Duration of Pharmacotherapy and Length of Studies

Combination Therapies

Combination treatment regimens may be more effective than monotherapy, especially when medications with differing mechanisms of action are assessed. The gains that can be obtained with this approach are illustrated by the pharmacotherapy of acquired immunodeficiency syndrome (AIDS) where combination therapy is yielding significant improvements in outcome. The trade-offs with combination therapy may be increases in adverse effects of medications and direct costs. To the extent that interest is rising in the use of multiple pharmaceutical agents in a variety of medical disciplines, practitioners and patients will need to have a better sense of what harms, as well as benefits, should be expected.

We recommend that investigators direct their efforts to combination therapy in two ways. First, more attention should be given to searching specifically for combination treatments in new or forthcoming publications of ongoing and future trials. Second, studies on promising combination therapies should be deliberately designed and supported, so that systematic information can be marshaled about this aspect of care, rather than relying on what may appear to be rather serendipitous "combinations" in existing work. Combination pharmacotherapy studies may require special funding mechanisms, including strong Federal support, as they are less likely to be funded by the pharmaceutical industry. Of note is that the NIAAA has recently funded a multisite study of the combination therapy of naltrexone and acamprosate compared with each drug alone and with placebo in alcohol-dependent subjects.

Co-Interventions

A majority of investigators believe that psychosocial co-interventions are essential to the success of pharmacotherapy in alcohol-dependent individuals. Unfortunately, these interventions often are not described in sufficient detail to permit the interventions to be characterized adequately for comprehensive reviews and analyses of the existing literature or to allow replication of the intervention in other studies or settings. Thus, we make three recommendations on this point. The first two relate more to thorough reporting of results than to the conduct of the research, although by implication investigators will need to ensure that adequate documentation is kept about the co-interventions. The third recommendation relates to types of studies that should be carried out to broaden the scope of understanding about the use of pharmacotherapies in all types of settings and patients.

First, we recommend that all future studies describe the timing and content of the intervention and the training of the therapists. Second, we recommend that those reporting on the studies include an assessment of patient compliance with the psychosocial intervention. Third, testing interventions that can be performed outside specialized treatment centers, such as brief motivational enhancement, will substantially aid the generalizability of information about the effectiveness (if not efficacy) of pharmacologic treatments for patients with this disorder. Thus, we recommend that investigators begin to design, and funding agencies support, studies of interventions that heretofore have not been well studied (if at all).

Treatment in a Variety of Settings

As noted elsewhere, most of our literature review and synthesis has focused on efficacy studies - i.e., the knowledge base derived from randomized controlled trials (RCTs). As is well understood, such studies may have serious limitations in terms of constraints on the types of patients included in the trials; they also may be carried out in specialized centers by practitioners with exceptionally high professional qualifications. In short, they may not provide as much information on the effectiveness of the pharmacologic agent among "ordinary" patient populations in "average" settings and circumstances.

The practical drawback to this fact is that, in the case of pharmaceuticals, once the FDA has approved a medication, any physician in any practice setting may prescribe it. Thus, we can expect to see considerable diffusion of at least certain kinds of medications to patient groups in which the drug in question has not been tested (or at least tested adequately). Moreover, most alcohol-dependent patients currently are not under treatment in centers similar to those in which the reviewed trials took place.

In short, we cannot generalize our findings and conclusions (much) beyond the existing trial data. This leaves important questions about the extent to which our findings will be applicable to the broader population of patients, centers and settings, and clinicians for whom the various medications reviewed here might be considered. For example, we might ask: Are similar outcomes possible in the offices of trained primary care physicians or in community mental health centers?

To address these questions, we recommend that a broader set of studies, conducted as trials if possible or as well-designed longitudinal cohort studies, be carried out in a wider array of typical settings in which patients with alcoholism receive care. These studies should include primary care physicians as the principal source of care; in some circumstances, other primary care clinicians may also be involved in the studies (e.g., psychiatric nurse practitioners for co-interventions). The aim is to replicate the types of trials that already have been conducted on the key drugs (particularly naltrexone and acamprosate), to the extent possible, with studies that will inform providers, patients, and policymakers about the effectiveness of these agents as they may be used by physicians, other providers, and treatment centers of all types. This body of work will provide information about the pharmaceutical interventions themselves; it also will yield data that may indicate whether this nation will have to devote substantially more resources to upgrading substance abuse treatment centers in the coming years.

Common Outcome Measurements and Definitions

We were hindered in our attempts to perform a meta-analysis of the efficacy of pharmacotherapy for alcohol dependence in part by the multiple and disparate outcomes that were used in the trials. The use of various outcomes across studies and differing methods of measuring those outcomes impede the clinical community in evaluating the relative efficacy of treatments. Similarly, differences in defining relapse rates and heavy drinking and in measuring craving also create problems for efficiently abstracting, comparing, or interpreting data across this body of literature. The problems are in two areas: ambiguous descriptions of the outcomes that are measured (including different usages of the same term, such as "relapse") and different choices of endpoints (some of which may be idiosyncratic to a particular study).

We recommend that investigators, professional societies involved in alcohol research, and the appropriate Federal funding agencies work together with investigators to ensure that future studies will have reasonably comparable outcomes. We also recommend that some attention be given to ensuring that outcomes important to patients and their families always be included in these studies. Very precise definitions of outcomes that are measured (i.e., clarity in describing the outcomes) will foster this goal. We also recommend that journal editors insist on unambiguous descriptions of all outcomes, so that even unusual or special ones can be completely understood by readers and reviewers of these articles. This is particularly germane to the measurement of craving.

High Dropout Rates

Many alcohol studies are hindered by high dropout rates. Many trials had dropout rates of 40 percent or more. This level of attrition seriously undermines the confidence that one can place in the work. This is especially true in the following circumstances:

• The original sample sizes are small.
• The dropout rates between experimental and control or comparison groups differ.
• Those lost to followup cannot be compared adequately with those who complete the study on a full range of baseline characteristics.
• Those lost to followup for different reasons (e.g., death, institutionalization, adverse side effects, or unknown) cannot be compared with each other on baseline characteristics.

We recognize that these patient populations (especially those outside specialized centers and those included in community-oriented or effectiveness studies) may be difficult to retain in studies, and that the challenge of doing so is only greater as researchers attempt to address some of our other recommendations (e.g., a broader set of treatment settings, longer treatment, and longer followup). Nevertheless, we recommend that alcohol investigators work with survey researchers to develop and apply methods of reducing dropout and loss to followup. In addition, they should devote some resources simply to determining the reasons for dropout from their efficacy trials or effectiveness studies, with a view to providing this information in published articles. This will enable readers and reviewers to gain a better appreciation of the dynamics of the studies (and the patient population), but it also will provide feedback on which issues need to be addressed (especially problematic circumstances affecting patients) that would permit researchers to target resources effectively to reduce dropout rates in future studies.

Pharmacokinetics

As can be inferred from our presentation in Chapter 1 on the presumptive mechanisms by which the various pharmacologic agents studied here affect patients with alcohol dependence, especially at the biochemical level, little is understood about exactly what these agents do and why (at least in humans). The theory on these matters is quite rich; the empirical data are relatively thin. Part of the deficit in the knowledge base relates specifically to the relationship of blood levels and active or toxic metabolites to treatment response; data of these sorts are essentially absent in the reported trials.

We recommend that alcohol researchers address this information gap by incorporating data into clinical trials on the pharmacokinetics of the compound under study. This would include basic information such as steady-state levels of the parent compound, measurement of metabolites as appropriate, and investigation of how drug-drug interactions may affect blood levels. Such data may lead to insights as to why some patients do not respond to a given agent while other patients develop toxicity. Advances in these areas will help the pharmacotherapy of alcoholism mature as a discipline.

Relationship of Pharmacotherapy to Patient Heterogeneity and Disease Biology

As can be inferred from the recommendations to this point, we believe that the heterogeneity of patients with alcoholism is substantial; this heterogeneity includes genetic, biological, environmental, and sociodemographic characteristics. Similarly, the underlying biology of this disease is itself now thought to exhibit signs of heterogeneity (as noted in Chapter 1 about the concept of "alcoholisms"). Putting these facts together suggests that how this disorder (or disorders) plays out in different populations may be a remarkably complex, multifaceted process. Coupling this observation with the multiple mechanisms by which pharmacotherapies may act at the basic biological (or psychological) level suggests that more attention must be paid to the combinations of patient characteristics, the disease they exhibit, and the specific treatments that are provided. Thus, we recommend that researchers devote more attention and resources into characterizing the biological/clinical, sociodemographic, and other attributes of their patient populations, specifying to the extent possible the manifestations of alcoholism(s) in these patients, and measuring the relationship between these sets of factors (on the one hand) and the response to pharmacotherapies (on the other). Some areas of particular concern include the growing number of patients with dependence on multiple substances and what implications this codependency has for pharmacotherapy and the effects of psychiatric comorbidity on pharmacotherapy outcomes. We recognize that this area of investigation may require maturation of the research on the heterogeneity of alcoholism.