1Introduction

Publication Details

Primary Objectives and Scope of This Evidence Report

Alcoholism has a high lifetime prevalence rate of 10 percent and costs the Nation more than $166.5 billion annually in direct and indirect health and social costs (National Institute on Drug Abuse [NIDA] and National Institute on Alcohol Abuse and Alcoholism [NIAAA], 1998). However, the actual health, social, and economic costs of this disorder encompass so many factors that they cannot be measured definitively. Table 1 summarizes basic information about alcoholism, its costs, and usual treatment.

Table 1. Facts about alcohol dependence.

Table

Table 1. Facts about alcohol dependence.

Treating patients with alcoholism generally occurs in two phases: dealing with acute withdrawal and detoxification and supporting patients in remission, rehabilitation, and abstinence. The focus for this evidence report is the latter area, which arguably poses far more complicated health care delivery and health policy and financing challenges.

Considerable uncertainty surrounds the question of how best to maintain abstinence in patients who are in remission. Some of this uncertainty stems from a debate between those who advocate a "drug-free" 12-step approach that does not incorporate use of any pharmacotherapies and those who, while accepting the great value of the 12-step approach, hold that use of appropriate, nonaddictive medications may be an important step toward improved outcomes. In addition, controversy and uncertainty remain about the advantages and drawbacks of those types of medications. Disulfiram ("Antabuse"®), for example, is not effective for many patients; they may consume alcohol while on disulfiram, or more likely, stop taking the medication entirely. Moreover, clinicians are not sure of the pharmacologic or physiologic mechanisms (or the potential side effects) of certain other agents. This hampers their ability to judge whether their patients are appropriate candidates for such therapy. Finally, pharmaceutical products that are either already in use in Europe or at least further along in clinical trials (e.g., acamprosate) offer the promise of a greatly enhanced array of treatments for this condition, once they are subjected to appropriate investigation and receive relevant approval in this country. As clinicians weigh all relevant options in this area, they will need to understand the trajectory of investigations of these drugs; this evidence report can provide them with an adequate context for decisionmaking about appropriate therapies in the next 2 to 3 years.

The Agency for Health Care Policy and Research (AHCPR), other public-sector agencies concerned with the treatment and costs of alcoholism, professional associations, patient groups, and health care delivery organizations all face the problem of scarce resources to invest in developing the knowledge base on the vast array of health care problems that might be tackled. Interest and financial support for evidence-based practice efforts must be leveraged by taking on topics that are of broad concern and for which the quality and appropriateness of health care can be improved. Increasingly, clinicians, patients and their families, administrators of health plans and clinics, and others are seeing the management of patients who abuse alcohol and are physiologically or psychologically dependent on alcohol as an important and rewarding enterprise in improving the health and well-being of both patients and their families.

Key Clinical Questions

The clinical questions have at least two major components (which are laid out in more detail in Chapter 2). Initially, they involve various established pharmaceutical agents that have been used in the United States and abroad in an effort to help patients abstain from alcohol use and to reduce their craving for alcohol. These include: disulfiram; the opiate antagonists (naltrexone and nalmefene); serotonergic agents such as buspirone, ondansetron, and the selective serotonin reuptake inhibitors (SSRIs) including citalopram, fluoxetine, fluvoxamine, paroxetine, and sertraline; and lithium.

Disulfiram and naltrexone, in particular, are mainstream agents in use in the United States today, and so they must be the first focus of our evidence report. Of special importance in this regard is that the field of pharmacotherapy for alcohol dependence has evolved substantially over the past 5 years, especially with the data on opiate antagonists.

Concomitantly, our clinical questions involve at least one promising pharmaceutical agent currently in use in Europe - acamprosate (calcium acetyl homotaurinate) - for treating alcoholism. An investigational new drug (IND) application is on file for this drug at the Food and Drug Administration (FDA) in the United States, and it is in Phase III trials. Acamprosate also will be included in a major clinical trial to be supported by NIAAA during the next 2 years. In short, acamprosate is expected to play an important role in treating patients with alcohol dependence within 2 years of the release of this report.

Technical Expert Advisory Group Involvement

Proposal criteria from AHCPR required the identification of technical experts in the field of alcohol dependence pharmacotherapies. The Technical Expert Advisory Group (TEAG) (see Appendix A for its composition) was expected to contribute to (a) advancing AHCPR's broader goals of creating and maintaining "science partnerships" and "public-private partnerships" and (b) meeting the needs of a broad array of potential customers and users of its products. Thus, it was both an adequate resource and a sounding board throughout the project. Our TEAG comprised: (1) technical/clinical experts [3 members]; (2) patients or representatives of organizations whose mission concerns the interests and perspectives of patients and consumers [2 members]; and (3) potential users of the final evidence report or other materials, including explicitly a representative of the organization that nominated the topic (i.e., the American Society of Addiction Medicine (ASAM) [3 members].

To ensure robust, scientifically relevant work, the TEAG was called on to provide reactions to work in progress and advice on substantive issues or possibly overlooked areas of research. TEAG members participated in several conference calls:

  • At the beginning of the project to discuss the key clinical questions.
  • During the project to discuss, for instance, the preliminary assessment of the literature and to provide input for our Data Extraction Form.
  • Toward the end of the project to discuss whether meta-analysis was appropriate for the therapies under evaluation.
Because of their extensive knowledge of the alcohol literature and their active involvement in professional societies, TEAG members were also asked to participate in the peer review process by providing comments on the draft report.

Alcohol Dependence: Background and Significance

Disease Biology

Two aspects of the biology of alcoholism warrant brief review in this report, because they drive a good deal of the theory behind the use of current pharmacotherapies and the development of new medications. An overview of the possible means by which certain classes of drugs may act, for instance on the brain, to control craving or intake of alcohol is presented first. This is followed by a brief discussion of the genetic aspects of this disorder. More explicit causal pathways for the key therapies that were reviewed for this report are presented in Chapter 2 on methodology.

Potential Mechanisms of Action of Drugs Used in the Treatment of Alcoholism

Although most alcohol researchers agree that alcoholism is a brain disease, the use of medications to treat the core symptoms of alcoholism has sometimes encountered resistance from the self-help community that endorses medication-free group therapy regimens. The high rate of relapse among patients with this disease, regardless of prior treatment modalities, has prompted researchers to develop appropriate medications to treat this illness.

Recent studies in neurobiology have provided a great opportunity to design radically new medications for the treatment of alcoholism and to understand the mechanisms of those treatments that have been in use for some time. Current research suggests that alcohol activates a reward pathway in the brain; this pathway involves serotonin, dopamine, glutamate, and endogenous opioids, as well as other neurotransmitters and neuromodulators.

Medications that have been used in the treatment of alcoholism vary widely in their mechanisms of action, from the effect of disulfiram inhibiting peripheral acetaldehyde metabolism to the central blockade of opioid receptors. Indeed, the mechanism of action of some compounds, such as acamprosate, is still uncertain. A brief summary follows of the known or presumed mechanisms of action of the major drugs considered in this report.

Disulfiram

The basis of disulfiram's effects on alcohol intake is its inhibition of aldehyde dehydrogenase (e.g., Wilson, Blanchard, Davidson, et al., 1984). Alcohol is metabolized to acetaldehyde and then to acetate, primarily in the liver. Because disulfiram inhibits aldehyde dehydrogenase, acetaldehyde levels increase in plasma. Acetaldehyde produces a variety of unpleasant side effects, including flushing, changes in blood pressure, and nausea. Disulfiram is used to reinforce the patient's desire to stop drinking by providing a psychological deterrent to consuming alcohol that is then reinforced by an unpleasant response should alcohol be consumed. Disulfiram has been in use for the treatment of alcoholism for close to 50 years.

Naltrexone

Naltrexone, an opiate antagonist, received FDA approval in December 1994 and became the first drug to be approved for the treatment of alcoholism since disulfiram's approval. Although its mechanism of action is very different from that for disulfiram, naltrexone's involvement in bringing about a reduction in alcohol intake in animals and in humans is not completely understood. Because the endogenous opioids are known to be a component of the brain reward pathway, naltrexone and other opiate antagonists that block these receptors are believed to reduce the craving for alcohol, the reward produced by alcohol, the intensity of intoxication, or all three of these alcohol outcomes (Hyytia and Sinclair, 1993; Volpicelli, Watson, King, et al., 1995; O'Malley, Jaffe, Rode, et al., 1996; Spanagel and Zieglgansberger, 1997; Wilcox and McMillen, 1998).

Acamprosate

Acamprosate, calcium acetyl homotaurinate, has been widely studied in European populations with alcoholism, but its mechanisms of action have not yet been clarified. When acamprosate was first introduced, it was promoted as a gamma-amino butyric acid (GABA) agonist (Lhuintre, Daoust, Moore, et al., 1985). More recent studies have focused on its possible interaction with the excitatory amino acid receptor ion channel complex known as the N-methyl D-Aspartate (NMDA) receptor. It seems clear that acamprosate does not bind directly to the NMDA receptor; whether it binds to a modulatory site on this receptor is still being debated (Spanagel and Zieglgansberger, 1997; Wilde and Wagstaff, 1997). Some evidence also indicates that acamprosate might block calcium flux into cells (Spanagel and Zieglgansberger, 1997). Such a mechanism might contribute to its inhibitory effects on alcohol intake; several preclinical studies demonstrate the effectiveness of calcium channel inhibitors in reducing alcohol intake (e.g., Rezvani, Pucilowski, Grady, et al., 1993). In general, the mechanism of action of acamprosate is not understood.

Serotonergic drugs

The involvement of the brain's serotonergic system in alcohol intake is well documented. Manipulations leading to decreased serotonin function seem to increase ethanol intake (for review see LeMarquand, Pihl, and Benkelfat, 1994). Systemic administration of agents that increase the physiologically active pool of serotonin produces attenuation of alcohol consumption in alcohol-preferring rats. Administration of the SSRIs such as fluoxetine, zimeldine, citalopram, and sertraline or serotonin releasers fenfluramine and dexfenfluramine reduces ethanol intake in rats (LeMarquand, Pihl, and Benkelfat, 1994). Low levels of serotonin are thought to result in increased impulsiveness and craving components that can contribute to the development of alcoholism.

When specific serotonin receptors are studied, the picture becomes less clear. There are numerous, i.e., more than 14, serotonin receptor subtypes with the 5-HT1A, 5-HT3, and 5-HT2 being those most strongly implicated as being involved in alcoholism. Analysis of the literature indicates that 5-HT2 and 5-HT3 receptors are probably involved in the regulation of alcohol intake by modulation of alcohol-reinforcing properties mediated by the dopamine (DA) system in the ventral tegmental area (VTA) (Koob and Bloom, 1988). The 5-HT2 agonist DOI has been shown to potentiate the effect of alcohol on dopamine release in the nucleus accumbens (Bowens and McBride, 1997). A role of the 5-HT3 receptors in mediating the excitatory actions of ethanol is also well documented. 5-HT3 receptor antagonists inhibited the increase in extracellular DA levels in the nucleus accumbens elicited by ethanol (Carboni, Acquas, Frau, et al., 1989; Wozniak, Pert, Linnoila, et al., 1990) and suppress voluntary alcohol intake in rats and humans (see LeMarquand, Pihl, and Benkelfat, 1994). On the other hand, 5-HT1 receptors have been implicated in the development of excessive alcohol drinking. Administration of a variety of 5-HT1 agonists (e.g., 8-OH-DPAT, mCPP, ipsapirone, buspirone) have been shown to reduce alcohol intake (LeMarquand, Pihl, and Benkelfat, 1994). Thus, a variety of 5-HT receptors appear to be involved in alcoholism.

Therefore, the brain 5-HT system may modulate alcohol intake by two different mechanisms: modulation of the dopamine-mediated reinforcing properties of alcohol via 5-HT2 and 5-HT3 receptors and suppression of alcohol intake by activation of 5-HT1A receptors.

Lithium

Lithium has been used successfully in treating patients with affective disorders (Coppen, Noguera, Baily, et al., 1971), and it was initially thought to be beneficial for the treatment of alcoholics who can exhibit cyclic mood swings and depression. Early publications suggested that the therapeutic effect of lithium derived from the stabilization of associated affective disorders, although the evidence to support this view was largely indirect. Judd, Hubbard, Huey, et al. (1977) and Judd and Huey (1984) found that lithium attenuates the subjective sensation of alcohol intoxication, including self-rating report of the desire to continue drinking after the ethanol challenge. In addition, Hirschowitz, Hitzemann, Kovasznay, et al. (1989) showed that lithium could produce a decrease in ethanol-induced intoxication, as measured by cognitive tests including the Minnesota Clerical Test. In summary, the neurobiological mechanisms of action of lithium are not clearly understood, although studies have found actions on phosphoinositide signaling, a major pathway for serotonin receptors.

Genetics

Abundant evidence indicates that alcoholism, like many behavioral disorders, derives from a complex interaction between the genetic makeup of an individual and his or her life experiences. Thus, adoption and twin studies consistently show evidence for a genetic contribution to alcohol dependence (Cloninger, 1987; Kendler, Heath, Neale, et al., 1992). In these same studies, however, the genetic effect was found not to be invariably associated with the development of alcoholism. Thus, identical twins may have one member with alcoholism and the other without alcoholism. This stands in contrast to diseases such as cystic fibrosis or Huntington's chorea where the genetic effect is in and of itself sufficient to lead to the illness.

The nature of the genetic vulnerability to alcoholism is not known. Some genes appear to confer protection from alcoholism by, for example, causing acute physical symptoms to be experienced when alcohol is consumed. Other genes may confer true vulnerability by enhancing the addictive process to alcohol, although the mechanism to achieve this is not known. Rather than one gene contributing to the vulnerability to alcoholism, it is likely that multiple genes are involved.

Some forms of alcoholism are quite likely not to be genetically based but instead derive from such environmental events as severe and chronic stress or peer and cultural influences. The variations in genetic and environmental risk help to explain the heterogeneity seen in alcoholism. They also suggest that no one medication is likely to be "best" or even always effective in treating patients with alcoholism, and that some forms of alcoholism may not respond to medication at all.

Summary

Detailed study of the neurochemistry and neuropharmacology of alcoholism is producing a more molecular view of alcohol's actions. Thus, alcohol can lead to a release of dopamine in reinforcing areas of the brain, and it affects serotonin, GABA, glutamate, opioids, and other neurotransmitters. These findings may provide a framework for understanding how medications might interact with disease biology to counteract loss of control, symptoms of protracted abstinence, craving, and relapse and thus enable clinicians to treat patients with alcoholism more successfully.

Epidemiology and Disease Burden

Natural History

According to the American Psychiatric Association (APA, 1994), alcohol dependence is a chronic disorder that manifests with a cluster of symptoms, including cognitive, behavioral, and physiologic symptoms that indicate an inability to stop drinking despite significant alcohol-related problems. In 1952, the APA developed the first edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM) (APA, 1952). However, not until the third edition (DSM-III and DSM-IIIR) did the APA provide rules for the diagnosis of alcohol problems, differentiating between alcohol abuse and dependence (APA, 1980, 1987). This differentiation between the two conditions continues in the DSM-IV and parallels that in the diagnostic criteria developed by the World Health Organization (WHO) in the tenth revision of the International Classification of Diseases (ICD-10) (WHO, 1993).

Alcoholism is considered to be a progressive disease with death representing a common outcome. Alcoholism causes premature death through overdose; through organic complications involving the brain, liver, heart, and other organs; and through its contribution to suicide, homicide, motor vehicle collisions, and other traumatic events (Morse and Flavin, 1992).

The average rate of progression from initial use of alcohol to overt alcohol dependence has been reported to be in the 6- to 8-year range (Schuckit, Tipp, Smith, et al., 1997). Inter-individual variation is considerable, however, and can range from rapid onset of dependence with quick deterioration to very slow progression from nondependent drinking to alcoholism over many years.

Progression of the illness is associated with increased health problems; interpersonal, social,and legal problems; and changes in the body's response to alcohol. For example, the individual who was able to drink large quantities of alcohol early in the course of illness may later develop the onset of alcohol-induced blackouts after only limited alcohol consumption. Nonetheless, long periods of abstinence can occur over the course of alcoholism including total cessation of drinking (Schuckit, Tipp, Smith, et al., 1997).

That the course of alcoholism varies is one indicator of heterogeneity within the broad diagnostic category of alcoholism. In fact, recent studies have found both genetic and pharmacological evidence in support of heterogeneity (Cloninger, 1987; Babor, Hofmann, DelBoca, et al., 1992). Data gathered from the adoption studies of Cloninger (1987) and other researchers indicate that the age of onset and progression of disease is partially determined by genetic factors. Babor, Hofmann, DelBoca, et al. (1992) have shown that patients presenting for treatment of alcoholism can be subtyped into distinctly different clinical populations based on a variety of clinical and historical variables such as age of onset, severity of dependence, and antisocial characteristics. Evidence from such studies supports the hypothesis that alcoholism may be better conceptualized as the "alcoholisms," an idea advanced by Jellinek (1960) decades ago. This concept suggests that different forms of alcoholism will likely respond differentially to various forms of pharmacological or psychosocial treatment. Although no evidence yet supports the concept of differential pharmacologic response among persons with alcoholism, the studies reviewed in this evidence report indicate that pharmacological responsiveness, when it occurs at all, is quite variable within the population of alcohol-dependent persons. Thus, the linkage between the heterogeneity of alcoholism and pharmacologic response deserves further study.

Disease Incidence and Prevalence

To determine the public health significance of alcohol dependence, policymakers and researchers typically rely on estimates of disease incidence or prevalence. Although the preferred measure of disease burden is incidence, which is the risk of developing the disease in a given period of time, studies of alcohol dependence typically report disease prevalence, which is the proportion of a population that is affected by a disease at a specific point in time (Rothman, 1986).

Prevalence rates of alcoholism in the U.S. population have been estimated through a variety of sampling techniques, but the definitions for alcoholism often vary. One of the more extensive studies that estimated prevalence based on direct interview is the Epidemiologic Catchment Area Survey sponsored by the National Institute of Mental Health (Regier, Farmer, Rae, et al., 1990). This survey involved interviews of a total of 20,291 adults aged 18 and older based on household surveys conducted between 1980 and 1984 in Baltimore, MD; St. Louis, MO; Durham, NC; New Haven, CT; and Los Angeles, CA. Residents of long-term mental hospitals, nursing homes, and penal institutions were sampled at higher probability rates than the community. The diagnosis of alcoholism was based on the DSM-IIIR (APA, 1987). The overall 6-month and lifetime prevalence rates of alcohol dependence were, respectively, 2.8 and 7.9 per 100 persons aged 18 and older.

Four national surveys that measured the 1-year prevalence of alcohol abuse and dependence were summarized in the 1997 Ninth Special Report to the U.S. Congress on Alcohol and Health (hereafter the Ninth Special Report), the 1988 National Health Interview Survey (NHIS), the 1990 National Alcohol Survey (NAS), the 1992 National Comorbidity Survey, and the 1992 National Longitudinal Alcohol Epidemiologic Survey (NLAES) (Ninth Special Report, 1997). As a result of the changing diagnostic criteria over time and differences in terminology between studies, the estimates of alcohol dependence are relatively inconsistent (Table 2). For example, using DSM-IV criteria, the NHIS and the NAS reported greatly divergent prevalence rates for alcohol dependence in males despite a 2-year time span between studies.

Table 2. One-year prevalence of alcohol dependence (DEP) reported by all national studies, in percentages, by sex.

Table

Table 2. One-year prevalence of alcohol dependence (DEP) reported by all national studies, in percentages, by sex.

More than 8 million Americans suffered from alcohol dependence in 1992 according to the NLAES (Grant, Harford, Dawson, et al., 1994). Alcohol dependence is more frequent in males and races other than black. The age range at highest risk are men and women of all races who are 18 to 29 years of age. The overall lifetime risk of alcohol dependence is estimated to be 9.6 percent for men and 3.2 percent for women (Grant, Harford, Hasin, et al., 1992).

Burden of Illness

Morbidity related to alcoholism is a significant problem in the United States. Based on data from the National Hospital Discharge Survey, approximately 1.5 percent of discharges from short-stay hospitals gave alcohol-related diagnoses as the first-listed diagnosis (Ninth Special Report, 1997); alcohol-related illness estimated from data on any-listed diagnosis would be much higher. Comparing discharge data with those from screening for alcohol disorders using several diagnostic instruments, Umbricht-Schneiter, Santora, and Moore (1991) found that alcohol-related problems were identified in only 7.4 percent of discharge diagnoses, whereas the screening data indicated that 22.4 percent of patients actually exhibited these problems. Thus, analyses of alcohol morbidity based on discharge diagnoses greatly underestimate the role of alcohol in morbidity. Evaluating alcohol-related discharge diagnoses does not account for conditions such as peptic ulcer disease, burns, and fractures that commonly have alcohol as a contributory factor. In short, these two estimates likely represent the lower and upper bounds for the true level of alcohol-related morbidity at least as reflected in hospitalization rates.

Mortality related to alcoholism and the use of alcohol is hard to estimate accurately because of the difficulty in identifying alcohol-related deaths. For example, many homicides involve alcohol, but these deaths may not be attributed, even partially, to alcohol. Similarly, many alcohol-related medical disorders that can lead to death, such as esophageal cancer or stomach cancer, are not coded as alcohol-related.

Two areas of alcohol-related mortality that are assessed regularly are traffic fatalities and cirrhosis. Recent evidence indicates that approximately 45 percent of the nearly 32,000 traffic fatalities that occur annually involve alcohol (Ninth Special Report, 1997). In 1992, cirrhosis accounted for 25,407 deaths in the United States, with alcohol consumption contributing to many of these deaths. The Centers for Disease Control and Prevention estimated alcohol-related mortality to be approximately 100,000 deaths annually in the United States (McGinnis and Foege, 1993).

Alcohol costs the United States an estimated $148 billion annually, taking health effects, lost productivity, and treatment of alcohol-related diseases into account (NIDA and NIAAA, 1998). For health care alone, people with alcoholism consume more than 15 percent of the national health care budget (Rice, Kelman, Miller, et al., 1986). Specialized private and public programs are only a small part of the total alcohol-related costs, approximately $3.8 billion in 1989 (Dayhoff, Pope, and Huber, 1994). The major costs are indirect, estimated at 75 percent of the total alcohol-related costs (McCrady and Langenbucher, 1996).

Treatment

Pharmacotherapy

The pharmacotherapy of alcoholism attempts to accomplish one or more of four goals:

1.

Treat the core dependence syndrome including craving, preoccupation, and loss of control.

2.

Enhance abstinence and minimize relapse by the threat of or the development of adversive consequences (or both) in response to alcohol consumption.

3.

Treat comorbid disorders that increase the likelihood of alcohol use.

4.

Treat the consequences of alcohol use such as protracted abstinence symptoms, cognitive impairment, and liver problems.

The possibility of using a medication to treat a behavioral disorder that involves the "voluntary" ingestion of a substance represents a point of view that appears to have originated in the latter half of the 20th century. In the mid-1940s, researchers serendipitously discovered a substance, disulfiram, that sensitized humans to the ingestion of alcohol; it produced headache, nausea, flushing, rapid heartbeat, weakness, and even collapse. It was quickly realized that disulfiram had the potential to modify the course of alcoholism by providing a psychological counter-force to the urge to drink and a physiological consequence should drinking occur. The FDA approved disulfiram for the treatment of alcoholism in the United States under the trade name Antabuse®. Antabuse® was widely prescribed, and many positive articles were written about its effectiveness; few carefully controlled studies were completed. Not until the 1980s were the results of placebo-controlled randomized controlled trials (RCTs) of disulfiram completed and published. Surprisingly, these trials were not uniformly supportive for a positive drug effect for disulfiram. However, the overall role of disulfiram in the treatment of alcoholism remains a complex issue involving the pharmacologic effect of the drug, the psychological effect of the drug, the beliefs of the practitioner and the patient, and the setting in which disulfiram is given. One area that would benefit from the completion of adequate placebo-controlled RCTs is the effectiveness of disulfiram when given under supervised circumstances. A body of mostly uncontrolled studies suggests that this methodology may be particularly effective (Brewer, 1993).

In the 1950s, with the advent of medications that were effective in treating patients with bipolar disorder, depression, and anxiety, questions arose regarding the ability of these medications to modify the course of alcoholism in a positive way. The self-medication hypothesis of alcoholism posits that alcohol is consumed to "treat" an underlying disorder of mood such as depression or anxiety. Conversely, certain mental states, such as mania, may lead to excessive alcohol consumption because of increased risk-taking or impairment of judgment. Therefore, medications that treat the underlying disorder should diminish the misuse of alcohol. As with disulfiram, the early studies of these medications consisted of case reports, case series, and open-label studies - and many of these reports were positive. However, with the completion of placebo-controlled, double-blind RCTs in primary alcoholics without preexisting and comorbid psychiatric disorders, the effectiveness of lithium, antidepressants, and the anxiolytic buspirone was not demonstrated. The results with regard to improvement in the symptoms of alcoholism in patients with mood or anxiety disorders have been more promising, but even here results are mixed.

As explained above, alcoholism is a primary disorder that is heterogeneous and derives from a complex interaction of genetic makeup and environmental experience. Furthermore, the majority of alcoholics do not have a comorbid mental illness such as depression, bipolar disorder, or an anxiety disorder. Therefore, some have reasoned that pharmacologic treatments can be developed successfully to address two aspects of this disorder: (1) the core syndrome of alcoholism (e.g., craving, loss of control, preoccupation with obtaining and using alcohol) and the consequences of heavy alcohol use and (2) the symptoms of protracted withdrawal (e.g., sleep problems and stress intolerance) that may persist long after acute withdrawal symptoms have subsided.

The development of such agents has been and is being greatly advanced by discoveries in the basic sciences of the underlying neurochemistry and neuropharmacology of alcohol and of the brain's reinforcement systems. Based on this work, investigations of agents that modify dopaminergic, serotonergic, opioidergic, GABAergic, and glutaminergic neurotransmission have been and are being undertaken to explore for a therapeutic effect in alcoholism (see Litten, Allen, and Fertig, 1996, for a recent review). It is of great importance to the field that two medications developed using this strategy, naltrexone and acamprosate, have shown efficacy for the treatment of primary alcoholism in placebo-controlled RCTs (see evidence tables). Larger studies of these and other agents are now being completed, and this body of work should change the landscape of the pharmacotherapy of alcoholism over the next 5 to 10 years.

Psychosocial Therapies

Psychosocial therapies have been the major form of therapy for alcoholism for decades. They include psychodynamic psychotherapy, behavioral therapies, cognitive therapies, motivational therapies, brief interventions, aversive therapies, group therapies, marital therapy, 12-step programs, and therapeutic communities, as well as various combinations of these modalities provided in a wide array of settings. One major debate in the field is the value of intensive inpatient programs vs. intensive (or less intensive) outpatient programs, with attention drawn to the long-term health outcomes and economic implications of these choices. Although RCTs for many of these therapeutic modalities are limited, several comprehensive reviews of this area have been completed (e.g., Miller and Hester, 1986; Finney and Monahan, 1996). On balance, the evidence supports the effectiveness of a number of psychosocial treatments for alcoholism; however, the issues of patient motivation and the heterogeneity of alcoholism have received only minimal attention.

Combined Pharmacotherapy and Psychotherapy

With the development of effective pharmacotherapies for alcoholism, the question of the interaction of pharmacotherapy with psychosocial therapy is emerging as an important avenue for treatment research. To date, very few trials have tested for differential effects of psychosocial therapies when combined with placebo-controlled pharmacotherapy, although pharmacotherapy studies typically (if not invariably) employ one form or another of psychosocial therapy because the latter is the "standard" treatment and ethically could not be withheld.

In one interesting, albeit complex, design, O'Malley, Jaffe, Chang, et al. (1992) examined the interaction of naltrexone vs. placebo with supportive therapy vs. relapse-prevention therapy. This research team found evidence for interactions - naltrexone and supportive therapy were associated with improved abstinence rates, whereas naltrexone and relapse-prevention therapy were associated with lower relapse rates, among those who were nonabstinent. Although these data need replication, they point to the possible synergistic effects of combining pharmacotherapy and psychosocial therapy, and they further underscore the heterogeneity of this patient population and this disorder.

Organization of This Report

The remainder of this evidence report is organized in the following sections. Chapter 2 provides details about our literature search and review methodology. Specifically included are the causal pathways for our key clinical questions; our approaches to conducting the systematic review, abstracting data from articles, maintaining quality control, and applying a quality rating system for individual articles; and similar details. In addition, we comment briefly on the peer review process and our decision not to conduct any supplemental analyses. Chapter 3 provides our results on the five main classes of drugs covered by this report (disulfiram, naltrexone, acamprosate, the serotonergic agents, and lithium). Chapter 4 provides some concluding remarks, and Chapter 5 offers our recommendations for a research agenda on alcohol dependence, focusing on remaining issues in treating adults with this disorder and on treating the illness in certain special populations (e.g., people less than 18 years of age, women, and ethnic groups).

All references cited in the report are included in the reference list. The report also includes three appendixes, evidence tables, and a bibliography, which is a complete list of the literature considered and used in developing the evidence report.