Publication
Article
Author(s):
This article briefly describes pharmacotherapy for the treatment of alcoholism and discusses strategies to address treatment challenges like medication nonadherence.
Psychiatric Times
March 2007
Vol. XXIV
Issue 3
This article was originally presented as an independent educational activity under the direction of CME LLC. The ability to receive CME credits has expired. The article is now presented here for your reference.
After reading this article, you will be familiar with:
-Identifying the typical daily doses of FDA-approved medications for the treatment of alcohol dependence.
-Describing the significance of medication adherence on treatment outcomes in patients with alcohol dependence.
Detailng strategies for recognizing, probing and improving medication nonadherence in patients who are prescribed a medication for a chronic illness like alcohol dependence.
Describing novel options that might enhance medication adherence.
Who will benefit from reading this article?
This activity was designed to meet the continuing education needs of psychiatrists and other physicians, physician assistants, registered nurses and advanced practice psychiatric nurses. Other mental health care professionals may find this activity informative.
Alcoholism is a chronic and serious disorder with often devastating consequences. One out of three families in the United States is negatively impacted by a family member's excessive drinking and painful drinking-related problems.1 Sadly, alcohol disorders are largely undertreated. The National Institute of Alcohol Abuse and Alcoholism (NIAAA) estimates that of the approximately 7.9 million people in the United States who suffer from alcohol dependence, only approximately 2.2 million people seek treatment.2
For decades, the standard in the United States for treating alcohol dependence, following detoxification, has been addiction-specialty counseling and regular attendance at mutual support groups like Alcoholics Anonymous. (Only a percentage of patients with alcohol dependence seek medical treatment for their disorder.3 This is highlighted by the fact that pharmacologic agents for the treatment of alcohol dependence have been available for over 10 years. However, historically, only a small fraction of patients are ever prescribed a pharmacologic agent-Ed.) Successful treatment recovery rates, while notable, still leave a large proportion of people who continue to drink heavily even in treatment or who rapidly relapse following treatment. Analysis of data from the NIAAA 2001-2002 U.S. National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) found that among alcohol-dependent patients who have received treatment, 54.5% relapsed soon after completing treatment. 2 Initiating pharmacotherapy with counseling, in a number of cases, has been shown to be a valuable tool toward reducing relapse rates and increasing abstinence in alcohol-dependent patients. The NIAAA advises clinicians to consider initiating medications that have been approved by the U.S. Food and Drug Administration in patients who are still drinking or are abstinent but continue to experience slips or intense cravings for alcohol.4
In addition, however, it is recognized that the success of pharmacotherapy is largely dependent on an individual's adherence to taking the medication as prescribed. Daily pill-taking over an extended period of time can often be problematic, not just for patients with addictions, but for anyone suffering from a chronic disorder. Higher levels of medication nonadherence have been associated with patients suffering from psychiatric disorders, including addictive disorders.5 This article briefly describes the FDA-approved medications for the treatment of alcohol dependence and discusses strategies to address treatment challenges like medication nonadherence.
Initiating pharmacotherapy
In the last decade, as more knowledge has been discovered about the neurobiology of addiction and the effects of alcohol on the brain, clinical studies have been completed on initiating medications with counseling (types of counseling have ranged from medical management and advice to addiction-specialty therapies). These studies have been systematically conducted for the purpose of evaluating the advantage of pharmacotherapy in improving early and long-term outcomes. Essentially, medications are evaluated as to whether or not they provide an incremental advantage to counseling, or whether or not they can be paired with a relatively modest type of medical management counseling as an alternative to more intensive and specialized addiction therapy. To date, the FDA has approved four medications for the treatment of alcohol dependence; they are listed in the Table with other relevant information.
Disulfiram's (Antabuse) mechanism of action is different from the other subsequent medications that have been approved by the FDA for alcoholism. Disulfiram blocks aldehyde dehydrogenase, an enzyme needed to convert acetaldehyde to acetate for the body to fully metabolize alcohol.6 When someone taking disulfiram consumes alcohol, its enzyme-blocking action creates disagreeable physical symptoms, primarily nausea and vomiting. The usual daily dose of disulfiram prescribed today is 250 mg/day, but the unpleasant symptoms are dose dependent. That is, the aversive reaction escalates with a higher daily dose of disulfiram and/or with greater quantities of alcohol. Severe ethanol-disulfiram reactions, even death, have been reported, but were almost always associated with much higher daily doses (e.g., 1000 mg/day to 3000 mg/day).
While usually the unpleasant feelings will cause most people to stop drinking, disulfiram prescribers primarily believe that it is the psychological threat of discomfort that deters a disulfiramtreated patient from drinking any alcohol, which, in turn, removes the patient's chances of actually experiencing aversive symptoms.6 When patients are highly motivated or are supervised in their taking of disulfiram, it has been reported that drinking is avoided and thus aversive symptoms are not experienced. Treatment regimens including incentivedriven interventions or disulfiram supervision by physicians have shown beneficial treatment outcomes, such as decreases in drinking and increases in the rate of abstinence.5 However, in many primarily unsupervised situations, patients essentially have not accepted this medication, and, rather than discontinuing their alcohol intake, they stop taking disulfiram. Fuller and colleagues7 reported an astoundingly high nonadherence rate of 80% in a large multisite study of alcohol-dependent patients who were asked to take 250 mg/day of disulfiram or placebo. See Suh et al.6 for more information on disulfiram.
Naltrexone (ReVia), an opioid receptor antagonist that originally had been approved for the treatment of opiate addiction in 1984, was the second medication to be approved by the FDA for the treatment of alcohol dependence in 1994. It utilizes a novel mechanism of action to reduce alcohol use. That is, naltrexone is thought to decrease drinking in alcohol-dependent individuals by blocking the release of alcohol-inducing endorphins via neural reward pathways. This action is thought to reduce the high or reinforcement that alcohol-dependent people report feeling when they drink alcohol. Naltrexone has also been shown to reduce craving for alcohol, which for motivated patients can promote long-term abstinence. Finally, naltrexone has proven to be a relatively safe medication for people suffering from alcoholism, with the most frequent side effect reported being nausea, which generally dissipates by the end of the second or third week of treatment. Also, when naltrexone-adherent patients with initial elevated liver enzyme levels reduce their alcohol intake over the treatment course, they show significant reductions in pre-treatment liver enzyme levels.8 The reduction in liver enzymes widely reported in alcohol-treatment populations is in direct contrast to naltrexone's black box warning of potential hepatoxicity, which actually was derived from a few very early studies using a 300 mg/day dose of naltrexone to treat obesity (not an alcohol-disordered population).
The usual daily dose of naltrexone prescribed today for alcohol treatment ranges from 50 mg/day to 100 mg/day. Naltrexone's efficacy in reducing heavy drinking (heavy = ≥5 drinks a day for men; ≥4 for women) and promoting abstinence has been demonstrated in numerous randomized, placebocontrolled clinical trials worldwide.9 Nonetheless, naltrexone has not been embraced by the clinical community since its original approval in 1994, due in part to problems with getting many patients to regularly take their daily pills for any length of time. Relapse rates are high in patients who are nonadherent to taking naltrexone as prescribed-a situation that usually appears to clinicians as patients unresponsive to naltrexone.5
Acamprosate (Campral), like naltrexone, affects neural reward pathways to reduce craving for alcohol and alcohol use, although this medication works through a different neurobiological remediation, i.e., acamprosate targets glutamatergic pathways. Acamprosate is thought to promote abstinence by first reducing physical and psychological discomfort often experienced during alcohol withdrawal, such as sweating, anxiety and sleep disturbances. Secondly, it is thought that acamprosate may have abstinent-sustaining properties. A number of randomized, controlled clinical trials conducted in Europe provide support for acamprosate's efficacy in maintaining abstinence.10
However, in a recently published U.S. randomized, placebo-controlled clinical trial of 1998 mg/day acamprosate for alcohol dependence in 601 patients, acamprosate was associated with higher rates of abstinence only in patients whose initial goal was to stop drinking completely.11 This study suggested that acamprosate might best target alcohol-dependent patients with greater motivation for abstinence. The usual daily dose of acamprosate prescribed today for alcohol treatment is 1998 mg/day (essentially two capsules of 333 mg tid).
Related to both naltrexone and acamprosate, the most recently published NIAAA-supported study, the Combining Medications and Behavioral Interventions for Alcoholism (COMBINE) study, is the largest investigation of the efficacy of pharmacotherapy in treating alcohol dependence that has been published to date.12 This study investigated the combination of naltrexone and acamprosate, and each alone, in treating 1,383 alcohol-dependent patients at 11 sites across the country. The findings of this large, well-controlled trial revealed that although all patients enrolled in this treatment study showed a considerable decrease in drinking, those who received naltrexone with a very basic type of medical counseling (called medical management [MM]13) reported the highest number of abstinent days while in treatment (approximately 80.6%).
Naltrexone with MM is a pharmacopsychosocial treatment combination that can be easily administered by primary health care providers, potentially reaching more patients than can be serviced in specialty care settings. Contrary to the original hypothesis, the results from the COMBINE study did not support better efficacy of a naltrexone and acamprosate combination, although this combination was safe and relatively well tolerated.12
With all three of these medications, there are a formidable number of patients in both clinical trials and private practices who skip medication doses and generally have difficulties taking pills on a daily basis for any length of time. These patients traditionally have poor outcomes and appear as medication nonresponders, when, in fact, they may not have been adequately exposed to the active medication that would allow clinicians to accurately assess treatment response. Most physicians feel their patients are good about taking prescribed medication(s), but these same clinicians also say that they cannot readily prove this assumption. All attempts to gather adherence data in research trials have documented that the prevalence of medication nonadherence in alcoholdependent patients is less than optimal and typically higher than anticipated.
If a patient's possibilities for being medication nonadherent can be minimized or bypassed, then patient nonadherence as a barrier to treatment response would be eliminated. This was the rationale for developing the fourth FDA-approved medication: a long-acting extendedrelease formulation (once-monthly) injection of naltrexone (Vivitrol). Each injection delivers 380 mg of naltrexone that remains in the body at therapeutic plasma concentrations for approximately a month's duration. This delivery system removes the patient burden of daily decision making. In addition, this type of delivery system for treating an alcohol disorder allows clinicians to be certain that their patients are taking and, hopefully, benefiting from treatment every day for 30 days.
Data have been reported that treatment response to injectable naltrexone is observable after only one injection, i.e., in the first month of treatment, according to a recent study.14 Thus, given that medication adherence is a non-issue in the month following a naltrexone injection, this type of innovative delivery system provides a simple and direct way for the clinician to evaluate whether or not the patient is responding to treatment.
Evidence for the efficacy of injectable naltrexone in reducing heavy drinking in the majority of patients, and in also maintaining abstinence in a number of patients, comes from a large, national, multisite study of two doses (380 mg or 190 mg) or placebo in 627 alcoholdependent participants who were given six injections over a six-month treatment period.15 Patients treated with the 380- mg dose monthly (the FDA-approved dose) demonstrated a greater reduction in heavy drinking than placebo-treated patients who had received counseling and a dummy injection.
Medication nonadherence
As we acquire promising pharmacotherapies that can be initiated with counseling for treating alcohol dependence, there are many challenges and barriers to fully introducing medication into mainstream clinical practice in the medical community. As mentioned above, the foremost challenge is a patient's nonadherence to prescribed treatment. All physicians understand that their patient's adherence to their medication(s) and attendance at their counseling sessions are critical for achieving successful outcomes. However, in practice, it has been demonstrated time after time how difficult it can be to ensure patient adherence to treatment, especially in chronic illnesses. For example, studies of insulin adherence in diabetes treatment report poor medication adherence in nearly 40% of patients, and the rate of medication nonadherence is as high as 50% in patients with hypertension.5
Adherence to treatment in alcoholdependent patients is no different from the rates reported above. Nonadherence rates, which vary by the type of medication and by the adherence definition, have ranged from 20% to 60% for daily naltrexone and as high as 80% for disulfiram.7 A recent meta-analysis found an average medication nonadherence rate of approximately 47% in acamprosate studies, which ranged from three to 24 months in duration.16
The Figure provides an illustration of relapse rates in a study group from the University of Pennsylvania that combined two double-blind, placebocontrolled clinical trials of 50 mg/day of naltrexone for 12 weeks to treat alcoholdependent outpatients.17 In the Figure, a patient who relapsed was defined as one who drank ≥5 drinks on one occasion. A patient who was medication adherent was defined as one who attended at least 80% of counseling sessions and took medications as prescribed. The data in the Figure illustrate the critical mediating role of medication adherence in evaluating treatment outcomes in daily pill-takers receiving medication with counseling for their alcohol dependence. That is, the proportion of patients who relapsed was relatively high and comparable among naltrexone-treated (42.9% relapsed) and placebo-treated (40% relapsed) patient groups in the medication nonadherent patients, as well as in placebo-treated patients who were medication adherent (38.6% relapsed). However, the naltrexonetreated, medication-adherent patient group had a dramatically smaller relapse rate of only 10%.
Thus, while the problem of medication nonadherence in alcohol-dependent patients is comparable to other disorders, it remains a formidable one for this patient population. This problem could cause the incorrect interpretation that a medication does not work, when in fact it was not taken as prescribed. The challenge is then how do we best identify, monitor and change medication nonadherence?
Clinicians need to inquire at each patient visit about patient pilltaking: Are there any problems or not? However, even when clinicians inquire at each visit about pill-taking problems, patients' responses may not be uniformly accurate. Identification of medication nonadherence may be the most difficult task the clinician faces, as most physicians want to trust that their patients are taking their medications as prescribed; and most patients want to tell their physicians that they are taking their medications according to their doctor's prescription. However, all published research data on this issue suggest that, overall, a number of patients with chronic disorders fail to take their medication as prescribed, with the most common occurrence being that patients just stop taking their medication(s).
Nonadherence in a patient is probably best identified by the clinician who proactively discusses nonadherence with the patient as a collaborator in trying to achieve 100% medication adherence. In addition, before nonadherence has a chance to happen, pitfalls associated with nonadherence and helpful strategies that produce good adherence can be discussed with the patient. Importantly, it is wise to establish a system for routinely monitoring medication adherence at each visit. This can be done using simple techniques such as asking patients to bring their pill bottles or blister cards to each visit, and/or asking them to keep a monthly calendar, indicating on each day the time that they took their medicine(s). Prevention and improvement of medication nonadherence can occur for some patients simply by the clinician routinely expressing interest and instructing patients on ways to selfmonitor their pill-taking as part of their treatment.
A selection of tools has been created to help monitor and validate patient pilltaking. 5 Of course, all of these tools vary in their cost, dependability and clinical function. The more frequently used tools include medication blister cards, where each dose per day is clearly marked and there is a clear plastic encasing per dose that indicates at a glance whether or not the dose was taken out of the card; Medication Event Monitoring System (MEMS) Caps, where each bottle cap encases a computer chip that records the day and time of each bottle cap opening, a sort of proxy for a patient taking their pills; B-vitamin riboflavin markers; and medication blood plasma level monitoring. These tools can be used in conjunction with routine clinician-patient discussions that target establishing true adherence rates. In addition, the clinician talking to the patient can resolve discrepancies between patient reports and any monitoring device. While none of these tools should be considered a gold standard, they all have their unique assets and liabilities and many of these can be helpful toward improving patient adherence. Medication blister cards possibly have the most number of desirable features and also have clinical utility as they are relatively inexpensive, easy for patients to use and transport, and appear to assist both the patient who wants to take the pills as prescribed and the practitioner who wants to monitor adherence.
At the start of pharmacotherapy, clinicians should proactively probe patients for typical medicationnonadherence risk factors. Once a pattern of medication nonadherence is identified, establishing more exact reasons for an individual patient's medication nonadherence should be probed so that strategies to improve medication adherence can be tailored to the individual patient and incorporated at each patient visit. As a word of caution, it is especially important to probe for reasons beyond the most socially acceptable and frequently stated reason of "I forgot." For example, if the patient truly forgot to take the pills because, say, they are not used to taking pills regularly, then strategies could be suggested like placing sticky notes on a bathroom mirror, patient's toothbrush or orange juice carton. However, if the problem is not simply forgetting, a clinician's attempt to organize the patient with systematic reminders, as just described, may not be useful.
There are many reasons for medication nonadherence, and most reasons are not easy for patients to admit. Some patients also may have difficulties identifying and/or articulating the actual reason(s) for not following the prescribed medication regimen. However, a more in depth discussion can be revealing. For example, it has been reported that patients who have better social support or believe that the medication will be effective tend to show better adherence.5 Worse adherence is predicted in patients who report side effects from treatment (particularly nausea or fatigue) or who abuse other drugs. In addition, adherence may be adversely affected by the treatment setting (e.g., a long wait time for appointments or staff or others at the facility who create an uncomfortable atmosphere or pessimistic outlook for recovery).
In summary, most patients' problems with medication adherence, if identified and probed, may be correctable with clinician diligence and advice. Clinician strategies for correcting medication nonadherence in a patient may be as simple as talking with the patient about things to do to overcome individual challenges. On the other hand, structured psychosocial interventions may aid in alleviating medication nonadherence beyond what can be achieved via a continued monitored basis or clinicianpatient dialogue.
Medical management is a manualbased, 13,18 minimally intensive intervention that can be employed by private, office-based medical practitioners. The goal of MM is to support the use of pharmacotherapy while providing individualized advice to patients to help them stop drinking. In addition, the MM clinician creates a dialogue with the patient meant to specifically increase medication adherence across treatment. All of this is done in a supportive atmosphere that conveys optimism for recovery.
The initial MM visit requires almost an hour to complete and involves reviewing results of the initial evaluation with emphasis on drinking-related problems and need for treatment. It includes patient education on alcohol dependence and the rationale for treating this disorder with medication. The clinician also gives the patient instructions on dosing and side-effect management, constructs a medication adherence plan for the patient to follow, and provides referral to local support groups. The plan for follow-up MM visits is also reviewed at the initial visit. Follow-up visits (lasting 15 to 30 minutes) involve a brief check of the patient's medical status, as well as the patient status regarding drinking and pill-taking. Advice given at followups is tailored to the current status of the patient.
Conclusion
Alcohol dependence is a treatable disease, but the proportion of affected individuals who come to treatment is less than 30%. The low proportion of patients seeking treatment may be due in part to the fact that successful treatment recovery rates, while notable, still leave a large number of treated people who rapidly relapse. In the last decade, advanced knowledge of the neurobiology of addiction plus the FDA approval of new medications for treating alcohol dependence provide some optimism that pharmacotherapy may be a valuable part of treatment for decreasing heavy drinking and maintaining abstinence and sobriety in many cases. Nonetheless, there are still challenges, with the primary barrier to successful treatment response being that a formidable number of patients tend to skip doses or stop their medication altogether for a variety of reasons. This, of course, ultimately compromises chances for a good treatment response. The most recent FDA-approved pharmacotherapy, however, is a long-acting, once-a-month injection that relieves patients of the daily decision to take their medication, as well as for the need for clinician monitoring of patients' pill adherence. This type of delivery system also aids the physician in evaluating the patient's response to treatment, i.e., in the initial clinical trial, positive changes in reducing alcohol use and craving were seen as early as the first injection in most patients.
An accumulation of clinical trial results have now indicated that following detoxification for alcohol dependence, initiating pharmacotherapy (FDAapproved for alcohol dependence) with counseling in the rehabilitation/recovery phase to treat alcohol dependence is very important to consider if we wish to provide recovery to more of our patients and, hopefully, attract more alcoholdependent patients to seek treatment. As new advances that change standard treatment unfold, it is essential to be attuned and prepared for the challenges that pharmacotherapy brings to our patients as we move forward with innovative addiction treatment in the 21st century.
The author, Dr. Pettinati, is professor of psychiatry and director of the division of treatment research in the Center for the Studies of Addiction at the University of Pennsylvania School of Medicine. Dr. Pettinati's co-author, Ms. Gallis, is a research assistant and the regulatory and monitoring coordinator at the Center for the Studies of Addiction at the University of Pennsylvania School of Medicine.
Disclosure:
Dr. Pettinati receives grants/research support from Alkermes, Inc., AstraZeneca, Bristol-Myers Squibb Company, Cephalon, Inc., Forest Laboratories, and Ortho-McNeil Pharmaceutical, Inc. She is also on the Advisory Board and Speakers’ Bureau and a consultant for Alkermes, Inc., AstraZeneca, Cephalon, Inc., and Forest Laboratories.
Ms. Gallis has indicated that she has nothing to disclose.
The editors of this educational activity have no financial relationship with the firm providing support in part through an unrestricted educational grant.
The content of this educational activity is determined by the authors of the piece and by the editors. The opinions expressed herein are not necessarily those of Cephalon, Inc., Alkermes, Inc., or the publishers.
References1. Dawson DA, Grant BF. Family history of alcoholism and gender:their combined effects on DSM-IV alcohol dependence and majordepression. J Stud Alcohol. 1998;59:97-106.
2. Dawson DA, Grant BF, Stinson FS et al. Recovery from DSMIValcohol dependence: United States, 2001-2002. Addiction.2005;100:281-292 [see comments].
3. Cohen E, Feinn R, Arias A, Kranzler HR. Alcohol treatment utilization:findings from the Natinal Epidemiologic Survey on Alcohol andRelated Conditions. Drug Alcohol Depend. 2007;86:214-221.
4. Helping Patients Who Drink Too Much: A Clinician's Guide-2005Edition. Available at: http://pubs.niaaa.nih.gov/publications/Practitioner/CliniciansGuide2005/guide.pdf. Accessed Dec.21, 2006.
5. Pettinati HM. Improving medication adherence in alcoholdependence. J Clin Psychiatry. 2006;67(suppl 14):23-29.
6. Suh JJ, Pettinati HM, Kampman KM, O'Brien CP. The statusof disulfiram: a half of a century later. J Clin Psychopharmacol.2006;26:290-302.
7. Fuller RK, Branchey L, Brightwell DR et al. Disulfiram treatmentof alcoholism. A Veterans Administration cooperative study.JAMA. 1986;256:1449-1455.
8. Anton RF, Moak DH, Latham P et al. Naltrexone combinedwith either cognitive behavioral or motivational enhancementtherapy for alcohol dependence. J Clin Psychopharmacol.2005;25:349-357.
9. Pettinati HM, O'Brien CP, Rabinowitz AR et al. The status ofnaltrexone in the treatment of alcohol dependence: specific effectson heavy drinking. J Clin Psychopharmacol. 2006;26:610-625.
10. Mann K, Lehert P, Morgan MY. The efficacy of acamprosatein the maintenance of abstinence in alcohol-dependentindividuals: results of a meta-analysis. Alcohol Clin Exp Res.2004;28:51-63.
11. Mason BJ, Goodman AM, Chabac S, Lehert P. Effect of oralacamprosate on abstinence in patients with alcohol dependencein a double-blind, placebo-controlled trial: the role of patientmotivation. J Psychiatr Res. 2006;40:383-393.
12. Anton RF, O'Malley SS, Ciraulo DA et al. Combined pharmacotherapiesand behavioral interventions for alcohol dependence:the COMBINE study: a randomized controlled trial. JAMA.2006;295:2003-2017 [see comments].
13. Pettinati HM, Weiss RD, Dundon W et al. A structured approachto medical management: a psychosocial intervention to supportpharmacotherapy in the treatment of alcohol dependence. J StudAlcohol Suppl. 2005;:170-178; discussion 168-169.
14. Ciraulo D, Pettinati HM, Dong Q et al. Early onset of effectof injectable extended-release naltrexone (XR-NTX) pharmacotherapyfor alcohol dependence. Poster 13. Presented at theannual meeting of the American Society of Addiction Medicine.May 4-6, 2006; San Diego.
15. Garbutt JC, Kranzler HR, O'Malley SS et al. Efficacy andtolerability of long-acting injectable naltrexone for alcoholdependence: a randomized controlled trial. [Published erratum inJAMA. 2005;293:1978; 293:2864.] JAMA. 2005;293:1617-1625[see comments].
16. Bouza C, Angeles M, Munoz A, Amate JM. Efficacy andsafety of naltrexone and acamprosate in the treatment of alcoholdependence: a systematic review. [Published erratum in Addiction.2005;100:573.] Addiction. 2004;99:811-828 [see comment].
17. Pettinati HM, Volpicelli JR, Pierce JD Jr, O'Brien CP. Improvingnaltrexone response: an intervention for medical practitioners toenhance medication compliance in alcohol dependent patients.J Addict Dis. 2000;19:71-83.
18. Pettinati HM, Weiss RD, Miller WR et al. COMBINE MonographSeries, Volume 2. Medical Management Treatment Manual: AClinical Research Guide for Medically Trained Clinicians ProvidingPharmacotherapy as Part of the Treatment for Alcohol Dependence.Bethesda, Md.: NIAAA. DHHS Publication 2004 No. (NIH) 04-5289. Available at: http://pubs.niaaa.nih.gov/publications/combine/Combine%202.pdf. Accessed Dec. 21, 2006.