its mechanism(s) of influence, are left unspecified. Possible Complexities It is important to keep in mind that a clear description of the processes underlying comorbidity may entail complex variations of the basic etiological explanations outlined above. First, processes responsible for the initiation of comorbidity may well differ from those responsible for the maintenance of this status. For example, an anxiety disorder might contribute to the initiation of and relapse to pathological substance use, whereas other processes could serve to maintain ongoing pathological substance use. Secondly, etiologic processes in comorbidity may be heterogeneous between individuals. In this regard, various populations (e.g., differing treatment settings or genders) may be more or less likely to include comorbid cases of varying etiologies. Lastly, etiologic processes in comorbidity may overlap and/or interact. For example, several of the models described above could interact synergistically to form a hybrid etiological process promoting comorbidity. The fact that complex models might apply to the etiology and maintenance of comorbidity could help to explain the proliferation of multiple perspectives on the nature of this phenomenon. This possibility evokes the classic example of several blind individuals each touching different parts of the same elephant—they can accurately understand the whole of what they are exploring only by sharing and integrating their separate perspectives. Similarly, differing perspectives and levels of analysis applied to the problem of comorbidity might be more informative when considered together. With this in mind, we turn next to a review of empirical findings from studies of comorbidity that span the laboratory, the clinic, the community and the family. STUDIES BEARING ON THE ETIOLOGY AND MAINTENANCE OF COMORBIDITY Laboratory Studies As noted above, the idea that some aspect of one disorder serves to promote the onset of a second disorder is a conspicuous possibility when considering the etiology of comorbidity. In fact, high percentages of comorbid patients report that anxiety symptoms cue alcohol use aimed at relief from those symptoms (e.g., Bibb & Chambless, 1986; Chambless, Cherney, Caputo, & Rheinstein, 1987; Cox, Norton, Dorward, & Fergusson, 1989: Quitkin et al., 1972; Smail, Stockwell, Canter, & Hodgson, 1984). Paradoxically, however, comorbid patients also report that periods of heavy drinking are associated with worsening of phobias (e.g., Stockwell, Smail, Hodgson, & Canter, 1984). These studies are suggestive of complex causal links between anxiety symptoms and alcohol use; however, the accuracy of retrospective self-reporting in comorbidity has been called into question (e.g., Kushner et al., 1990; Schuckit & Hesselbrock, 1994). An alternative approach is to examine this relationship in the laboratory. Direct examination of the psychopharmacology of alcohol and other drugs, in conjunction with knowledge of the biology and psychology of anxiety, might provide important The Relationship Between Anxiety Disorders and Alcohol Use Disorders 155 clues as to the effects of alcohol use and anxiety symptoms on each other. With this in mind, we review relevant laboratory studies while attending to potentially important caveats related to their generalizability. Alcohol use and anxiety reduction. Based upon their acute neurochemical effects, drugs such as alcohol and the opiates might be expected to temporarily relieve panic and other anxiety states. For example, acute alcohol intoxication has been shown to affect gamma-aminobutyric acid (GABA) receptors in a way that is similar to that of the antianxiety/panic class of drugs, benzodiazepine (BZ) (e.g., Liljequist & Engel, 1984; Nestoros, 1980). Further supporting the overlap of effect for alcohol and BZ drugs is the finding of cross-tolerance for these substances (Eriksson et al., 1989). Also consistent with a specific anti-panic effect for acute opiate and alcohol intoxication are studies showing that these states are associated with an inhibition of norepinephrine (NE) systems in the brain (e.g., Gold, Pottash, Sweeney, Kleber, & Redmond, 1979; Gold, Redmond, & Kleber, 1979; Meltzer, 1987). Cognitive-based theories also predict anxiety reduction during acute intoxication from alcohol (e.g., Wilson, 1987). For example, it has been proposed that alcohol lowers anxiety by diverting attention from anxiogenic cues (e.g., Steele & Josephs, 1988), as well as by undermining cognitive processes that make anxiogenic cues seem self-relevant (e.g., Hull, 1981). Social learning theory predicts anxiety reduction from substance use through effects on self-efficacy (e.g., Bandura, 1985). For example, when alcohol is expected to aid in coping with a given situation or feeling, then drinking should increase self-efficacy thereby decreasing anxiety level (e.g., Marlatt, 1984). It has also been proposed that alcohol use may lower anxiety by providing an externalized excuse for anticipated failures (“self-handicapping”) (e.g., Jones & Berglas, 1978; Tucker, Vuchinich, & Sobell, 1981). Many dozens of studies have evaluated whether alcohol modulates stress responding in the laboratory (e.g., see excellent reviews by Cappell & Greeley, 1987; Pohorecky, 1991; Sher, 1987). These reviews agree that whether alcohol dampens stress-responding depends on a rather large array of design variables (e.g., intensity of stressor, type of stressor, temporal relationship of drinking to the stressor, and dose). Also of apparent importance to alcohol’s stress-response dampening effects are a number of subject variables (e.g., typical drinking patterns, personality, physical fitness, gender, expectancies for alcohol effects, and family history of alcoholism). Within the context of parameters such as these, studies have routinely found stress-response dampening effects associated with acute alcohol intoxication (Cappell & Greeley, 1987; Pohorecky, 1991; Sher, 1987). Unlike the large number of studies into alcohol’s effects on stress-responding, only a small number of studies have examined the effects of alcohol on clinical anxiety phenomena. Kushner et al. (1996) recruited nonalcoholic individuals with panic disorder to determine the acute impact of alcohol on CO2 challenge-induced panic symptoms. They found that, relative to a placebo, alcohol significantly decreased panic response to the CO2 challenge, as well as state anxiety before and after the challenge. Using a similar approach, Abrams, Kushner, Lisdahl, and Voight (1997) found that, relative to both a placebo and a control condition, alcohol reduced the performance anxiety of socially phobic subjects during a public speaking challenge. Both of these studies pointed toward a pharmacological mechanism underlying the effects obtained. Four studies have examined alcohol’s acute impact on phobic reactions. Using snakes or mice as the phobic stimulus, two studies found alcohol reduced self-reported 156 M. G. Kushner, K. Abrams, and C. Borchardt anxiety and behavioral avoidance (Alexanderson & Lindman, 1980; Lindman, Alexanderson, & Lindfors, 1980), one study found alcohol reduced self-reported anxiety but not approach behavior (Rimm, Briddell, Zimmerman, & Caddy, 1981) and one study found no alcohol effects on anxiety or behavior (Thyer & Curtis, 1984). Notably, positive alcohol effects in these studies were, once again, attributed to pharmacological rather than psychological mechanisms. Alcohol use and anxiety induction. Although increases in GABA-ergic activity following acute alcohol administration support an anxiolytic effect for alcohol (above), decreased GABA tone stemming from chronic alcohol use and withdrawal may serve to generate anxiety. For example, lower plasma levels of GABA, as well as increased GABA receptors in the brain (postmortem), have been noted among chronic alcoholics (Coffman & Petty, 1985; Tran, Snyder, Major, & Hawley, 1981). The capacity of such a GABA deficiency to induce anxiety is evidenced by the anxiogenic qualities of BZ inverse agonists and GABA antagonists (e.g., Dorow, Horowski, Paschelke, & Amin, 1983; Pellow, 1980). Similarly, while alcohol and the opiates serve to reduce acutely the activity of NE systems in the brain, studies have documented increased NE activity during alcohol withdrawal (e.g., Borg, Kvande, & Sedvall, 1981; Hawley, Major, Schulman, & Linnoila, 1985). Importantly, overly active NE systems in the brain have been implicated in the pathophysiology of panic (e.g., Redmond, 1985). These findings, then, provide a hypothetical, albeit straightforward, causal link between alcoholism and the genesis of anxiety and panic mediated through modifications in neurotransmitter activity associated with chronic alcohol use and withdrawal. The psycho-social consequences of pathological alcohol use might also promote anxiety disorder. For example, extinction of fear responses can be significantly impeded by intoxication from alcohol (Cameron, Liepman, Curtis, & Thyer, 1987), barbiturates (Chambless, Foa, Groves, & Goldstein, 1979) and BZs (e.g., Bouton, Kenney, & Rosengard, 1990; Marks et al., 1993). Therefore, chronic alcohol or drug use might contribute to the risk for anxiety disorder by impeding naturally occurring desensitization. Notably, this view suggests that alcoholism promotes anxiety disorder by inhibiting mechanisms that normally govern the persistence and escalation of otherwise normal anxiety reactions (e.g., hetero-social fears in adolescents). Social and occupational disruptions resulting from pathologic alcohol or drug use may also trigger anxiety symptoms and disorders in vulnerable individuals (e.g., Kushner et al., 1990). For example, the loss of one’s job, divorce and involuntary commitment to treatment are all high-order stressors that are more likely to be experienced by alcoholics than nonalcoholics. Also, attention-narrowing from alcohol, although described above as a possible mechanism by which drinking exerts anxiolytic effects, is thought to promote anxiety when used in environments rich in fear provoking cues (Steele & Josephs, 1988). That is, alcoholics may, under certain circumstances, focus more attention on life’s problems and anxiogenic cues than do nonalcoholics. Consistent with these biological and psychological links between alcoholism and the development of anxiety problems, studies do suggest that protracted drinking episodes as well as drinking very high doses of alcohol may worsen anxiety in non-comorbid alcoholics (e.g., Freed, 1978; Mello & Mendelson, 1978; Peyser, 1982; Stockwell, Hodgson, & Rankin, 1982). For example, Stockwell et al. (1982) evaluated the impact on mood and physiological responses of a high dose of alcohol (6 gm/kg) consumed over several drinking sessions during a 2-day period. Interestingly, they reported that The Relationship Between Anxiety Disorders and Alcohol Use Disorders 157 there was temporary improvement in affective state after each drinking session but an overall increase in anxiety throughout the 2-day drinking period. However, we are not aware of any studies documenting the onset of anxiety disorders, per se, as a result of pathological alcohol use. Anxiety effects on drinking behavior. Like the literature on stress response dampening, many studies have examined the impact of stressors on self-administration of alcohol in non-anxiety-disordered subjects. Sher (1987) reviewed these studies and concluded: “Stress-related drinking is a function of a number of psychological and social factors, especially the ability to cope effectively with the stressor without alcohol and the potentially negative consequences of alcohol consumption” (p. 255). Again, however, the generalizability of laboratory alcohol self-administration studies to the natural drinking behavior of individuals with anxiety disorders is questionable. Kushner et al. (1997) had subjects with panic disorder inhale either 35% CO2 (panic challenge) or room air (control challenge) both before and after a 30-minute alcohol self-administration period. The primary analysis showed that challenge group assignment did not significantly impact upon the amount of alcohol subjects consumed. However, a secondary analysis showed that those most strongly expecting panic at the post-drinking challenge did self-administered more alcohol than other subjects; especially at those choice points closest to the second challenge. Once again, the generalizability of these findings to anxiety-induced drinking behavior outside of the laboratory was not directly addressed by this study. Summary. Retrospective self-report studies and experimental studies are each subject to their own set of limitations. Notably, however, both approaches point to the shortterm anxiolytic effects of alcohol and the long-term anxiogenic effects of alcohol. Further, psychological and biological mechanisms associated with the use and withdrawal from alcohol and opiates are highly consistent with this pattern of effects. Alcohol use in response to anxiety is thought to be influenced by a confluence of factors. Because of this complexity, behavioral laboratory studies are potentially limited in terms of external validity. This caveat notwithstanding, limited laboratory evidence suggests that anxiety can promote alcohol use under some circumstances. Clinical Studies Clinical studies can provide information relevant to etiological processes in comorbidity. For example, if anxiety disorder promotes pathological alcohol use then active anxiety symptoms should increase risk for relapse following alcoholism treatment. Alternatively, if pathological alcohol use promotes anxiety symptoms then successful alcoholism treatment should reduce or eliminate anxiety symptoms. Below we selectively review clinical studies of comorbidity that help to address these and related issues. Effects of successful anxiety treatment on drinking behavior.3 Successful treatment of anxiety symptoms with buspirone appears to reduce alcohol use among comorbid individ3It is important to note that the focus of this section is the impact of successful anxiety treatment on substance use behavior. Therefore, we did not include studies that failed to find anti- 158 M. G. Kushner, K. Abrams, and C. Borchardt uals. Tollefson, Montague-Clouse, and Tollefson (1992) administered buspirone or a placebo to 51 individuals with both GAD (diagnosed at least 30 days after detoxification) and alcohol abuse/dependency over a 24-week study period. They reported that buspirone was associated with a reduction in number of days desiring alcohol and an overall improvement in clinical global assessment. Kranzler et al. (1994) administered buspirone or a placebo to 61 anxious alcoholics, all of whom were also receiving psychotherapy aimed at relapse prevention. They noted nonsignificant trends toward less alcohol use in the buspirone group at the end of treatment, and significantly lower alcohol use in the buspirone group at the 6-month follow-up assessment. The conclusion that buspirone reduced drinking by reducing anxiety is called into question by the possibility that such drugs have a direct effect on craving for alcohol (e.g., Tollefson et al., 1992). However, this alternative interpretation could not be applied to the results of the study reported by Fals-Stewart and Schafer (1992). In this study, 60 patients with both substance abuse and OCD were given either: (a) substance abuse treatment plus a behavioral treatment for OCD; (b) substance abuse treatment plus relaxation training (as an “attention control”); or (c) substance abuse treatment only. They reported that those who received the OCD treatment showed the greatest reduction in OCD symptoms and abstinence rates at a 12-month follow-up. Effects of persistent anxiety disorder symptoms on alcoholism treatment outcome. It appears that general psychopathology predicts a poorer outcome in response to treatment for pathological use of alcohol, cocaine, and opiates (e.g., Carroll, Power, Bryant, & Rounsaville, 1993; Ojehagen, Berglund, Appel, Nilsson, & Skjaerris, 1990; Rounsaville, Dolinsky, Babor, & Meyer, 1987; Rounsaville, Kosten, Weissman, & Kleber, 1986). Also, stressful events following alcoholism treatment appear to promote relapse (e.g., Billing & Moos, 1983; Brown, Vik, Patterson, Grant, & Schuckit, 1995; Brown et al., 1990; Rosenberg, 1983). Referring to anxiety disorders, per se, LaBounty, Hatsukami, Morgan, and Nelson (1992) reported 6-month posttreatment relapse rates and reasons for relapse in a group of 35 alcoholics with symptoms of phobia and/or panic and in a group of alcoholics without comorbid anxiety disorder. They found that although relapse rates were similar between the groups, significantly more anxious subjects reported relapsing to cope with aversive emotional symptoms. Tomasson and Vaglum (1996) looked at the association between anxiety disorder and alcohol use disorder relapse in a nationwide sample of inpatient alcoholics. They reported that for men, panic disorder predicted a greater likelihood of relapse following treatment. Further, women who were abstinent at follow-up reported higher levels of psychiatric distress than did other study groups. Effect of successful alcoholism treatment on comorbid anxiety. Not surprisingly, state anxiety tends to decrease from the beginning to the end of inpatient alcoholism treatment anxiety effects for a given intervention. (For example, we did not include studies that have failed to find an anti-anxiety effect for buspirone in comorbid alcoholics, such as that reported by Malcolm et al., 1992.) Although this is consistent with our aim of addressing theoretical versus clinical issues related to comorbidity, studies showing no effect for a presumptive anxiety treatment in comorbid samples are obviously relevant to the issue of clinical management of comorbid patients. The Relationship Between Anxiety Disorders and Alcohol Use Disorders 159 (Brown, Irwin, & Schuckit, 1991). Also reinforcing common sense expectations, studies have shown that, compared to non-comorbid alcoholics, state anxiety is higher during and following treatment among comorbid alcoholics (Brown et al., 1991; Thevos et al., 1991). Although several studies have examined the stability of lifetime anxiety disorder diagnoses that were first made during alcoholism treatment (Penick, Powell, Liskow, Jackson, & Nickel, 1988; Rounsaville, Cacciola, Weissman, & Kleber, 1981), no study to date has satisfactorily examined the effect of alcoholism treatment on the status of comorbid anxiety disorders. Clearly, this will be an important issue for future research. Summary. Clinical studies address the validity of some theoretical predictions concerning comorbidity. We reviewed findings showing that reducing clinical anxiety symptoms improves alcoholism treatment outcome and that ongoing problems with panic and anxiety predict relapse. These findings are consistent with the idea that anxiety disorder promotes continued substance use among comorbid individuals. The converse question—how successful alcoholism treatment effects anxiety disorder status—remains unresolved at this time. Family Studies As noted earlier, it is well-established that risk for anxiety disorder as well as for alcoholism is, in part, transmitted within the family (e.g., McGue, 1994; Noyes et al., 1978). Further, several studies have shown that the family members of individuals with anxiety disorders are at increased risk for alcoholism (e.g., Cloninger, Martin, Clayton, & Guze, 1981; Crowe et al., 1983; Noyes et al., 1986) and that the family members of individuals with alcoholism are at increased risk for anxiety disorder (e.g., Mathew, Wilson, Blazer, & George, 1993; Reich, Earls, Frankel, & Shayka, 1993; Sher, Walitzer, Wood, & Brent, 1991); however, not all studies obtain the latter pattern of findings (e.g., Schuckit, 1994). In any case, as reviewed below, the absence of specificity concerning the status of both comorbid disorders in probands and family members limits the amount of information these studies can provide about the etiology of comorbidity. Family studies and comorbidity models. Merikangas et al. (1996) give a detailed description of how specific patterns of familial transmission provide evidence for or against specific etiological models of comorbidity. For instance, probands experiencing one disorder that has the potential to cause a second disorder should have family members at an increased risk for both the causal disorder alone (“pure form”) and the causal disorder plus the noncausal disorder (“combined form”); however, these family members should not be at an increased risk for the noncausal disorder alone. As exemplified by Merikangas et al. (1996), “. . . if anxiety caused alcoholism, relatives of probands with anxiety should have an increased rate of pure anxiety or anxiety combined with alcoholism, but should have normal rates of pure alcoholism” (p. 102). On the other hand, if both disorders are caused by a third variable transmitted within the family, then the family members of probands with the pure form of either disorder should have family members at risk for: (a) the pure form of the proband’s disorder; (b) the combined (comorbid) disorder form; and, most importantly, (c) the pure form of the comorbid disorder that the proband does not have (“cross-transmission”). Studies allowing for examination of some or all of these patterns of familial transmission are reviewed below. 160 M. G. Kushner, K. Abrams, and C. Borchardt Family studies of pure anxiety disorder probands. Maier, Lichtermann, Minges, Oehrlein, and Franke (1993) examined the family members of 40 probands with “pure” panic disorder (i.e., no history of psychotic disorders, major depression, or alcoholism) and those of 80 controls. Not surprisingly, they found a significantly elevated agecorrected risk for panic disorder in family members of panic probands. More important to the present topic, they found that alcoholism was significantly more common in the relatives of pure panic probands (17.5%) than was true for healthy controls (11.9%). This effect held for both men (24.8% vs. 13.0%) and women (10.0% vs. 2.7%). Unfortunately, these authors did not distinguish pure form alcoholism from combined form alcoholism in family members. Therefore, it cannot be distinguished whether this familial segregation pattern is more consistent with the view that anxiety disorder promotes alcoholism (e.g., panic was transmitted within the family of probands with pure panic which, in turn, increased affected family members’ risk for comorbid alcoholism); or, whether this pattern is more consistent with the view that a third factor promotes both anxiety disorder and alcoholism (i.e., a factor was transmitted within the family of probands with pure panic which promotes both panic and alcoholism). Family studies of pure vs. comorbid probands. Maier, Minges, and Lichtermann (1993) report a family study using proband groups with: (a) no disorder (controls); (b) pure panic; (c) pure alcoholism, or (d) comorbid panic and alcoholism. Again, and, not surprisingly, the data show that panic and alcoholism aggregate separately in the families of probands with the same disorder. Of more interest, they report that, compared to controls, the risk for comorbidity in the family members of probands with pure panic disorder was increased by a factor of eight. Similarly, the risk for comorbidity in the family members of probands with pure alcoholism was increased above controls by a factor of five. These findings are consistent with the view that either panic or alcoholism transmitted to family members can serve as a causal factor in the onset of the other condition. Regarding cross-transmission, the risk for pure alcoholism was modestly elevated in the family members of probands with pure panic disorder as compared to the family members of controls probands (OR 5 1.7). Further, the risk for pure panic disorder was also somewhat elevated in the family members of probands with pure alcoholism as compared to the family members of control probands (OR 5 1.5). While neither of these odds ratios are of an impressive magnitude relative to effects noted above, a proportional hazard model analysis found the former, but not the latter, to be statistically significant. Notably, they reported that these results were similar even when the diagnostic parameters were expanded to include other anxiety disorders (excepting simple phobia and GAD) and substance use disorders other than alcoholism. Preliminary reports in the Yale Family Study of the Comorbidity of Alcoholism and Anxiety (e.g., Merikangas et al., 1996) provide data on familial transmission patterns for 226 probands with pure alcoholism, pure anxiety disorder, combined (comorbid) disorder, or no disorder (controls). As in previous reports, they found an increased risk in the family members for the disorder(s) carried by the proband; however, this finding was more true for alcohol dependence than abuse. The family members of probands with pure anxiety disorder, compared to controls, showed an increased risk for comorbidity (6% vs. 3%) as did the family members of probands with pure alcohol dependence (7% vs. 3%). Although no statistical tests were reported, these data appear to be consistent with the view that either an anxiety The Relationship Between Anxiety Disorders and Alcohol Use Disorders 161 disorder or alcoholism that is transmitted in the family can, in turn, promote the development of the comorbid disorder. Regarding cross-transmission, Merikangas et al. (1996) reported that the family members of probands with pure anxiety disorder were about twice as likely to have pure alcohol dependence as compared to the family members of control probands (8% vs. 4%). However, this pattern of cross-transmission did not hold for probands with pure alcoholism. That is, the family members of probands with pure alcoholism experienced pure anxiety disorder at a rate similar to that found in the family members of control probands (11% vs. 13%, respectively). Merikangas et al. (1996) suggested that their findings regarding cross-transmission indicate “. . . some role for shared etiologic factors in the comorbidity of alcoholism and anxiety disorders that could account for the increased risk of alcoholism in relatives of probands with anxiety disorders” (p. 103). Notably, by implying that a shared etiology model applies only to a subset of comorbidity cases (i.e., those having probands with pure anxiety disorder), Merikangas et al. (1996) raise the possibility discussed earlier that the etiology of comorbidity is heterogeneous. Twin studies examining common versus distinct etiologies for anxiety disorder and alcoholism. Kendler et al. (1995) provide a unique twin study of the interrelationship between environmental (both familial and non-familial) and genetic risk classes for several anxiety disorders (phobias, GAD, panic) and alcoholism. Complex analyses of twin pairs suggest that genetic factors, along with nonfamilial environmental factors, do contribute significantly to the development of both anxiety disorder and alcoholism; however, there was little overlap in the specific risk factors relevant to the two conditions. For example, while 12% of genetic variance for alcoholism was associated with two general genetic factors, 76% was associated with genetic factors unique to alcoholism. Similarly, nonfamilial environmental factors associated with alcoholism had a negligible relationship to those associated with anxiety disorders. Of some interest, shared environmental familial factors assessed (e.g., parental child-rearing style) had little relationship to the etiology of either alcoholism or anxiety disorders. Mullan, Gurling, Oppenheim, and Murray (1986) identified alcoholics who have a monozygotic (MZ) or dizygotic (DZ) twin. Studying 54 such twin pairs, they reported that rates of anxiety disorder were comparable between the alcoholic probands and their alcoholic co-twins; however, anxiety disorder rates were higher in the alcoholic probands than in their nonalcoholic co-twins. Notwithstanding the absence of statistical tests showing these differences to be significant (presumably due to small cell sizes), these findings appear to suggest that being alcoholic promotes the development of anxiety disorder. Further, the absence of differences in risk for anxiety disorder between MZ and DZ alcoholic co-twins, along with the likelihood that all twins were raised in the same household, strongly implies that these findings cannot be attributed to genetic or other familial effects. Replication of this study with larger sample sizes would increase our confidence in the generalizability of these interesting findings. Summary. Studies reviewed are consistent with the idea that either alcoholism or anxiety disorder can, to some extent, serve as a causal stimulus vis a vis the other. Probands with either disorder alone have family members at a substantially increased risk for comorbidity (i.e., Maier, Minges, & Lichtermann, 1993; Merikangas et al., 1996). Also, alcoholic co-twins (MZ and DZ) share the same increased risk for anxiety disorder as 162 M. G. Kushner, K. Abrams, and C. Borchardt their alcoholic probands; however, the nonalcoholic co-twins of alcoholics probands show no increased risk for anxiety disorder (Mullan et al., 1986). Evidence of cross-transmission of pure disorders would indicate that a common etiology, transmitted within the family, may play a role in some cases of comorbidity. However, because cross-transmission effects were unstable and relatively small in the studies by Maier, Minges, and Lichtermann (1993) and Merikangas et al. (1996), it is probably premature to conclude that a common factor contributes to the development of comorbidity. Also suggesting caution concerning cross-transmission are the findings reported by Kendler et al. (1995) showing no evidence for a common etiology of anxiety disorder and alcoholism in a large sample of twin pairs. Prospective Studies Using Nonclinical Samples When considering the possible causal associations between comorbid disorders, there are few more intuitive questions than which disorder tends to begin first; or, put another way, what if any impact does either disorder alone have upon the longitudinal likelihood for a new onset of the other. This question has been assessed retrospectively in clinical and community samples a number of times. These data suggest that anxiety disorder tends to begin prior to alcoholism for a small majority of comorbid individuals (e.g., Christie et al., 1988; Wittchen, Essau, & Krieg, 1991). Once again, however, retrospective reports of disorder onset are subject to significant biases and errors, rendering the validity of data obtained by this method questionable (Kushner et al., 1990; Schuckit & Hesselbrock, 1994). The prospective study approach has been recommended as a potentially valid method for examining the sequencing of comorbid disorder onset (e.g., Anthony, 1991). Unfortunately, there have been only a few such studies, and these are rather mixed in terms of diagnostic rigor, population, and methodologic approach. Nonetheless, we believe that the potential for prospective studies to inform us about processes associated with comorbidity is sufficiently great as to warrant their liberal inclusion in a review of this literature.4 Prospective studies of anxiety-spectrum clinical phenomena and substance use—Children. Shyness has been shown to be a fairly stable trait that manifests in very young children and may share biological substrates with social anxiety in adults (e.g., Kagan, Reznick, & Snidman, 1998). Ensminger, Brown, and Kellam (1982) reported that boys, but not girls, rated as shy by their first-grade teacher showed a trend toward more alcohol use 10 years later; however, this effect did not hold for marijuana or cigarettes. Further, first-grade boys rated at as both shy and aggressive demonstrated the heaviest use of all three substances at follow-up. More than shyness, behavioral inhibition is a well-researched childhood temperamental construct that has been directly linked to childhood anxiety disorder (e.g., Biederman et al., 1993). Caspi, Moffitt, Newman, and Silva (1996) assessed the behav4Past reviews of this literature (e.g., Schuckit & Hesselbrock, 1994) have included the longitudinal data reported by Vaillant (1980). We did not include this work here because outcome measures (e.g., orality as reflected by pessimism, passivity, self-doubt, fear of sex, suggestibility, or dependent personality) bare no straightforward relationship to anxiety disorder or anxietyspectrum disability. The Relationship Between Anxiety Disorders and Alcohol Use Disorders 163 ioral style of 3-year-old boys and girls based on 22 behavioral characteristics rated during a 90-minute testing session. They found that boys rated as behaviorally inhibited at age 3 were significantly more likely to report problems with alcohol use during an interview conducted when they were 21 years of age; however, when considering alcohol dependence as the outcome (vs. alcohol problems), this effect only approached significance. Prospective studies of anxiety-spectrum phenomena and substance use — Adults. Hagnell, Lanke, Rorsman, and Ohman (1986) examined a community-based sample (N 5 2,612) that included all inhabitants in a delimited geographical area in Sweden. They reported that alcoholism was 13.5- to nearly 16-fold greater at the follow-up among those who, 15 years earlier, demonstrated “symptom neurosis” (included symptoms of anxiety, depression, and asthenia/tiredness) in combination with a “subsolid” personality (defined as a high degree of emotional lability). Interestingly, however, neither of these categories were predictive of alcoholism when considered separately. Unfortunately, this report is not specific as to the diagnostic status of subjects at baseline, nor do their diagnostic categories map easily onto a modern diagnostic nosology (e.g., Diagnostic and Statistical Manual of Mental Disorders, fourth edition [DSM-IV]; American Psychiatric Association, 1994). Kammeier, Hoffmann, and Loper (1973) identified 38 men who had first taken a personality inventory while in college and then again while in alcoholism treatment (13-year interval, on average). They reported significantly more “deviance” in subjects’ treatment profiles—including psychasthenia—than in their college profiles. They noted that the lack of controls makes it impossible to rule out the possibility that the treatment profiles were elevated in response to factors other than alcoholism, such as the general negative circumstances of being in inpatient treatment, aging, or physical health problems. Also, it cannot be determined if psychasthenia worsened after college but prior to the onset of alcoholism. Prospective studies of anxiety symptoms/disorders and substance use—Adolescents and young adults. Friedman, Utada, Glickman, and Morrissey (1987) recruited 232 high school boys and girls who reported recent drug or alcohol use. Subjects were first assessed at an average age of 15 years (time 1) with additional assessments conducted 8 and 17 months later (times 2 & 3). Cross-lagged panel correlations between times 1 & 3 were interpreted to indicate reciprocal causal influence between general psychological symptomatology (measured by a self-report scale with good psychometric properties) and level of self-reported drug/alcohol use. These effects remained robust for all anxiety subscales of the self-report instrument (i.e., “interpersonal sensitivity,” “anxiety” and “obsessive-compulsive”) except “phobic anxiety.” Kushner, Sher, and Erickson (1999) assessed 454 men and women college students as freshmen (time 1) and again both 3 and 6 years later (times 2 & 3, respectively). (Subjects with a family history of alcoholism were oversampled to provide a roughly even split in terms of those who were family history positive and family history negative.) Assessments included selected 12-month DIS/DSM-III anxiety disorders (GAD, agoraphobia, social phobia, and/or panic) and 12-month DIS/DSM-III alcohol use disorders (alcohol dependence alone, and alcohol abuse with or without dependence). They found that having an anxiety disorder diagnosis alone at time 1 increased by 3.5 times the odds of developing a new alcohol dependence diagnosis by time 3. Similarly, having an anxiety disorder diagnosis alone at time 2 increased by five times the 164 M. G. Kushner, K. Abrams, and C. Borchardt odds of developing a new alcohol dependence diagnosis by time 3. Conversely, those with an alcohol dependence diagnosis alone at either times 1 or 2 experienced an approximate fourfold increase in the odds of developing a new anxiety disorder diagnosis by time 3. These effects held after statistically controlling any influence exerted by family history for alcoholism or by gender. Importantly, findings were not robust when alcohol abuse (vs. dependence alone) was also considered. Summary. Prospective studies converge on the conclusion that problems with either anxiety or alcohol use alone can set the stage for developing the other. The anxietyspectrum phenomena of shyness and behavioral inhibition in children predicted the development of alcohol problems in adolescence and young adulthood; however, this effect applied to boys only (Caspi et al., 1996; Ensminger et al., 1982). The anxietyspectrum conditions of anxiety neurosis and subsolid personality in adults predicted the later development of alcoholism (Hagnell et al., 1986). Psychasthenia was shown to be worse during alcoholism treatment than was true years before alcoholism had developed (Kammeier et al., 1973). Friedman et al. (1987) found the prospective predictive value between anxiety symptoms and substance use to be bidirectional in adolescents. Similarly, Kushner et al. (1999) found the prospective predictive value between anxiety disorder and alcohol dependence to be bidirectional in young adults. DISCUSSION Primary Conclusions Perhaps the strongest conclusion that can be drawn from this review is that anxiety disorder and alcohol use disorder each appear to have the capacity to serve as a causal stimulus in the development of the other. Providing strong support for this conclusion, prospective studies show that having either an anxiety disorder alone or an alcohol use disorder alone predicts the later development of the other (e.g., Kushner et al., 1999). Patterns of familial transmission—that is, increased comorbidity in the family members of probands with anxiety disorder alone or alcohol use disorder alone— are also indicative of a reciprocally causal connection between comorbid disorders (e.g., Maier, Minges, & Lichtermann, 1993). However, weak and inconsistent evidence for cross-transmission of pure disorders within families (Maier, Minges, & Lichtermann, 1993; Merikangas et al., 1996), along with findings from twin studies (Kendler et al., 1995; Mullan et al., 1986), provide little support for the idea that a shared third factor operates to cause both comorbid disorders. Notably, and discussed in more detail below, the findings supporting these conclusions are generally more robust for cases involving alcohol dependence versus those involving alcohol abuse without dependence. A small number of clinical studies also show that anxiety disorder can contribute to a poorer outcome in alcoholism treatment and an increase in the risk of relapse (LaBounty et al., 1992; Tomasson & Vaglum, 1996). Similarly, successfully treating comorbid anxiety disorder appears to improve alcoholism treatment outcome (e.g., Fals-Stewart & Schafer, 1992; Tollefson et al., 1992). These findings suggest that, beyond the capacity for anxiety disorder and alcohol use disorder to contribute to the initiation of the other (above), anxiety disorder also appears to contribute to the persistence of alcohol use disorder. Unfortunately, however, the literature does not clarify the effect of alcoholism on the persistence of anxiety disorder. The Relationship Between Anxiety Disorders and Alcohol Use Disorders 165 Explanatory Models Revisited Although we concluded from the findings reviewed that both anxiety disorder and alcohol use disorder can contribute to the etiology of the other, the specific mechanism(s) underlying this influence are not known. However, we do believe that sufficient evidence exists to allow us to formulate some likely possibilities. In this section, therefore, we consider the evidence for and against the operation of several putative causal mechanisms in the development of comorbidity. Anxiety disorder promoting comorbid alcoholism. Laboratory studies provide solid grounds for concluding that the acute pharmacological effects of alcohol intoxication can include stress-response dampening and reduction of clinical anxiety symptoms. Further, alcohol self-administration studies suggest that some individuals engage in a self-medicating style of drinking. However, and notwithstanding the questionable external validity of laboratory self-administration studies, the likelihood that a self-medicating style of drinking would lead to frank alcoholism is unknown at the present time. Nonetheless, some evidence suggests that a self-medicating style of drinking— whether emerging before or after alcoholism is established—can contribute to a relapse to problem drinking following a period of abstinence among comorbid individuals (LaBounty et al., 1992). An additional mechanism that could promote comorbidity was suggested to us by the notable overlap in the neuro-processes associated with anxiety disorder and alcohol withdrawal states (e.g., GABA and NE) (e.g., George et al., 1990; Meltzer, 1987). We conjecture that this overlap may confer an increase in risk for the development of a withdrawal syndrome for those with anxiety disorder. That is, because anxiety-disordered individuals already manifest neuro-abnormalities thought to underlie the experience of withdrawal symptoms, they may require less involvement with alcohol for a withdrawal syndrome to manifest. Importantly, however, this possibility has not been evaluated empirically. Alcoholism promoting comorbid anxiety disorder. Laboratory studies reviewed show that chronic alcohol use, especially that involving dependence and withdrawal, can induce changes in neuro-chemical systems linked to the induction of anxiety and panic (e.g., Borg et al., 1981; Coffman & Petty, 1985). We must again wonder, however, whether such processes are sufficient to initiate an independent anxiety disorder versus a temporary anxiety syndrome that is dependent upon continued pathological use of alcohol. Some alcoholism-induced neuronal changes have been found to persist for months or years following abstinence (e.g., Gross & Hastey, 1976; Roelofs, 1985). Therefore, it is conceivable that alcoholism-induced changes in the central nervous system that promote anxiety symptoms might also persist long after abstinence. However, additional studies will be required to resolve this issue firmly. Notably, even if alcoholism initiates but does not maintain anxiety disorder symptoms, other cognitive and/or biological processes might come to maintain anxiety problems following abstinence (e.g., Barlow, 1988; Klein, 1980). For example, according to Barlow (1988), changes in cognitive processes following initial experiences with panic (e.g., increased worry over symptoms and self-focused scanning for signs of another episode) can increase the likelihood of future attacks regardless of whether the original panic impetus is still operative. Again, establishing the validity of this possibility will require additional research. 166 M. G. Kushner, K. Abrams, and C. Borchardt A hybrid explanatory model of comorbidity. We propose that an interaction of the anxiogenic and anxiolytic processes outlined above might best explain the etiology of comorbidity. Consider once again the study reported by Stockwell et al. (1982) in which relief from anxiety immediately following drinking, along with the expectation that anxiety would increase if drinking were discontinued, appeared alongside an overall worsening of anxiety as subjects continued to drink. This outcome exemplifies how a feed-forward cycle could emerge in which drinking maintained by the immediate reinforcement of anxiety reduction contributes to an overall worsening of anxiety that, in turn, promotes more drinking, and so on. While those individuals with alcoholism involving withdrawal would be routinely exposed to this cycle of drinking-related anxiety reduction and induction (e.g., Stockwell et al., 1982), this is not necessarily so among individuals with an anxiety disorder alone. Only those who drink chronically enough to initiate anxiogenic consequences would activate this hypothetical feed-forward cycle. While this would imply that alcoholism should most often predate anxiety disorder in comorbidity, longitudinal and retrospective studies show that anxiety disorder is as, or more likely to predate alcoholism in cases of comorbidity (Kushner et al., 1990). In spite of the absence of a perfect fit of the feed-forward model with empirical observations, we suggest that this view fits better with what is presently known about comorbidity than do the alternatives discussed (e.g., either disorder as the sole cause of the other or a common factor causing both). Therefore, efforts at refining this model may make the most sense at present. For example, as noted above, it may be that individuals with anxiety disorders are especially prone to the development of withdrawal symptomatology. Such processes could help to explain why those with anxiety disorder are apparently so vulnerable to entering into the vicious feed-forward cycle of drinking-related anxiety reduction and induction. Interpretational Caveats Comorbidity and type of substance use disorder. As alluded to above, the significant findings reviewed herein were generally associated with cases involving alcohol dependence versus those involving alcohol abuse without dependence. This pattern was evident in key studies looking at the prevalence of comorbidity (e.g., Regier et al., 1990), the familial links between comorbid disorders (e.g., Merikangas et al., 1996), and the prospective relationship between comorbid disorders (e.g., Kushner et al., 1999). We suggest, therefore, that the aforementioned conclusions and conjectures apply primarily to comorbidity involving alcohol dependence rather than alcohol abuse without dependence. Unfortunately, a firm explanation for the distinction between alcohol dependence and alcohol abuse in comorbidity is presently lacking. However, the feed-forward model of comorbidity, as proposed above, does postulate a central role for alcohol disorder with physical dependence in the etiology of comorbidity. Beyond the distinction between alcohol dependence and abuse, the conclusions noted above cannot be assumed to apply to SUDs involving drugs other than alcohol. With this said, the significant odds ratios between drug dependence and anxiety disorder in the ECA survey noted earlier (e.g., Regier et al., 1990) do suggest a meaningful relationship between these conditions. In this regard, we tentatively suggest that drugs associated with short-term anxiety dampening and a clear physical dependence syndrome (e.g., opiates) may be the most likely to parallel alcohol dependence in their relation to comorbidity as discussed above. However, the acute effect of drugs such as The Relationship Between Anxiety Disorders and Alcohol Use Disorders 167 marijuana and cocaine may actually induce anxiety and panic feelings (e.g., Rosenbaum, 1986; Szuster, Pontius, & Campos, 1988). Thus, while the pathological use of these drugs may well be related to anxiety problems, the nature of this relationship is likely to differ from that of alcohol dependence. Comorbidity and type of anxiety disorder. The general conclusions drawn from this review may also be more or less valid when applied to specific anxiety disorders. Epidemiological data from the ECA survey linked alcohol dependence to nearly all of the anxiety disorders except for simple phobia (Himle & Hill, 1991). Family studies obtained stronger effects when considering panic disorder (Maier, Minges, & Lichtermann, 1993) than when considering several anxiety disorders together (Merikangas et al., 1996). Similarly, the association of anxiety disorder to alcoholism relapse was more robust for panic disorder (Tomasson & Vaglum, 1996) than for several anxiety disorders considered together (LaBounty et al., 1992). Thus, while all of the anxiety disorders reviewed—with the possible exception of simple phobia—appear to have a significant association with alcohol dependence, the strength of that association may vary between anxiety disorders. Further, we cannot rule out the possibility that the various anxiety disorders differ in the mechanisms underlying their comorbidity with alcoholism. In fact, Kushner et al. (1990) pointed to biological, behavioral, and clinical differences among the anxiety disorders as suggestive of a unique relationship to comorbid alcoholism for each. CONCLUSION Comorbidity constitutes one of the most perplexing problems faced by taxonomists, epidemiologists, and psychopathologists today. The main goal of this article was to identify seemingly competing views on the etiology of comorbidity and to review literature bearing on the validity of each. Major conclusions are that either an anxiety disorder or alcohol dependence can serve as a causal stimulus for the other and that anxiety disorder can contribute to the persistence of pathological alcohol use and relapse to drinking among those with both disorders. We propose a feed-forward vicious cycle model in which the interaction between anxiety disorder symptoms and alcohol dependence contributes to the etiology of comorbidity. Finally, we note that while hundreds of studies relate to comorbidity, few studies address rigorously the direct interaction of comorbid disorders using laboratory, clinical, prospective or family methods. Therefore, while we were able to reach some conclusions here, it is our hope that this review promotes coordinated and focused efforts toward a better understanding of comorbidity. Acknowledgment—This research was supported, in part, by National Institute on Alcohol Abuse and Alcholism grant no. R29-AA09871 to the first author. REFERENCES Abrams, K., Kushner, M. G., Lisdahl, K., & Voight, A. (1997). The expectancy and pharmacologic effects of alcohol on social anxiety in individuals with social phobia. Paper presented at the 31st annual convention of the Association for the Advancement of Behavior Therapy, Miami, FL. 168 M. G. Kushner, K. Abrams, and C. Borchardt Alexanderson, G., & Lindman, R. (1980). Of mice and women. 2. Differential effects of dose information and administration of alcohol on fear. In Anxiety and Alcohol: Limitation of Tension Reduction Theory in Nonalcoholics, Report from the Department of Psychology at Abo Akademi, Monograph Supplement 1. Allan, C. A. (1995). Alcohol problems and anxiety disorders—A critical review. Alcohol and Alcoholism, 30, 145–151. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Anthony, J. C. (1991). Epidemiology of drug dependence and illicit drug use. Current Opinion in Psychiatry, 4, 435–439. Bandura, A. (1985). Social foundations of thought and actions. Englewood Cliffs, NJ: Prentice Hall. Barlow, D. H. (1988). Anxiety and its disorders. New York: Guilford Press. Berkson. J. (1949). Limitations of the application of four-fold tables to hospital data. Biological Bulletin, 2, 47–53. Bibb, J. L., & Chambless, D. L. (1986). Alcohol use and abuse among diagnosed agoraphobics. Behaviour Research and Therapy, 24, 49–58. Biederman, J., Rosenbaum, J. F., Bolduc-Murphy, E. A., Faraone, S. V., Chaloff, J., Hirsfeld, D. R., & Kagan, J. (1993). A 3-year follow-up of children with and without behavioral inhibition. Journal of the American Academy of Adolescent Psychiatry, 32, 814–821. Billing, A. G., & Moos, R. H. (1983). Psychosocial processes of recovery among alcoholics and their families: Implications for clinicians and program evaluators. Addictive Behaviors, 8, 205–218. Borg, S., Kvande, H., & Sedvall, G. (1981). Central norepinephrine metabolism during alcohol intoxication in addicts and healthy volunteers. Science, 213, 1135–1137. Bouton, M. E., Kenney, F. A., & Rosengard, C. (1990). State-dependent fear extinction with two benzodiazepine tranquilizers. Behavioral Neuroscience, 104, 44–55. Brady, K. T., & Lydiard, R. B. (1993). The association of alcoholism and anxiety. Psychiatric Quarterly, 64, 135–149. Brown, S. A., Irwin, M., & Schuckit, M. A. (1991). Changes in anxiety among abstinent male alcoholics. Journal of Studies on Alcohol, 52, 55–61. Brown, S. A., Vik, P. W., McQuaid, J. R., Patterson, T. L., Irwin, M. R., & Grant, I. (1990). Severity of psychosocial stress and outcome of alcoholism treatment. Journal of Abnormal Psychology, 99, 344–348. Brown, S. A., Vik, P. W., Patterson, T. L., Grant, I., & Schuckit, M. A. (1995). Stress, vulnerability and adult alcohol relapse. Journal of Studies on Alcohol, 56, 538–545. Cameron, O. G., Liepman, M. R., Curtis, G. C., & Thyer, B. A. (1987). Ethanol retards desensitization of simple phobias in non-alcoholics. British Journal of Psychiatry, 150, 845–849. Cappell, H., & Greeley, J. (1987). Alcohol and tension reduction: An update on research and theory. In H. T. Blane & K. E. Leonard (Eds.), Psychological theories of drinking and alcoholism (pp. 15–55). New York: Guilford Press. Carroll, K. M., Power, M. D., Bryant, K., & Rounsaville, B. J. (1993). One-year follow-up status of treatmentseeking cocaine abusers: Psychopathology and dependence severity as predictors of outcome. The Journal of Nervous and Mental Disease, 181, 71–79. Caspi, A., Moffitt, T. E., Newman, D. L., & Silva, P. A. (1996). Behavioral observations at age 3 years predict adult psychiatric disorders. Archives of General Psychiatry, 53, 1033–1039. Chambless, D. L., Cherney, J., Caputo, G. C., & Rheinstein, B. J. G. (1987). Anxiety disorders and alcoholism. Journal of Anxiety Disorders, 1, 29–40. Chambless, D. L., Foa, E. B., Groves, G. A., & Goldstein, A. J. (1979). Flooding with brevital in the treatment of agoraphobia: Countereffective? Behaviour Research and Therapy, 17, 243–251. Christie, K. A., Burke, J. D., Regier, D. A., Rae, D. S., Boyd, J. H., & Locke, B. Z. (1988). Epidemiologic evidence for early onset of mental disorders and higher risk of drug abuse in young adults. American Journal of Psychiatry, 145, 971–975. Cloninger, C. R., Martin, R. L., Clayton, P., & Guze, S. B. (1981). A blind follow-up and family study of anxiety neurosis: Preliminary analysis of the St. Louis 500. In D. F. Klein & J. Rabkin (Eds.), Anxiety: New research and changing concepts (pp. 137–150). New York: Raven Press. Coffman, J. A., & Petty, F. (1985). Plasma GABA levels in chronic alcoholics. American Journal of Psychiatry, 142, 1204–1205. Cowley, D. S. (1992). Alcohol abuse, substance abuse, and panic disorder. The American Journal of Medicine, 92(Suppl. 1A), 41–48. Cox, B. J., Norton, G. R., Dorward, J., & Fergusson, P. A. (1989). The relationship between panic attacks and chemical dependencies. Addictive Behaviors, 14, 53–60. Cox, B. J., Norton, G. R., Swinson, R. P., & Endler, N. S. (1990). Substance abuse and panic-related anxiety: A critical review. Behaviour Research and Therapy, 28, 385–393. Crowe, R. R., Noyes, R., Pauls, D. L., & Slymen, D. (1983). A family study of panic disorder. Archives of General Psychiatry, 40, 1065–1069. The Relationship Between Anxiety Disorders and Alcohol Use Disorders 169 Dorow, R., Horowski, R., Paschelke, G., & Amin, M. (1983). Severe anxiety induced by FG7142, a B carboline ligand for benzodiazepine receptors. Lancet, II, 98–99. Ensminger, M. E., Brown, C. H., & Kellam, S. G. (1982). Sex differences in antecedents of substance use among adolescents. Journal of Social Issues, 38, 25–42. Eriksson, E., Westberg, P., Thuresson, K., Mordigh, K., Ekman, R., & Widerlov, E. (1989). Increased cerebrospinal fluid levels of endorphin immunoreactivity in panic disorder. Neuropschopharmacology, 2, 225–228. Fals-Stewart, W., & Schafer, J. (1992). The treatment of substance abusers diagnosed with obsessive-compulsive disorder: An outcome study. Journal of Substance Abuse Treatment, 9, 365–370. Freed, E. X. (1978). Alcohol and mood: An updated review. International Journal of Addictions, 13, 173–200. Friedman, A. S., Utada, A. T., Glickman, N. W., & Morrissey, M. R. (1987). Psychopathology as an antecedent to, and as a “consequence” of, substance use, in adolescence. Journal of Drug Education, 17, 233–244. George, D. T., Nutt, D. J., Dwyer, B. A., & Linnoila, M. (1990). Alcoholism and panic disorder: Is the comorbidity more than coincidence? Acta Psychiatrica Scandinavica, 81, 97–107. Gold, M. S., Pottash, A. L. C., Sweeney, D. R., Kleber, H. D., & Redmond, D. E., Jr. (1979). Rapid opiate detoxification: Clinical evidence of antidepressant and antipanic effects of opiates. American Journal of Psychiatry, 136, 982–983. Gold, M. S., Redmond, D. E., Jr., & Kleber, H. D. (1979). Noradrenergic hyperactivity in opiate withdrawal supported by clonidine reversal of opiate withdrawal. American Journal of Psychiatry, 136, 100–102. Gross, M. M., & Hastey, J. M. (1976). Sleep disturbances in alcoholism. In R. E. Tarter & A. A. Sugerman (Eds.), Alcoholism: Interdisciplinary approaches to an enduring problem (pp. 257–308). Reading, MA: AddisonWesley Publishing Company. Hagnell, O., Lanke, J., Rorsman, B., & Ohman, R. (1986). Predictors of alcoholism in the Lundby study II: Personality traits as risk factors for alcoholism. European Archives of Psychiatry and Neurological Science, 235, 192–196. Hawley, R., Major, L., Schulman, E., & Linnoila, M. (1985). Cerebrospinal fluid 3-methyloxy4hydroxyphenylglycol and norepinephrine levels in alcohol withdrawal. Archives of General Psychiatry, 42, 1056–1062. Himle, J. A., & Hill, E. M. (1991). Alcohol abuse and the anxiety disorders: Evidence from the Epidemiologic Catchment Area survey. Journal of Anxiety Disorders, 5, 237–245. Hull, J. (1981). A self-awareness model of the causes and effects of alcohol consumption. Journal of Abnormal Psychology, 90, 586–600. Jones, E. E., & Berglas, S. (1978). Control of attributions about the self through self-handicapping strategies: The appeal of alcohol and the role of underachievement. Personality and Social Psychology Bulletin, 4, 200–206. Kagan, J., Reznick, S., & Snidman, N. (1998). Biological bases of childhood shyness. Science, 240, 167–171. Kammeier, M. L., Hoffmann, H., & Loper, R. G. (1973). Personality characteristics of alcoholics as college freshmen and at time of treatment. Quarterly Journal on the Study of Alcoholism, 34, 390–399. Kendler, K. S., Walters, E. E., Neale, M. C., Kessler, R. C., Heath, A. C., & Eaves. L. J. (1995). The structure of the genetic and environmental risk factors for six major psychiatric disorders in women: Phobia, generalized anxiety disorder, panic disorder, bulimia, major depression, and alcoholism. Archives of General Psychiatry, 52, 374–383. Kessler, R. C., Crum, R. M., Warner, L. A., Nelson, C. B., Schulenberg, J., & Anthony, J. C. (1997). Lifetime co-occurrence of DSM-III-R alcohol abuse and dependence with other psychiatric disorders in the National Comorbidity Survey. Archives of General Psychiatry, 54, 313–321. Klein, D. F. (1980). Anxiety reconceptualized. Comprehensive Psychiatry, 2, 411–427. Kranzler, H. R., Burleson, J. A., Del Boca, F. K., Babor, T. F., Korner, P., Brown, J., & Bohn, M. J. (1994). Buspirone treatment of anxious alcoholics: A placebo-controlled trail. Archives of General Psychiatry, 51, 720–731. Kushner, M. G. (1996). Relationship between alcohol problems and anxiety disorders (Letter). American Journal of Psychiatry, 153, 139. Kushner, M. G., Mackenzie, T. B., Fiszdon, J., Valentiner, D. P., Foa, E., & Wangensteen, D. (1996). The effects of alcohol consumption on laboratory induced panic and state anxiety. Archives of General Psychiatry, 53, 264–270. Kushner, M. G., Rassovsky, Y., Abrams, K., Whaley, G., Schwarze, N., Kruckeberg, J., & Mackenzie, T. (1997). Effects of panic provocation on the self-administration of alcohol in individuals with panic disorder. Paper presented at the 31st annual meeting of the Association for the Advancement of Behavior Therapy, Miami, FL. Kushner, M. G., Sher, K. J., & Beitman, B. D. (1990). The relation between alcohol problems and the anxiety disorders. American Journal of Psychiatry, 147, 685–695. Kushner, M. G., Sher, K. J., & Erickson, D. J. (1999). Prospective analysis of the relation between DSM-III anxiety disorders and alcohol use disorders. American Journal of Psychiatry, 156, 723–732. LaBounty, L. P., Hatsukami, D., Morgan, S. F., & Nelson, L. (1992). Relapse among alcoholics with phobic and panic symptoms. Addictive Behaviors, 17, 9–15. 170 M. G. Kushner, K. Abrams, and C. Borchardt Liljequist, S., & Engel, J. (1984). The effects of GABA and benzodiazepine receptor antagonist on the anticonflict actions of diazepam or ethanol. Pharmacology Biochemistry and Behavior, 21, 521–525. Lindman, R., Alexanderson, G., & Lindfors, B. (1980). In Anxiety and Alcohol: Limitation of Tension Reduction Theory in Nonalcoholics, Report from the Department of Psychology at Abo Akademi, Monograph Supplement 1. Maier, W., Lichtermann, D., Minges, J., Oehrlein, A., & Franke, P. (1993). A controlled family study in panic disorder. Journal of Psychiatric Research, 27(Suppl. 1), 79–87. Maier, W., Minges, J., & Lichtermann, D. (1993). Alcoholism and panic disorder: Co-occurrence and cotransmission in families. European Archives of Psychiatry and Clinical Neuroscience, 243, 205–211. Malcolm, R., Anton, R. F., Randall, C. L., Johnston, A., Brady, K., & Thevos, A. (1992). A placebo-controlled trial of buspirone in anxious inpatient alcoholics. Alcoholism, Clinical and Experimental Research, 16, 1007–1013. Marks, I. M., Swinson, R. P., Basoglu, M., Kuch, K., Noshirvani, H., O’Sullivan, G., Lelliot, P. T., Kirby, M., McNeamee, G., Sengun, S., & Wickwire, K. (1993). Alprazolam and exposure alone and combined in panic disorder with agoraphobia: A controlled study in London and Toronto. British Journal of Psychiatry, 162, 776–787. Marlatt, G. A. (1984). Alcohol, the magic elixir: Stress, expectancy, and the transformation of emotional states. Paper presented to seventh annual Coatesville Jefferson Conference, VA Medical Center, Coatesville, PA. Marshall, J. R. (1997). Alcohol and substance abuse in panic disorder. Journal of Clinical Psychiatry, 58(Suppl. 2), 46–49. Mathew, R. J., Wilson, W. H., Blazer, D. G., & George, L. K. (1993). Psychiatric disorders in adult children of alcoholics. Data from the epidemiologic catchment area project. American Journal of Psychiatry, 150, 793–800. McGue, M. (1994). Genes, environment, and the etiology of alcoholism. In R. Zucker, G. Boyd, & J. Howard (Eds.), Development of alcohol problems: Exploring the biopsychosocial matrix: Monograph No. 26 (pp. 1–40). Bethesda, MD: National Institute on Alcohol Abuse and Alcoholism. Mello, N. K., & Mendelson, J. H. (1978). Alcohol and human behavior. In L. L. Iversen, S. D. Iversen, & S. H. Snyder (Eds.), Handbook of psychopharmacology. Vol. 12: Drugs of abuse, section III (pp. 235–317). New York: Plenum Press. Meltzer, H. Y. (1987). Psychopharmacology: The third generation of progress. New York: Raven Press. Merikangas, K. R., & Gelernter, C. S. (1990). Comorbidity for alcoholism and depression. Psychiatric Clinics of North America, 13, 613–632. Merikangas, K. A., Stevens, D., & Fenton, B. (1996). Comorbidity of alcoholism and anxiety disorders. Alcohol Health and Research World, 20, 100–105. Mullan, M. J., Gurling, H. M. D., Oppenheim, B. E., & Murray, R. M. (1986). The relationship between alcoholism and neurosis: Evidence from a twin study. British Jouranl of Psychiatry, 148, 435–441. Nestoros, J. (1980). Ethanol specifically potentiates GABA-mediated neurotransmission in feline cerebral cortex. Science, 209, 708–710. Noyes, R., Clancy, J., Crowe, R., Hoenk, R. P., & Slymen, D. J. (1978). The family prevalence of anxiety neurosis. Archives of General Psychiatry, 35, 1057–1059. Noyes R., Crowe, R. R., Harris, E. L., Hamra, B. J., McChesney, C. M., & Chaudhry, D. R. (1986). Relationship between panic disorder and agoraphobia: A family study. Archives of General Psychiatry, 43, 227–232. Ojehagen, A., Berglund, M., Appel, C., Nilsson, B., & Skjaerris, A. (1990). Psychiatric symptoms in alcoholics attending outpatient treatment. Alcoholism: Clinical and Experimental Research, 15, 640–646. Pellow, S. (1980). Can drug effect on anxiety and convulsions be separated? Neuroscience and Biobehavioral Review, 9, 55–73. Penick, E. C., Powell, B. J., Liskow, B. I., Jackson, J. O., & Nickel, E. J. (1988). The stability of coexistent psychiatric syndromes in alcoholic men after one year. Journal of Studies on Alcohol, 49, 395–405. Peyser, H. (1982). Stress and alcohol. In L. Goldberger & S. Breznotz (Eds.), Handbook of stress (pp. 585– 598). New York: Free Press. Pohorecky, L. A. (1991). Stress and alcohol interaction: An update of human research. Alcoholism: Clinical and Experimental Research, 15, 438–459. Quitkin, F. M., Rifkin, A., Kaplan, J., & Klein, D. F. (1972). Phobic anxiety syndrome complicated by drug dependence and addiction. Archives of General Psychiatry, 27, 159–162. Redmond, D. E. (1985). Neurochemical basis for anxiety and anxiety disorders: Evidence from drugs which decrease human fear and anxiety. In A. H. Tuma & J. D. Maser (Eds.), Anxiety and the anxiety disorders (pp. 533–555). Englewood Cliffs, NJ: Lawrence Erlbaum Associates. Regier, D. A., Farmer, M. E., Rae, D. S., Locke, B. Z., Keith, S. J., Judd, L. L., & Goodwin, F. K. (1990). Comorbidity of mental disorders with alcohol and other drug abuse: Results from the Epidemiologic Catchment Area (ECA) Study. Journal of the American Medical Association, 264, 2511–2518. Reich, W., Earls, F., Frankel, O., & Shayka, J. J. (1993). Psychopathology in children of alcoholics. Journal of the American Academy of Child and Adolescent Psychiatry, 32, 995–1002. The Relationship Between Anxiety Disorders and Alcohol Use Disorders 171 Rimm, D., Briddell, D., Zimmerman, M., & Caddy, G. (1981). The effects of alcohol and the expectancy effects of alcohol on snake fears. Addictive Behaviors, 6, 47–51. Roelofs, S. (1985). Hyperventilation, anxiety, and craving for alcohol: A subacute alcohol withdrawal syndrome. Alcohol, 2, 501–505. Rosenbaum, J. F. (1986). Cocaine and panic disorder (Letter). American Journal of Psychiatry, 143, 1320. Rosenberg, H. (1983). Relapsed versus non-relapsed alcohol abusers: Coping skills, life events, and social support. Addictive Behaviors, 8, 183–186. Rounsaville, B. J., Dolinsky, Z. S., Babor, T. F., & Meyer, R. E. (1987). Psychopathology as a predictor of treatment outcome in alcoholics. Archives of General Psychiatry, 44, 505–513. Rounsaville, B. J., Kosten, T. R., Weissman, M. M., & Kleber, H. D. (1986). Prognostic significance of psychopathology in treated opiate addicts: A 2.5-year follow-up study. Archives of General Psychiatry, 43, 739–745. Rounsaville, B. J., Cacciola, J., Weissman, M. M., & Kleber, H. D. (1981). Diagnostic concordance in a followup study of opiate addicts. Journal of Psychiatric Research, 16, 191–201. Schuckit, M. A. (1994). Low level of response to alcohol as a predictor of future alcoholism. American Journal of Psychiatry, 151, 184–189. Schuckit, M. A. (1996). Alcohol, anxiety and depressive disorders. Alcohol Health and Research World, 20, 81–85. Schuckit, M. A., & Hesselbrock, V. (1994). Alcohol dependence and anxiety disorders: What is the relationship? American Journal of Psychiatry, 151, 1723–1734. Schuckit, M. A., & Hesselbrock, V. (1996). Schuckit & Hesselbrock reply (Letter). American Journal of Psychiatry, 153, 140. Sher, K. J. (1987). Stress response dampening. In H. T. Blane & K. E. Leonard (Eds.), Psychological theories of drinking and alcoholism (pp. 227–271). New York: Guilford Press. Sher, K. J., Walitzer, K. S., Wood, P. K., & Brent, E. E. (1991). Characteristics of children of alcoholics: Putative risk factors, substance use and abuse and psychopathology. Journal of Abnormal Psychology, 100, 427– 448. Smail, P., Stockwell, T., Canter, S., & Hodgson, R. (1984). Alcohol dependence and phobic anxiety states I. A prevalence study. British Journal of Psychiatry, 144, 53–57. Steele, C. M., & Josephs, R. A. (1988). Drinking your troubles away II: An attention-allocation model of alcohol’s effect on psychological stress. Journal of Abnormal Psychology, 97, 196–205. Stewart, S. H. (1996). Alcohol abuse in individuals exposed to trauma: A critical review. Psychological Bulletin, 120, 83–112. Stockwell, T., & Bolderston, H. (1987). Alcohol and phobias. British Journal of Addiction, 82, 971–979. Stockwell, T., Hodgson, R., & Rankin, H. (1982). Tension reduction and the effects of prolonged alcohol consumption. British Journal of Addiction, 77, 65–73. Stockwell, T., Smail, P., Hodgson, R., & Canter, S. (1984). Alcohol dependence and phobic anxiety states II. A retrospective study. British Journal of Psychiatry, 144, 58–63. Szuster, R. R., Pontius, E. B., & Campos, P. E. (1988). Marijuana sensitivity and panic anxiety. Journal of Clinical Psychiatry, 49, 427–429. Thevos, A. K., Johnston, A. L., Latham, P. K., Randall, C. L., Adinoff, B., & Malcolm, R. (1991). Symptoms of anxiety in inpatient alcoholics with and without DSM-III-R anxiety diagnoses. Alcoholism: Clinical and Experimental Research, 15, 102–105. Thyer, B. A., & Curtis, G. C. (1984). The effects of ethanol intoxication on phobic anxiety. Behaviour Research and Therapy, 22, 599–610. Tollefson, G. D., Montague-Clouse, J., & Tollefson, S. L. (1992). Treatment of comorbid generalized anxiety in a recently detoxified alcoholic population with a selective serotonergic drug (buspirone). Journal of Clinical Psychopharmacology, 12, 19–26. Tomasson, K., & Vaglum, P. (1996). Psychopathology and alcohol consumption among treatment-seeking alcoholics: A prospective study. Addiction, 91, 1019–1030. Tran, V. T., Snyder, S. H., Major, L. F., & Hawley, R. J. (1981). GABA receptors are increased in brains of alcoholics. Annals of Neurology, 9, 289–292. Tucker, J. A., Vuchinich, R. E., & Sobell, M. B. (1981). Alcohol consumption as a self-handicapping strategy. Journal of Abnormal Psychology, 90, 220–230. Vaillant, G. (1980). Natural history of male psychological health: VIII. Antecedents of alcoholism and “orality.” American Journal of Psychiatry, 137, 181–186. Wilson, G. T. (1987). Cognitive processes in addiction. British Journal of Addiction, 82, 343–353. Wilson, G. T. (1988). Alcohol and anxiety. Behaviour Research and Therapy, 26, 369–381. Wittchen, H., Essau, C. A., & Krieg, J. (1991). Anxiety disorders: Similarities and differences of comorbidity in treated and untreated groups. British Journal of Psychiatry, 159(Suppl. 12), 23–33.