IX. Review and Synthesis of Available Evidence

text A A A

A. Nicotine Dependence

1. Somatic treatments

Pharmacotherapies for nicotine dependence can be divided into NRTs and nonnicotine medications. The following sections describe these therapies in terms of treatment goals, efficacy in smoking cessation, side effects, and implementation. For more information, the reader is referred to descriptive (602, 706, 750, 1544–1548) and meta-analytic (789, 790, 793, 795, 1549) reviews.

a) Nicotine replacement therapy
+ (1) Goals

The primary goal of NRT is to relieve withdrawal symptoms when patients stop smoking and thus allow them to focus on conditioning factors that typically trigger a relapse to smoking. After the acute withdrawal period, NRT is reduced gradually so that little withdrawal should occur. A secondary goal of NRT may be to reduce smoking, but this approach has not been well studied.

+ (2) Description of products
+ a. Nicotine transdermal patch

The nicotine transdermal patch takes advantage of the ready absorption of nicotine across the skin (799, 1550). It is available in four formulations: three 24-hour patches and one 16-hour patch for use while an individual is awake. Treatment initiation typically uses a 21- to 22-mg 24-hour patch or a 15-mg 16-hour patch. Patches are applied on the first morning of smoking cessation and then each morning thereafter. The nicotine is slowly absorbed so that on the first day venous nicotine levels peak 6–10 hours after administration. Thereafter, nicotine levels remain fairly steady, with a decline from peak to trough of 25%–40% with 24-hour patches (799). The 16-hour nicotine patches demonstrate similar nicotine pharmacokinetics after discontinuation of smoking. Nicotine levels obtained with the use of patches are typically half those obtained by smoking (799).

After 4–6 weeks, the nicotine patch dose is tapered to a middle dose (e.g., 14 mg/24 hours or 10 mg/16 hours) and then tapered again in 2–4 weeks to the lowest dose (7 mg/24 hours or 5 mg/16 hours). The recommended total duration of treatment is usually 6–12 weeks (71, 1550). Most, but not all, studies indicate abrupt discontinuation of the patch usually causes no significant withdrawal so that tapering may not be necessary (71). There are now two nicotine patches available over the counter: a 21-, 14-, and 7-mg strength patch (24-hour application) and a 22- and 11-mg strength patch (16-hour application). The 15-mg patch does not require tapering.

+ b. Nicotine gum

Nicotine ingested through the gastrointestinal tract is extensively metabolized on first pass through the liver (1551). In addition, nicotine is a gastrointestinal irritant that makes orally ingested nicotine an unpleasant treatment for patients who are already experiencing nicotine withdrawal. Nicotine gum (nicotine polacrilex) avoids this problem via buccal absorption (1551). The gum, which is available as an over-the-counter medication, contains 2 or 4 mg of nicotine that can be released from a resin by chewing (886).

When compared with nicotine delivery via smoking, nicotine gum produces lower steady-state blood levels of nicotine and does not reach peak levels of nicotine absorption for >30 minutes (1551). Because cigarette nicotine is absorbed directly into the arterial circulation, arterial blood levels from smoking are 5–10 times higher than those from the 2- and 4-mg gum (1552). Venous nicotine levels from the 2- and 4-mg gum are about 33% and 67% of the steady-state (i.e., between cigarettes) levels of nicotine achieved by cigarette smoking, respectively (1551). Absorption of nicotine in the buccal mucosa and the resulting steady-state levels are further decreased by an acidic environment; thus, patients should not drink acidic beverages (e.g., coffee, soda, juice) immediately before, during, or after nicotine gum use (1552).

In terms of dosing, the original recommendation was to use one piece of 2-mg gum every 15–30 minutes as needed for craving. More recent work, however, suggests greater efficacy with scheduled dosing (e.g., 1 piece of 2-mg gum/hour) and 4-mg gum for highly nicotine-dependent smokers (724). Although the original recommended treatment duration was 3 months, many experts believe longer treatment is more effective. However, two trials of longer treatment durations produced contradictory results (801).

+ c. Nicotine lozenges

The nicotine lozenge contains nicotine bound to a polacrilex ion-exchange resin, similar to the nicotine gum. It is available in 2- and 4-mg dose formulations. Because it does not have to be chewed, the lozenge may be preferable for smokers with dental problems or for those who do not like to chew gum. Compared with an equal dose of nicotine gum, the lozenge delivers 25% more nicotine (802). It is recommended that the lozenge be used every 1–2 hours for the first 2–4 weeks of treatment, with the frequency of use reduced to every 2–4 hours in subsequent weeks.

+ d. Nicotine nasal spray

This type of NRT consists of a nicotine solution in a nasal spray bottle similar to those used with saline sprays and antihistamines (71, 809, 1553). Nasal sprays produce droplets that average 1 mg per administration (spray solutions are dispensed at 10 mg/ml); patients administer the spray to each nostril every 1–2 hours. This formulation produces a more rapid rise in and higher nicotine levels than the nicotine gum but less than that obtained with cigarettes. There is approximately a 30% replacement of plasma nicotine levels with nasal spray use (809). Peak nicotine levels occur within 10 minutes, and venous nicotine levels are about 67% of between-cigarette levels (782). Smokers are to use the product ad libitum up to 30 times/day for 12 weeks, including a tapering period.

+ e. Nicotine vapor inhalers

Nicotine vapor inhalers are cartridges of nicotine containing about 1 mg of nicotine each placed inside hollow cigarette-like plastic rods. The cartridges produce a nicotine vapor when warm air is passed through them (805, 806). Absorption from the inhaler is primarily buccal rather than respiratory (1554). Nicotine blood levels produced by the nicotine vapor inhaler are about 33% of between-cigarette levels (1555). Smokers are instructed to puff continuously on the inhaler (0.013 mg/puff) during the day, with a recommended dose of 6–16 cartridges daily. The inhaler is to be used ad libitum for about 12 weeks.

+ (3) Efficacy of nicotine replacement therapies
+ a. General issues

In general, the various NRTs have shown comparable efficacy in studies using the rigorous outcome measure of at least 6 months of abstinence from smoking (790, 793, 1556). Although specific rates of smoking cessation vary across studies, there is typically a twofold increase in rates of smoking abstinence with an NRT. Many studies have shown that NRTs decrease withdrawal symptoms such as anxiety, anger/irritability, depression, difficulty concentrating, and impatience in outpatient settings (602, 755); insomnia and weight gain, however, are not consistently decreased by all forms of NRT (801). For example, nicotine gum at 2 mg ad libitum appears less likely to reduce craving compared with the nicotine patch or nasal spray; however, this may be a dose-related issue, as the 4-mg gum does appear to better reduce craving (1557). Finally, higher doses of an NRT only marginally decrease withdrawal symptoms (801).

+ b. Nicotine transdermal patch

The overall efficacy of the nicotine transdermal patch for smoking cessation has been well documented. A 1994 meta-analysis of 17 randomized, placebo-controlled trials of the patch (71) documented abstinence rates of 27% vs. 13% at the end of treatment and 22% vs. 9% at 6-month follow-up for the nicotine and placebo patch, respectively. These effects of the nicotine transdermal patch were independent of patch type, treatment duration, tapering procedures, and behavioral therapy format or intensity, although it should be noted that behavioral treatment plus nicotine patch enhanced treatment outcomes compared with the patch alone. A recent meta-analysis of patch trials suggests an odds ratio of 1.81 (95% confidence interval [CI] = 1.63–2.02) for nicotine patch versus placebo patch treatment effects (790). Because the nicotine patch is available as an over-the-counter product, it is also noteworthy that randomized, double-blind, placebo-controlled studies (1549, 1558) as well as a meta-analysis (793) have demonstrated the efficacy of nicotine patch therapy in over-the-counter settings.

Several studies have addressed the issue of matching the patch dose to the patient's nicotine-dependence level, and there is modest evidence that higher patch doses are more helpful for more heavy and dependent smokers, at least with short-term outcomes. For example, Hughes et al. (1544) evaluated the effects of four doses of the patch (0, 21, 35, and 42 mg/day) for smoking cessation in 1,039 cigarette smokers using >30 cigarettes/day. In conjunction with weekly smoking cessation group therapy, patches were used daily for 6 weeks, after which the dose was tapered over an additional 10 weeks. At 12 weeks, there was a dose-dependent effect of nicotine patch therapy on smoking cessation rates, but it was less pronounced at 1-year follow-up. In an 8-week trial of 504 cigarette smokers using at least 15 cigarettes per day, Jorenby et al. (1559) compared two doses of the patch (44 and 22 mg/day) with three different intensities of counseling and found superior smoking cessation rates with the higher-dose patch only in combination with the minimal therapy condition and at the 4-week time point. Similar findings were reported by Killen et al. (1560) in their double-blind trial of 408 cigarette smokers who smoked >25 cigarettes per day. The subjects in that trial were randomized to receive either a high- (25-mg) or low- (15-mg) dose 16 hour/day patch for 6 weeks. Paoletti et al. (1561) stratified 297 cigarette smokers to a dose of patch based on low (250 ng/ml) or high (>250 ng/ml) baseline plasma cotinine levels and found comparable cessation rates for the 25- and 15-mg patches in individuals with high baseline plasma cotinine levels. Thus, further treatment studies are needed to determine if higher-dose patches are effective for heavier and more dependent smokers. In addition, evidence from one randomized, double-blind, placebo-controlled study suggests that nicotine patch therapy may improve smoking cessation outcomes even in individuals who had resumed smoking after previous nicotine patch treatment and were smoking at least 15 cigarettes per day (1562).

Other studies have assessed whether combining nicotine transdermal patch and other smoking cessation treatments results in improved outcomes. A randomized, double-blind, placebo-controlled trial by Bohadana et al. (1563) studied the efficacy of the nicotine inhaler in combination with a nicotine or placebo patch for up to 12 weeks in 400 nicotine-dependent subjects. The inhaler plus nicotine patch group had higher abstinence rates than the inhaler plus placebo patch group at 6 and 12 weeks but not at 6- or 12-month follow-up. These results suggest that the use of the nicotine patch can augment outcomes with the nicotine inhaler and these two NRTs can be safely combined. The combination of nicotine patch with behavioral therapy has been studied in several trials. One trial suggested a modest additional benefit of nicotine patch at 10 weeks when combined with CBT (1564), but no difference was found at 6- and 12-month follow-ups. Two other trials (1559, 1565) used other behavioral interventions (e.g., self-help pamphlets, self-help video-enhanced manual, physician motivational messages, brief follow-up nurse visits, weekly cessation counseling groups) and found no augmenting effects of behavioral treatments over the nicotine patch alone.

Some studies have also shown efficacy for nicotine patch treatment in a variety of patient subgroups and specialized contexts. For example, a large double-blind, randomized, placebo-controlled study by Ahluwalia et al. (669) specifically examined the usefulness of the nicotine patch in African Americans, an important subset of the smoking population for which there is an insufficient number of studies assessing the efficacy of standard smoking cessation pharmacotherapies. Greater smoking cessation rates were observed with active patch (21 mg/day) relative to placebo patch at the end of the 10-week trial, with a trend toward significance at 6-month follow-up.

Three randomized, placebo-controlled trials have examined the efficacy of the nicotine patch in outpatient or hospitalized medical patients with smoking-related cardiac and pulmonary diseases. These trials demonstrated short- but not longer-term (i.e., >6 month) improvements in smoking cessation rates (910, 916, 1566). The negative outcomes, including mortality, were no greater with nicotine patch treatment than with placebo patch.

There have also been three preliminary controlled studies of the nicotine patch in patients with a psychiatric or substance use disorder. There have been two open-label studies of the patch (21 mg/day for 6 weeks, followed by 14 and 7 mg/day for 2 weeks each) for patients with co-occurring schizophrenia and nicotine dependence that demonstrated short-term smoking cessation in about 33% of patients (702, 703). In the study by George et al. (703), treatment of subjects with second-generation as compared with first-generation antipsychotics led to enhanced smoking cessation outcomes. However, in both studies there was a significant return to smoking in patients with schizophrenia 6 months after patch discontinuation. Finally, a single small, randomized, controlled trial in alcohol-dependent smokers (705) showed no significant effect of nicotine patch therapy on smoking cessation rates. Nevertheless, the patch was well tolerated by patients in all three studies. It is clear that placebo-controlled trials to establish the relative efficacy of the patch in psychiatric and substance-using populations are needed, but it appears that the patch has utility in treating smokers with a co-occurring psychiatric disorder.

+ c. Nicotine gum

Several placebo-controlled trials have established the safety and efficacy of nicotine gum for smoking cessation (790), with a pooled odds ratio of 1.66 (95% CI = 1.52–1.81). There is some evidence from randomized, controlled trials supporting the use of 4-mg nicotine gum in more highly dependent cigarette smokers (803, 1557, 1567, 1568). This finding supports the idea of matching the nicotine gum dose to the smoker's dependence level. The efficacy of matching gum dose to the patient's smoking dependence level appears to be independent of the intensity of behavioral therapy support, although this needs further study.

To manage the heavy smoker with treatment-resistant dependence, the combination of nicotine gum with nicotine patch is often used clinically, and its efficacy is supported by some evidence. For example, in a randomized trial of 374 nicotine-dependent cigarette smokers, Kornitzer et al. (888) found that the combination of active nicotine gum (2 mg) and active nicotine patch (15 mg/day) was superior to active patch plus placebo gum and placebo gum plus placebo patch at the end of the 12-week trial (34.2% vs. 22.7% vs. 17.3%) as well as at 1-year follow-up (18.1% vs. 12.7% vs. 13.3%).

+ d. Nicotine lozenges

Nicotine lozenges have been less well studied but also appear to be efficacious in improving short- and long-term smoking cessation rates. In a 6-week double-blind, randomized, placebo-controlled, multicenter study of 2- and 4-mg nicotine lozenges compared with placebo gum (802), smokers with low nicotine dependence (first cigarette >30 minutes after waking) were assigned to the 2-mg lozenge or placebo and those with a high nicotine dependence (first cigarette <30 minutes after waking) were assigned to the 4-mg lozenge or placebo. Both doses of the lozenge significantly increased carbon monoxide-verified continuous abstinence rates, with significant reduction in nicotine craving and withdrawal. At the 4-mg dose, abstinence rates were more than doubled (48.7% vs. 20.8%; p<0.001), suggesting that the higher 4-mg lozenge may be more efficacious in more highly dependent smokers and that the lozenge dose can be matched with the dependence level. The efficacy of the lozenge compared with placebo was also demonstrated at 12-month follow-up.

+ e. Nicotine nasal spray

Nicotine nasal spray has been found to be a safe and effective aid for smoking cessation in two randomized, double-blind, placebo-controlled trials (809, 1569). Both trials provided treatment for 3–6 months. Active nasal spray led to a doubling of quit rates during active use, but the differences between the active and placebo treatment were reduced or absent with extended follow-up, suggesting the need for maintenance use of this agent. The results of a recent Cochrane meta-analysis (790) suggested that the odds ratio for nicotine nasal spray versus placebo is 2.35 (95% CI=1.63–3.38). To date, long-term studies of the spray as well as studies of the spray in combination with the patch, gum, or bupropion have not been published.

+ f. Nicotine inhalers

Two randomized, double-blind, placebo-controlled trials (1570, 1571) have demonstrated the superiority of nicotine vapor inhalers to placebo inhalers for smoking cessation in trials of 4–6 months' duration, with two- to threefold increases in quit rates (17%–26%), compared with placebo at the trial endpoints, but smaller between-group differences at follow-up periods 1 year. The results of the Cochrane meta-analysis (790) suggest that the odds ratio for nicotine inhaler versus placebo is 2.14 (95% CI=1.44–3.18). Although these data support the short-term efficacy of the inhaler in cigarette smokers, longer-term trials are needed.

+ (4) Side effects of nicotine replacement therapies
+ a. Nicotine patch

No significant medical problems with nicotine patches have been found (798, 801, 1550, 1551). The most common minor side effects are skin reactions (50%), insomnia and increased or vivid dreams (15% with 24-hour patches), and nausea (5%–10%) (798, 801, 1550). Tolerance to these side effects usually develops within a week. In addition, the rotation of patch sites can decrease skin irritation, and a 24-hour patch can be removed before bedtime or changed to a 16-hour patch to determine if nicotine replacement is contributing to insomnia. Although the results of an early study suggested that the concomitant use of cigarettes and nicotine patches caused myocardial infarction (800), later analyses and prospective empirical studies in smokers with active heart disease indicated that the use of nicotine patches is safe in cardiac patients (836, 916). Abrupt cessation of the nicotine patch does not appear to produce significant withdrawal symptoms, and long-term use of the patch has not been associated with any long-term medical or psychiatric sequelae (140, 602, 798). There appears to be little dependence liability associated with patch use, as only 2% of patch users continue to use this product for an extended period after a cessation trial (796).

+ b. Nicotine gum

Major side effects from nicotine gum are uncommon and rarely deter use (790, 793, 804); minor side effects are of mechanical (e.g., difficulty chewing, sore jaw) or local pharmacological (e.g., burning in mouth, throat irritation) origin. Tolerance develops to most side effects over the first week (804). Education about the proper use of the gum (e.g., do not chew too vigorously) also decreases the side effects (804). In earlier research, some disorders were listed as contraindications to the use of nicotine gum (e.g., cardiovascular disease, pregnancy, hypertension). However, because nicotine blood levels are much lower with nicotine gum than with cigarettes, these contraindications have been removed (804, 1551).

The only potential psychological side effect of nicotine gum is the continuance of nicotine dependence (758). Abrupt cessation of nicotine gum can produce withdrawal symptoms similar to but less intense than that from cigarettes (758), whereas gradual reduction in the use of nicotine gum usually produces very minor or no withdrawal symptoms (758, 1572). There are several lines of evidence indicating that most long-term use is not dependence. Instead, long-term use appears to represent patients' desire to extend the duration of therapy based on their fear that they will return to smoking if they stop using the nicotine gum. West et al. (796) conducted a study on the continued use of NRT 15 weeks after smoking cessation and found that only 7% of smokers continued to use the gum after a cessation attempt compared with 2% continuing the patch and 10% continuing the nasal spray or inhaler. By 2 years, all but 1%–2% of smokers had stopped gum use, and the amount of gum use at long-term follow-up was minimal (usually 12 mg/day) (797). The potential harmful effects of long-term use of nicotine gum have not been studied; however, it is unlikely there are any, given the absence of exposure to carcinogens or carbon monoxide and the much lower levels of nicotine obtained from nicotine gum than from cigarettes (804, 1551).

+ c. Nicotine lozenges

Mild throat and mouth irritation have been reported in preliminary trials (802). Side effects of the nicotine lozenge include heartburn, hiccups, and nausea (802). In addition, because the lozenge contains phenylalanine, it should not be used by individuals with a history of phenylketonuria.

+ d. Nicotine nasal spray

The major short-term side effects of nicotine nasal spray are nasal and throat irritation, rhinitis, sneezing, coughing, and watering eyes (807–809). One or more of these occur in >75% of patients, although long-term nasal problems from use of nicotine nasal spray do not usually occur (807). Whether abrupt cessation of the spray produces withdrawal has not been studied. Nicotine nasal spray may have some potential to induce dependence. Several patients who quit smoking with nicotine nasal spray in some studies continued to use it for long periods (782, 811). Indeed, one controlled study of nicotine nasal spray by West et al. (796) determined that continued use of the spray occurred in 10% of smokers, and the results of a human laboratory study (811) suggest that nicotine nasal spray has modest reinforcing effects. Thus, follow-up of smokers using nasal spray is recommended.

+ e. Nicotine inhaler

No serious medical side effects have been reported with nicotine inhalers (810). About 50% of subjects report throat irritation or coughing (806). The results of the controlled study by West et al. (796) suggest that about 10% of smokers quitting with a nicotine inhaler continue to use it for extended periods, and thus follow-up of inhaler users is recommended.

+ (5) Implementation

Nicotine-dependent smokers benefit equally from nicotine gum, patch, inhaler, or nasal spray (790, 793, 1573, 1574). Some have suggested that NRT be used only if the patient is enrolled in behavioral therapy; however, the data clearly show that NRT is effective in the absence of behavioral therapy (790) and that not all individuals are receptive to concomitant behavioral therapy. Although nicotine gum and patches are available over the counter, psychiatrists may still encourage appropriate use of the patch consistent with package instructions, provide adjunctive psychiatric management and, when appropriate, psychosocial and other pharmacological therapies. It has been suggested that a patient's precessation blood level of cotinine, a metabolite of nicotine with a half-life of about 20 hours, may be a useful benchmark to examine the percent of nicotine replaced by NRTs (1575), presumably because the level from nicotine replacement is higher; however, this area is controversial and requires further study (1576).

Although most smoking cessation treatment lasts 6–12 weeks, some authors (71, 140, 602, 1577) have advocated longer-term use of NRT and even a nicotine replacement program. However, the European multicenter CEASE trial compared high (25-mg) versus low (15-mg) patch doses and shorter (22 weeks) versus longer (8 weeks) periods of treatment and found that although higher doses of the nicotine patch led to better outcomes, extending patch treatment beyond 8–12 weeks did not increase cessation rates (1578). In addition, a meta-analysis of nicotine patches did not find that longer treatment was associated with higher quit rates (71). However, no data examining whether longer NRT treatment is associated with higher abstinence rates are available for nicotine nasal spray, inhaler, or lozenges.

b) Sustained-release bupropion
+ (1) Goals

Bupropion is a first-line pharmacological treatment for nicotine-dependent smokers who want to quit smoking. It is a phenylaminoketone that is an atypical antidepressant and is available in immediate-release, sustained-release (SR), and extended-release (XR) formulations. Bupropion's mechanism of action in the treatment of nicotine dependence likely involves blockade of dopamine and norepinephrine reuptake (161) as well as antagonism of high-affinity nicotinic acetylcholine receptors (162). The goals of bupropion therapy entail cessation of smoking behavior, reduction of nicotine craving and withdrawal symptoms, and prevention of cessation-induced weight gain.

+ (2) Efficacy

Since the pivotal study by Hurt et al. (158) established the efficacy and safety of bupropion SR for treatment of nicotine dependence, which led to its FDA approval in 1998, multiple studies have confirmed its utility (160, 795, 812, 1579, 1580, 1582). A recent meta-analysis showed that rates of smoking cessation are doubled by bupropion treatment (795).

The 7-week double-blind, placebo-controlled multicenter trial of Hurt et al. (158) assigned 615 subjects who smoked 15 cigarettes per day to placebo or to 100, 150, or 300 mg/day bupropion SR; all treatment conditions also included weekly individual smoking cessation counseling. The end-of-trial, 7-week prevalence cessation rates were 19.0%, 28.8%, 38.6%, and 44.2%, respectively. At 1-year follow-up, the cessation rates were 12.4%, 19.6%, 22.9%, and 23.1%, respectively. In addition, bupropion SR treatment reduced weight gain associated with smoking cessation in a dose-dependent manner and significantly reduced nicotine withdrawal symptoms at the 150 and 300 mg/day doses. In this study, the major side effects associated with bupropion SR compared with placebo were insomnia and dry mouth. In accordance with these results, the 300 mg/day dose (150 mg b.i.d.) was recommended as the target dose for bupropion therapy for smoking cessation, with an initial dose of 150 mg/day for 3–4 days.

Several recent studies have extended the use of bupropion for smoking cessation. Hays et al. (160) examined the effects of bupropion versus placebo on the prevention of smoking relapse in 784 cigarette smokers who achieved smoking abstinence after a 7-week open-label trial of bupropion (300 mg/day). Those subjects who were abstinent from smoking at the end of the open-label phase of the trial (58.8%) were then randomized to receive bupropion (300 mg/day) or placebo for an additional 45 weeks. At the end of the 52-week treatment period, significantly more smokers in the bupropion group were abstinent than in the placebo group (55.1% vs. 42.3%; p<0.01) but not at the 1-year postmedication follow-up assessment (41.6% vs. 40.0%). The number of days to smoking relapse was higher in the bupropion than in the placebo group (156 vs. 65; p<0.05), and weight gain was significantly less in the bupropion group at the end of the 52-week treatment and at 1-year follow-up.

Dale et al. (1580) randomly assigned 615 subjects who smoked at least 15 cigarettes/day to placebo or to 100, 150, or 300 mg/day bupropion SR in a double-blind fashion. At the end of the 7 weeks of active treatment, bupropion SR showed improved rates of smoking abstinence in a dose-dependent manner.

It is not clear whether bupropion is superior to NRT or whether the addition of NRT to bupropion improves overall cessation rates. In one study, the combination of bupropion with the nicotine patch was evaluated in a double-blind, placebo-controlled, randomized, multicenter trial (159). The 893 cigarette smokers in that study, who smoked at least 15 cigarettes/day, were randomized to one of four experimental groups: 1) placebo bupropion (0 mg/day) plus placebo patch; 2) bupropion (300 mg/day) plus placebo patch; 3) placebo bupropion plus nicotine patch (21 mg/day for 4 weeks, with 2 weeks of 14 mg/day and 2 weeks of 7 mg/day); or 4) bupropion plus nicotine patch. Bupropion was administered 1 week before the target quit date (day 15), at which time patch treatment was initiated for a total of 8 weeks. All subjects received weekly individual smoking cessation counseling. Cessation rates at the 1-year follow-up assessment were 15.6% for placebo, 16.4% for active patch alone, 30.3% for bupropion alone, and 35.5% for the combination of nicotine patch and bupropion. The bupropion plus nicotine patch and bupropion alone conditions were significantly better than the placebo and nicotine patch alone conditions, but the combination (35.5%) was not significantly better than bupropion alone (30.3%). Weight suppression after cessation was most robust in the combination therapy group. Side effects were consistent with the profiles of patch and bupropion, and the combination was well tolerated. However, a higher-than-expected rate of treatment-emergent hypertension (4%–5%) was noted with the combination of bupropion and nicotine patch (159). The patch alone treatment was significantly different from placebo at the end of the trial but not at the follow-up assessments.

Other studies have suggested the efficacy of bupropion SR for smoking cessation in African Americans (670) and for smokers who have not responded to initial bupropion therapy (1583), smokers with chronic obstructive pulmonary disease (918) or cardiovascular disease (917), and users of smokeless tobacco (1584, 1585).

Finally, studies have suggested the utility of bupropion SR for smoking cessation/reduction in psychiatric and substance-abusing smokers. Using a secondary analysis of the Hurt et al. (158) study, Hayford et al. (1586) found that bupropion SR was equally efficacious for smoking cessation in individuals with or without a history of major depression or alcoholism. Similarly, Cox et al. (454), in a multicenter, randomized, double-blind relapse prevention trial of bupropion (300 mg/day) with 784 smokers, found bupropion to be effective regardless of whether patients had a history of depression or not. Finally, Chengappa et al. (458) found good rates of response to an open-label trial of bupropion for smoking cessation in 25 smokers with remitted major depression who were receiving maintenance SSRI, with 8 of the 25 (32%) subjects stopping smoking at the end of the 9-week trial. These findings suggest that bupropion's actions on nicotine dependence are independent of its antidepressant effects.

Bupropion SR has also been evaluated in three trials of patients with schizophrenia and nicotine dependence, including an open-label trial of 300 mg/day (415) and placebo-controlled trials of 150 (704) and 300 (414) mg/day. Weiner et al. (415) conducted a 26-week open-label trial of bupropion SR (300 mg/day), with 14 weeks of initial group CBT in eight patients with schizophrenia and nicotine dependence. Although none of the patients quit smoking during the trial, they did decrease their cigarette use, as indicated by an overall reduction (about 40%) in expired carbon monoxide levels at the end of the trial compared with pretrial levels. Of the four patients who chose to continue bupropion treatment for an additional 12 weeks after the trial, one stopped smoking completely, two had further decreases in expired carbon monoxide levels, and one maintained the decrease attained during the trial. In additional, negative symptoms of schizophrenia were reduced by bupropion during the trial, but this finding was not significant.

In a 12-week double-blind, placebo-controlled trial, Evins and colleagues (704) added bupropion SR or placebo to concurrent CBT for 18 patients with schizophrenia and nicotine dependence. Compared with placebo, treatment with bupropion led to an approximately 40%–50% reduction in carbon monoxide levels by the end of the trial. In addition, one of the nine subjects in the bupropion group versus none of the nine subjects in the placebo group achieved smoking cessation by the end of the trial. Both the positive and negative symptoms of schizophrenia were found to be reduced by bupropion during the trial.

In the third study, George et al. (414) conducted a 10-week double-blind, placebo-controlled trial of bupropion SR (300 mg/day) in 32 nicotine-dependent smokers with schizophrenia or schizoaffective disorder. All subjects received weekly group therapy emphasizing motivational enhancement, relapse prevention, and social skills training. Trial endpoint cessation rates (confirmed by a carbon monoxide level <10 ppm) were 50% (8 of 16) in the bupropion group and 12.5% (2 of 16) in the placebo group (p<0.05). Positive symptoms of schizophrenia were not affected, but negative symptom scores were reduced by about 15% in the bupropion group. In addition, treatment with a second- versus a first-generation antipsychotic medication strongly predicted success in smoking cessation in patients with schizophrenia.

The results from these preliminary placebo-controlled trials of bupropion suggest that smoking reduction and cessation are possible in patients with schizophrenia, exacerbation of psychotic symptoms is unlikely, and negative symptoms of schizophrenia may be reduced. With endpoint cessation rates of 11%–50%, bupropion may be more effective at higher doses (300 vs. 150 mg/day) in this population.

+ (3) Side effects

The primary side effects reported with bupropion administration in cigarette smokers are headache, nausea and vomiting, insomnia, and activation, most of which occur during the first week of treatment, although insomnia can persist (158, 159, 795, 1587–1590). Seizures are of exceedingly low occurrence (<0.5%) at doses of 300 mg/day, but a history of seizures or a seizure disorder would suggest a need for caution in prescribing bupropion (795, 1590). In addition, the use of bupropion is not recommended in individuals with a past or particularly a current diagnosis of an eating disorder because one study found an increased risk of generalized tonic-clonic (grand mal) seizures in bupropion-treated patients with bulimia (1591). The risk of hypertension with bupropion is <1% in smokers but may be higher in combination with an NRT (159).

+ (4) Implementation of bupropion therapy

For the treatment of nicotine dependence, the target dose of bupropion is 300 mg/day, taken in two doses of 150 mg each. Bupropion is started 7 days before the target quit date at 150 mg/day and, after 4–5 days, the dose is increased to 150 mg b.i.d. Currently there are few data as to which subgroups of smokers may benefit most for treatment with bupropion, although smokers with depressive symptoms may benefit from the medication's antidepressant properties independent of its efficacy in treating nicotine dependence.

c) Nortriptyline

Nortriptyline is a TCA that has been shown to be superior to placebo for smoking cessation in randomized, double-blind, placebo-controlled trials (456, 795, 814, 815, 1592) and to have comparable efficacy to bupropion (795, 814, 1579). Some evidence suggests that its benefits persist for up to a year of treatment (895) and that its efficacy can be augmented by concomitant use of the nicotine patch (455); however, data on adding psychosocial therapies to nortriptyline treatment are mixed (456, 895, 1579).

Nortriptyline's mechanism of action in treating nicotine dependence is unclear, although the medication does block reuptake of norepinephrine and serotonin and may have indirect effects on dopamine (816). In addition, its mechanism in nicotine dependence may be distinct from its mechanism in treating depression, because its efficacy in smoking cessation is unrelated to the presence or absence of depressive symptoms or major depressive disorder (456, 795, 815, 816).

Side effects of nortriptyline are frequent (455, 815, 816) and include anticholinergic effects (e.g., dry mouth, blurred vision, constipation, tachycardia, and urinary hesitancy or retention), sedation, orthostatic hypotension, and cardiac conduction changes (817). The toxicity of nortriptyline in overdose amounts also needs to be taken into consideration when prescribing this medication. Nevertheless, nortriptyline appears to have some utility in smokers and can be recommended as a second-line agent after NRT and bupropion, although further study of this agent is necessary.

d) Clonidine

Clonidine is a presynaptic 2-adrenergic agonist that dampens sympathetic activity originating at the locus coeruleus (602, 749, 1593). Because of its suggested efficacy for alcohol and opioid withdrawal, it was tried with nicotine withdrawal as well (749, 1593). Several clinical trials have used oral or transdermal clonidine in doses of 0.1–0.4 mg/day for 2–6 weeks with and without behavioral therapy; a meta-analysis of these trials suggested that clonidine is efficacious (818). In general, however, the effects of clonidine have not been as robust as the effects of NRTs or bupropion. The most common side effects of clonidine treatment are dry mouth, sedation, and constipation (818, 819, 1594). Postural hypotension, rebound hypertension, and depression are rare when clonidine is used for smoking cessation treatment (1594).

e) Other pharmacotherapies

A number of other pharmacotherapies for nicotine dependence have been studied, but there is insufficient evidence to support their use. Mecamylamine, an oral antihypertensive agent, is a noncompetitive ion channel blocker of high-affinity CNS and peripheral nicotinic receptors (820, 821, 1595) that decreases the positive subjective effects from cigarettes. In several small trials (820, 821), it was shown to result in an initial increase but a subsequent decrease in cigarette use. It has been postulated that naltrexone, the long-acting oral form of the short-acting intravenous opioid antagonist naloxone, would be useful in treating nicotine dependence because the performance-enhancing and other positive effects of nicotine may be opioid mediated (1596). Although one preliminary study showed benefits of naltrexone in combination with nicotine patch therapy (1597), naltrexone did not appear to decrease smoking in other studies (603, 822) and may even have increased smoking in some individuals (749, 1598). Buspirone, a 5-HT1A receptor partial agonist anxiolytic agent, has been noted to improve short-term smoking cessation rates in a general population of smokers; it improved abstinence in high-anxiety smokers in one study (1599) but not in another (823, 1600). The alkaloid lobeline, which appears to have dopamine reuptake blockade and nicotinic receptor antagonist properties (1601), was evaluated in a 6-week, randomized, double-blind, placebo-controlled, multicenter trial but did not produce any differences in smoking cessation rates (1602). Thus, there is little evidence that these diverse pharmacotherapies are useful for smoking cessation.

In addition to bupropion and nortriptyline, other antidepressive agents have been studied in nicotine-dependent patients. The available evidence gives little support for the use of SSRIs to assist in smoking cessation, either alone (1603, 1604) or in combination with NRT (1605, 1606). Although possible benefits have been found for the monoamine oxidase A inhibitor moclobemide (1607) and the monoamine oxidase B inhibitor selegiline hydrochloride (1608) for smoking cessation, larger trials of these agents are warranted. At the present time there is no evidence that other antidepressants are efficacious in treating nicotine dependence (795).

f) Somatic treatments

Acupuncture is often proposed for use in nicotine dependence, with the rationale that it may release endorphins that assist in smoking cessation. The efficacy of auricular acupuncture has been supported by some (786) but not other (1609) studies. The results of two more recent controlled studies (785, 1610) suggest that active versus sham acupuncture can lead to short- and long-term reduction in cigarette smoking but not smoking cessation. In addition, a meta-analysis of multiple other smaller studies of acupuncture has found no evidence for acupuncture efficacy in smoking cessation (788). In accordance with these results, acupuncture cannot be recommended as a treatment for smoking cessation.

Other somatic treatments that have been proposed include black pepper extracts (1611), capsaicin (1612), denicotinized tobacco (1613–1615), cigarette flavorings (1616), regenerated (denicotinized) smoke (1617), and citric acid inhaler (1618, 1619), all of which decrease cigarette craving or withdrawal symptoms or provide a substitute for the satisfaction from cigarettes in laboratory tests. However, none of these modalities have been sufficiently studied to recommend their use.

Cigarette filters have also been used to help smokers gradually reduce the amount of nicotine inhalation; however, studies of the efficacy of filters are inconclusive (1620, 1621), and smokers may to some extent adjust the frequency or depth of inhalation to compensate for such changes (1620, 1621). Thus, there is insufficient evidence to recommend filters as a treatment at this time.

2. Psychosocial treatments
a) Goals

The goals of psychosocial therapies for nicotine dependence include 1) providing individuals with the necessary skills to initiate abstinence, 2) teaching skills to avoid relapse to tobacco use in high-risk situations, 3) supporting and extending the effects of proven pharmacotherapies for nicotine dependence, and 4) facilitating a transition toward eventual smoking abstinence in smokers who are unable to quit immediately.

b) Efficacy

Psychosocial therapies appear to be effective treatments for nicotine dependence either alone or in combination with pharmacotherapies. The descriptive (827, 1580, 1622) and meta-analytic (720, 824–826, 828) reviews of psychosocial interventions for smoking cessation provide more information on efficacy.

c) Side effects

The side effects of psychosocial therapies for smoking cessation are considered minimal. Some factors should be taken into account when more intensive psychosocial therapies are considered, including their length and intensity. For these reasons, the majority (about 95%) of smokers do not desire or tolerate psychosocial therapies beyond minimal interventions.

d) Implementation

Although pharmacotherapy (bupropion or NRT) is recommended for all patients, a stepped-care approach in implementing psychosocial treatments for smoking cessation in general medical populations is recommended; thus, in initial smoking cessation attempts, social support and brief interventions (e.g., brief counseling, telephone support) should be tried first (826). If they are ineffective, more intensive behavioral treatments can be considered because there is a strong association between the intensity (826) and duration (895) of tobacco dependence counseling and its effectiveness. This strategy of a stepped-care approach to psychosocial treatment is supported by the fact that most smokers will continue to seek treatment despite repeated unsuccessful attempts at abstaining from smoking (1623, 1624).

Both group and individual formats of psychosocial therapies have been used for smoking cessation (720, 721, 742); groups can increase social support and individual treatment can address the particular problems of the smoker. In terms of the format of interventions, it is not clear whether individual or group psychosocial treatments are superior to one another; this may depend on the subtype of smoker and/or the treatment setting. Although studies of matching therapies to particular types of smokers have been published (1625, 1626), at present there are insufficient data to recommend specific matching strategies. Thus, patient preference should be considered when recommending group or individual treatment or when choosing a specific psychosocial treatment approach.

Because nearly 67% of patients relapse in the first week after a smoking cessation attempt (698, 794), most treatment is timed to occur over the few weeks immediately before and after the quit date (742). The most common providers of psychosocial therapies for smoking cessation are self-help groups (e.g., Nicotine Anonymous), state health departments, community organizations (e.g., American Cancer Society, American Lung Association), wellness programs, clinics for the treatment of substance use disorders, or health educators and psychologists in health care organizations (172). However, some treatments may not be available or be available only intermittently to patients, be costly, and not be integrated into the health care system. Consequently, many of those motivated to quit smoking forgo psychosocial therapy (172).

e) Specific techniques

The specific psychosocial interventions that have been evaluated for the treatment of nicotine dependence are described below.

+ (1) Social support

A number of studies have demonstrated the effectiveness of social supports (e.g., buddy systems, partner support) as treatments for smoking cessation (718, 826, 830–833). In some research, social support has been provided as an active intervention, whereas in other studies the supportiveness of a spouse or partner has been assessed independently without a specific intervention occurring (829). Interventions to enhance the degree of support provided by a spouse or partner have not produced significant differences in support levels and, not surprisingly, have not influenced smoking cessation rates (1627). Nevertheless, the effect of social support as an active intervention is significant, and it is thus recommended as a treatment for smoking cessation.

+ (2) Brief therapies

Several studies in controlled settings have documented that smoking cessation rates can be augmented by minimal behavioral interventions such as community support groups (845) and telephone counseling (846, 847). MET, which is often a component of brief interventions, encourages the smoker to weigh the reasons for and against quitting, thereby facilitating a choice of whether to quit or not. Several studies have suggested that brief interventions may be effective in promoting confidence in smoking cessation and treatment retention, even when cessation rates do not seem to be significantly improved (639, 834, 1628). In addition, training programs for physicians in basic behavioral counseling to support cessation may lead to enhanced cessation rates simply because physicians are more likely to engage patients in a discussion about smoking cessation (835). Such brief interventions also seem to have efficacy in promoting smoking cessation in hospitalized medical patients (836, 837), although this was not true across all studies (855). Thus, there is good evidence to recommend brief interventions in smokers making initial attempts at cessation and without prior unsuccessful treatment (826, 828).

+ (3) Behavioral therapies

Behavioral therapy is based on the theory that learning processes operate in the development, maintenance, and cessation of smoking. There have been >100 controlled prospective studies verifying the efficacy of behavioral therapy (720, 734, 735, 838). Behavioral therapies are typically a multimodal package of several specific treatments. In most reviews and meta-analyses, 6-month quit rates with behavioral therapy packages have been 20%–25%, and groups treated with behavioral therapy typically have had a twofold increase in quit rates as compared with control groups (718–720, 734, 735, 824, 825, 838, 1620). These cessation rates with behavioral therapy are similar to long-term outcomes obtained with medications such as NRTs and bupropion. Given this large database of efficacy, multimodal behavioral therapy is a recommended first-line treatment.

Several specific types of behavioral therapies have also been studied, but none are recommended at the present time. With contingency management, which has some evidence for short-term efficacy, behaviors consistent with smoking cessation are reinforced by giving a reward (e.g., money, vouchers) when patients are abstinent and denying them the same rewards if they are not (1629). However, once the contingencies are removed, smoking tends to return quickly. Thus, contingency management approaches might be best indicated in settings where a finite period of smoking abstinence is needed (e.g., during pregnancy, before surgical procedures). In addition, these techniques are probably best combined with CBT strategies.

Cue exposure involves repeatedly exposing patients to real or imagined situations that evoke potent urges to smoke to extinguish the ability of these situations to evoke these urges (1630–1633). Controlled studies (718), however, do not support the efficacy of cue exposure.

In aversion therapy, patients are asked to engage in "rapid smoking," in which inhaling cigarette smoke every few seconds produces a state of mild nicotine intoxication accompanied by nausea, dizziness, tremors, and other symptoms that will negatively reinforce smoking behavior (693, 734, 735). This technique has shown efficacy in many controlled studies, and most reviews and meta-analyses have concluded that rapid smoking is efficacious; however, the available studies have had methodological problems (1634) and adherence to the technique is low (718, 719, 734, 735, 741, 742, 838).

+ (4) Cognitive-behavioral therapies

For smokers who have failed initial treatments, the use of more specialized and higher-intensity behavioral treatments such as CBT in combination with NRT or bupropion is recommended. In CBT, patients anticipate situations where they are likely to smoke (e.g., at a party, during an argument) and plan strategies to cope with these situations without smoking. Behavioral coping includes removing oneself from the situation, substituting other behaviors (e.g., walking, exercising), or using skills (e.g., assertiveness, refusal skills, time management) to manage triggers. Cognitive coping includes identifying maladaptive thoughts, challenging them, and substituting more effective thought patterns (e.g., reminding oneself of why it is important to stop smoking or that the urge will pass) or trying to prevent a slip from becoming a relapse (e.g., not viewing the slip as a catastrophe). Some degree of efficacy of CBT has been demonstrated in smokers with a history of depression, alcohol dependence, and substance abuse (456, 459, 841–844). CBT has also demonstrated some degree of efficacy in addressing weight concerns associated with smoking cessation (840). However, differences in study design and control groups make comparisons of the studies difficult.

+ (5) Self-guided therapies

The major goals of self-help materials and procedures are to increase a smoker's motivation to quit and to teach smoking cessation skills. Most self-help materials are behaviorally oriented, and written manuals are the most common form of self-help material, although computer and video versions have also been developed (856, 859). In controlled settings, computer-generated tailored self-help materials (848, 849, 852, 853) can augment smoking cessation rates in those who adhere to the self-help program. Whether self-help interventions used without additional contact or support increase smoking cessation is debatable (720, 734, 735, 742, 838, 851, 859, 1451). Self-help materials appear to be more effective in patients who are less nicotine dependent (860, 1635) and more motivated (859). The use of multiple modes of therapy (e.g., written materials plus phone contact) appears to enhance the effectiveness of self-help (721, 856, 859). Tailoring materials to the specific needs and concerns of each patient also appears helpful (736, 851, 860). Telephone counseling in response to smokers' calls to a hot line appears to increase cessation when added to other self-help interventions (856–858), and meta-analyses suggest a small positive effect for such combined support (718, 720).

+ (6) Other therapies

A number of other therapies have been studied in small samples or used in clinical settings. One well-controlled randomized study provided evidence that exercise may assist with the prevention of smoking relapse and weight gain related to cessation (1636); however, another did not (863). In fact, a recent meta-analysis found that only 1 out of 11 studies examining the effects of exercise for smoking cessation had positive results; most of these studies were of small sample size and insufficient design (864). Nonetheless, it is recommended that exercise and increased activity be encouraged in smokers attempting to quit or who have recently quit smoking. Biofeedback, 12-step programs, family therapy, and psychodynamic therapies for the treatment of nicotine dependence have been minimally studied, and scientific evidence for their use is limited. Nevertheless, there is clinical consensus that such therapies may be useful in some patients.

Stimulus control is probably best used in the context of multicomponent therapies. It involves initially removing or avoiding cues associated with smoking to reduce urges to smoke. These strategies include discarding cigarettes; removing ashtrays, lighters, and matches; avoiding smokers; and avoiding situations associated with smoking. There is some support for the effectiveness of these techniques alone (718), but they require further study.

Physiological feedback, which presumes that abstinence will be reinforced by giving smokers immediate positive feedback about the decline in carbon monoxide levels when they stop smoking (718), has also been assessed. Although the rationale behind physiological feedback seems logical, actual findings are weak.

Gradual cessation procedures require smokers to gradually reduce the nicotine yield of their cigarettes by 1) increasing the time between cigarettes, 2) switching to cigarette brands with a lower nicotine content, or 3) using graduated filters ("faders") to progressively reduce the delivery of nicotine from the same brand of cigarettes (743, 1637). However, evidence for the efficacy of these treatments in improving quit rates is mixed, and meta-analyses have not supported their efficacy (718, 719).

Relaxation techniques are often taught to smokers to help them manage relapse situations that are associated with anxiety. Although often used in multicomponent programs, relaxation itself has not been shown to increase smoking cessation in most studies (734, 735, 741, 742, 1620) or in a meta-analysis (718). There is also little scientific validation for the effectiveness of hypnotherapies for smoking cessation, despite the fact that they are commonly used by individuals who wish to stop smoking.


B. Alcohol-Related Disorders

1. Treatment settings

Studies of specific settings for the treatment of alcohol-related disorders have been relatively limited. Such studies may also be subject to selection bias in terms of the characteristics of patients willing to be randomized to different treatment settings (1638).

Most (52, 1639–1641) but not all (1303) studies that have randomly assigned patients to different levels of treatment have not found an advantage for inpatient care over less restrictive settings. However, for safety reasons, these studies have been required to exclude from randomization individuals who would ordinarily be considered to require inpatient treatment (1642). For example, the study by Hayashida et al. (48) randomly assigned alcohol-dependent patients coming to a VA hospital to inpatient or outpatient detoxification but excluded those who required hospitalization for medical reasons. However, those excluded were only 10% of the population applying to this clinic. Another study among male veterans (1643) showed that the mortality rates for individuals with an alcohol use disorder 3 years after discharge varied with the initial treatment setting. Veterans who completed inpatient rehabilitation had the lowest mortality rate, whereas those in the following groups had an increasingly high mortality rate, respectively: 1) those who had at least 6 days of inpatient treatment (but did not complete the program), 2) those who were admitted for brief detoxification lasting <5 days, and 3) those who received no specific treatment for their alcohol use disorder. This study provided preliminary evidence that more intensive treatment may lower the mortality associated with a chronic alcohol use disorder. However, patients in this study were not randomly assigned to treatment conditions, so it is possible that self-selection influenced the results.

Some evidence suggests that longer treatment stays and treatment completion may be associated with better outcomes (959, 1304). This probably reflects the fact that more motivated patients are more likely to stay in treatment and have better outcomes, because differences in outcome are not typically observed when patients are randomly assigned to shorter versus longer treatments (51, 956).

Other evidence suggests that the association between treatment setting and outcome may be a complex one that is influenced by the characteristics and treatment needs of the individual patient. Magura et al. (965) studied a cohort of 248 patients who were newly admitted to inpatient rehabilitation or intensive or regular outpatient care and determined whether they were naturalistically matched or mismatched to care according to ASAM patient placement criteria. At 3 months after intake, individuals who received regular outpatient care when intensive outpatient care would have been recommended as more appropriate had poorer drinking outcomes. In individuals who received residential as compared with intensive outpatient treatment, there also was a trend for a better outcome. Rychtarik et al. (966) also examined individual factors that might determine the appropriateness of a given treatment setting for an individual patient. They found that individuals with a high level of involvement with alcohol and lower cognitive abilities had better outcomes when treated in inpatient settings, whereas those with lower levels of alcohol involvement did better in outpatient settings.

2. Pharmacological treatments for withdrawal

Benzodiazepines are effective in treating alcohol withdrawal symptoms, particularly when compared with placebo in the prevention of withdrawal seizures (991, 992, 995). In preventing seizures, benzodiazepines appear comparable in efficacy to anticonvulsants, although the significant heterogeneity of relevant clinical trials makes comparisons difficult (995). Nevertheless, meta-analyses have consistently demonstrated the benefits of benzodiazepines in treating alcohol withdrawal. One meta-analysis (992) examined 134 studies, including 65 prospective controlled trials involving 42 different medications. This analysis concluded that benzodiazepines reduce withdrawal severity and the incidence of seizures and delirium (992). Another meta-analysis of 11 randomized, controlled trials found benzodiazepines to be superior to placebo, with no study demonstrating that any other class of agent, including beta-blockers, carbamazepine, and clonidine, was more beneficial than benzodiazepines (991). Most recently, Ntais et al. (995) analyzed 57 trials that included a total of 4,051 people and concluded that benzodiazepines are effective against alcohol withdrawal symptoms when compared with placebo.

Evidence from multiple randomized, controlled trials also supports the use of symptom-triggered therapy, with symptom-triggered detoxification protocols leading to less use of medication as well as shorter duration of treatment than fixed-dose protocols (998, 1001–1003). An additional clinical trial of outpatient detoxification using chlordiazepoxide prescribed according to a symptom-triggered detoxification protocol showed that, on average, patients received 167 mg of the medication over 2.7 days and 85% completed the protocol (1644); however, this study did not use a randomized, placebo-controlled design.

One randomized, double-blind, placebo-controlled study compared lorazepam with placebo for the prevention of recurrent seizures related to alcohol withdrawal (1005). This study found that treatment with intravenous lorazepam was associated with a significant reduction in the risk of recurrent seizures related to alcohol.

One meta-analysis (992) considered four randomized, controlled trials of carbamazepine and concluded that carbamazepine was superior to placebo and equal in efficacy to phenobarbital and oxazepam for patients with mild to moderate withdrawal. Another study compared the efficacy of carbamazepine and lorazepam for detoxification in patients with more than one previous detoxification for alcohol withdrawal compared with those with one or no previous alcohol withdrawals (977, 1019). The authors found that healthy middle-age male outpatients randomized to carbamazepine or lorazepam had comparable outcomes in terms of symptoms of alcohol withdrawal. However, carbamazepine was superior to lorazepam in preventing rebound withdrawal symptoms and reducing posttreatment drinking, especially among those with multiple past detoxifications (977).

With respect to the use of other anticonvulsants for detoxification, one recent review concluded that there is no current evidence that divalproex sodium is effective in treating alcohol withdrawal (987), but one study demonstrated that divalproex sodium used as adjunctive pharmacotherapy with oxazepam reduced the total amount of oxazepam needed for detoxification (994). There have been two other open-label trials of divalproex to treat alcohol withdrawal: one compared divalproex with lorazepam in 11 patients and concluded that divalproex ameliorated alcohol withdrawal symptoms and decreased the amount of lorazepam needed for detoxification (1020), and another found that valproate was as effective as phenobarbital in managing acute withdrawal (1018).

Studies of adrenergic agents have been quite small, limiting the generalizability of their conclusions. For example, one randomized, controlled trial of 37 male patients admitted for uncomplicated detoxification compared diazepam 10 mg p.o. with propranolol 20 mg p.o., repeating the medication dose every 4 hours for continued withdrawal symptoms (1645). The patients randomized to diazepam or propranolol showed improvements in blood pressure level, pulse rate, and withdrawal tremors. However, one subject in the propranolol group had a withdrawal seizure, whereas no subject randomized to diazepam manifested a withdrawal seizure or hallucinations. Another study compared intravenous diazepam, clonidine, and alprazolam treatment of alcohol withdrawal and found diazepam was more effective than clonidine and placebo for some measures of withdrawal (1646).

3. Pharmacological treatments for dependence and abuse
a) Naltrexone
+ (1) Goals

The goals of pharmacotherapy for alcohol dependence can be multifaceted. Medications may help maintain or improve abstinence, reduce heavy alcohol consumption, decrease relapse rates to heavy drinking, or more generically reduce craving, an intermediary step to the above clinical goals.

+ (2) Efficacy

It is thought that alcohol exerts its rewarding, euphoric, and subsequent craving effects, at least in part, by stimulating opiate receptors. Naltrexone, an opiate receptor antagonist, blocks this effect. Alternatively, opiate receptor antagonism may play a role in diminishing the rewarding aspects of alcohol-induced dopamine release. Naltrexone is one of the most widely studied medications for the treatment of alcohol dependence, with randomized, controlled clinical trials having been performed not only in the United States but also in many European countries and Australia. Most published studies have evaluated naltrexone against placebo for its ability to promote abstinence, reduce heavy drinking days, and decrease rates of relapse. Enough data from randomized, controlled trials have been published to allow for several meta-analyses (e.g., references 152, 153). These meta-analyses concluded that naltrexone was more efficacious than placebo across most of the drinking outcome variables referred to above.

This advantage of naltrexone is small to moderate, with effect sizes of 0.1–0.5 and/or relative risk decreases of 10%–14% depending on the drinking measure. It also appears from research studies and clinical experience that some patients respond to naltrexone and others do not. It has been suggested that a family history of alcoholism, genetic differences in opiate alleles (1647), and high levels of craving may be predictive of naltrexone response. However, when evaluating published studies on naltrexone, there are two significant issues to consider: patient adherence to treatment and the concomitant psychosocial intervention used.

The role of medication adherence was emphasized by Pettinati et al. (1648) in their reanalysis of data from several clinical trials completed at their site. In that review, they observed a marked difference in naltrexone response between patients who adhered to the medication regimen and those who did not. This finding has led others to evaluate the factors that predict adherence to naltrexone treatment among alcohol-dependent subjects (1513). These authors and others (1649) have suggested that high levels of side effects during the first several weeks of treatment reduces adherence and that higher craving levels at study entry, as well as a greater belief in naltrexone's efficacy, may predict better medication adherence. Day-to-day fluctuations in patient motivation to take naltrexone may be overcome through the use of a long-acting injectable form of the medication. This formulation has been designed to release a small amount of naltrexone in the first few days (1650) and has been shown to have good tolerability in a multisite study (24 sites with 624 subjects) without dose titration or initial oral naltrexone (1044).

Several studies (187, 1039) have indicated that naltrexone works best when combined with a relapse prevention (coping skills, CBT) approach. This observation is supported by other studies (1038, 1041). In general, when naltrexone was given to more severely alcohol-dependent subjects without a concomitant defined relapse prevention intervention, its efficacy was less robust (1045) or nonexistent (1043).

The FDA approved naltrexone for use in treating alcohol dependence in 1994 based on several small single-site studies (187, 1037). Although the results from a number of single-site and small multisite studies from several countries have supported the utility of naltrexone (954, 1038–1041) in intent-to-treat analyses, those from a few larger multisite studies have been negative (1042, 1043). In one multisite randomized, controlled trial (1043), 627 predominantly male alcohol-dependent veterans were treated for up to 52 weeks with 50 mg naltrexone or placebo. Participants also received TSF and other treatment "as usual." The results showed that compared with placebo, naltrexone had no effect on abstinence rates, number of days to relapse, percentage of drinking days, or number of drinks per drinking day. In this trial there were several differences compared with other naltrexone studies that had positive results; the most significant of these were 1) the subjects were older and more severely affected by alcohol, 2) the background psychosocial intervention (TSF) may have been more varied, 3) the dropout rate was slightly higher, and 4) there was more variability inherent in this multisite study when compared with single-site studies.

In a 12-week German multisite study (1042), 171 "detoxified" alcoholic subjects from seven centers were treated with 50 mg/day naltrexone and at least 1 hour a week of psychosocial intervention "as usual." There was no statistical difference between naltrexone and placebo on time to first heavy drinking episode, time to first drink, or overall alcohol consumption or craving. However, the naltrexone group had a statistically significant reduction in -glutamyl transpeptidase (a liver enzyme measured in blood that serves as a marker of heavy alcohol consumption) compared with placebo. These study results need to be interpreted cautiously, however, because 1) the average participant was abstinent for almost 20 days before starting naltrexone and the relapse rate overall was lower than that in most other studies, 2) the psychosocial treatment was not standardized, and 3) the dropout rate was relatively high (33%–38%).

+ (3) Side effects

Naltrexone's side effects have generally been moderate at the 50-mg dose used in most studies, and have included gastrointestinal problems (nausea, vomiting, abdominal pain) and CNS-related symptoms (headache, fatigue). The U.S. product label (as described in the Physicians' Desk Reference) warns that significant hepatoxicity may occur with naltrexone. This complication has been observed mostly in morbidly obese subjects receiving higher doses of naltrexone, whereas many clinical trials, including a safety study conducted for FDA review (1046), did not observe significant hepatoxicity. However, a more recent report (1047) describing an interaction between NSAIDs and high-dose naltrexone (>100 mg/day) that leads to hepatoxicity should be noted. Although the exact risk of this interaction is unknown, clinicians should use high doses of naltrexone cautiously and warn their patients of this potentially dangerous medication interaction.

+ (4) Implementation issues

To minimize any potential interaction of side effects caused by naltrexone and those normally occurring during alcohol withdrawal, a few days of alcohol abstinence should be achieved before naltrexone is initiated. In general, naltrexone is begun at 25 mg/day for a few days, followed by an increase in the dose to 50 mg/day, as tolerated. A gradual dose increase might reduce the gastrointestinal side effects that occasionally emerge early in treatment and generally moderate over time. Women might be more sensitive to the gastrointestinal effects of naltrexone and could be especially aided by a slower titration. At this time, there is no strong evidence that naltrexone doses of >50 mg/day are more efficacious, but some individual cases of greater efficacy at higher doses have been noted. Clearly explaining to patients the potential mild adverse effects of naltrexone and the tendency of these effects to improve over time and providing motivational enhancement to promote patient adherence to medication and alcohol reduction/abstinence may all lead to improved patient cooperation with medication and psychosocial treatments. Although it is not clear how specific psychosocial interventions interact with naltrexone, there is good evidence that relapse prevention therapies (e.g., CBT, coping skills) may allow naltrexone to have its maximal effect.

Because naltrexone is an opiate antagonist, individuals abusing opiates may experience opiate withdrawal when treatment is initiated, and those taking opiates for analgesic effects will find them ineffective during naltrexone treatment. Therefore, a complete medication history and urine toxicology screen for opiate medication may be indicated before naltrexone therapy is initiated. In addition, caution should be taken if acute opiate analgesia is required during the course of treatment. For example, higher doses of opiates may be required, in which case the signs of respiratory distress should be monitored and treated appropriately. The naltrexone-treated patient should carry a card explaining these issues and provide it to health care personnel in an emergency.

b) Disulfiram
+ (1) Goals

The goal of treatment with disulfiram is to enhance the motivation of abstinent alcoholic individuals to resist alcohol consumption. Pretreatment with disulfiram establishes the conditions in which the subsequent use of alcohol will result in a toxic and highly aversive reaction. In theory, this knowledge should improve the patient's cognitive control over any urge to drink.

+ (2) Efficacy

Disulfiram inhibits the activity of aldehyde dehydrogenase, the enzyme that metabolizes acetaldehyde, a major metabolite of alcohol. The usual therapeutic dose is 250 mg/day, although some patients achieve optimal benefit at either a higher or a lower dose (range 125–500 mg/day). In the presence of disulfiram, alcohol consumption results in the accumulation of toxic levels of acetaldehyde, which in turn produces a host of unpleasant signs and symptoms including a sensation of heat in the face and neck, headache, flushing, nausea, vomiting, hypotension, and anxiety. Chest pain, seizures, liver dysfunction, respiratory depression, cardiac arrhythmias, myocardial infarction, and death have also been reported.

Controlled trials have not demonstrated any advantage of disulfiram over placebo in achieving total abstinence, delaying relapse, or improving employment status or social stability (1048, 1049). However, some clinicians believe that this medication, when combined with other therapeutic interventions, has benefit for some individuals who remain employed and socially stable (150, 1048, 1050–1052). Treatment effectiveness is enhanced when adherence is encouraged through frequent behavioral monitoring (e.g., breath tests), group support for remaining abstinent (e.g., group therapy, AA) (1053), contingency contracting, or, where feasible, supervised administration of disulfiram. Patients who are intelligent, motivated, and not impulsive and whose drinking is often triggered by unanticipated internal or external cues that increase alcohol craving are the best candidates for disulfiram treatment. Poor candidates might include patients who are impulsive, have poor judgment, or have a co-occurring psychiatric disorder (e.g., schizophrenia) that is severe enough to make the patient unreliable or self-destructive (149, 1060). However, some patients with schizophrenia might be able to use disulfiram correctly while under active maintenance with antipsychotics (363).

+ (3) Side effects

Disulfiram can cause a variety of adverse effects; hepatotoxicity and neuropathies with use of this medication are rare but potentially severe. In patients with moderate to severe hepatic dysfunction, peripheral neuropathies, pregnancy, renal failure, or cardiac disease, disulfiram should be used cautiously and potential benefits and risks for the individual patient should be considered (1048). Because one of the metabolites of disulfiram inhibits dopamine beta-hydroxylase, resulting in increased levels of dopamine, there is a theoretical risk of augmenting psychotic symptoms (409, 410). Thus, patients with a psychotic disorder should be observed for evidence of a worsening psychosis if treated with disulfiram.

+ (4) Implementation issues

Understanding and explaining disulfiram's toxic or lethal effects to patients are a prerequisite for its use (1056–1058), so it should never be used without the patient's knowledge and consent. Patients taking disulfiram must be advised to avoid all forms of ethanol (including, for example, that found in cough syrup). In addition, disulfiram interferes with the metabolism of many medications, including TCAs, so that care must be taken to avoid toxicity (1651).

c) Acamprosate
+ (1) Goals

The goal of using a medication like acamprosate is to decrease the relapse rate to any drinking after an initial abstinence period has been achieved. It may work best in combination with a psychosocial intervention that promotes and facilitates abstinence.

+ (2) Efficacy

In 2004, the FDA approved acamprosate for the treatment of alcohol dependence, primarily based on data derived from European studies (reviewed in references 1062, 1063). Although the neuropharmacological action of acamprosate is not completely known, it is an amino acid derivative of taurine that is thought to work at brain glutamate receptor sites. In general, it is thought to stabilize glutamatergic function (155). As such, it has been hypothesized that it normalizes an aberrant glutamate system present during early abstinence that might be the basis of protracted withdrawal and craving during that period (1064).

European studies evaluating patients who generally started on acamprosate while in a hospitalized setting and were abstinent for at least 7–10 days have shown that with acamprosate treatment, there is an increase in the number of patients who maintain abstinence. Those who relapse have more abstinent time before their first drinking day and also more overall abstinent days during a year or more of treatment (1062, 1063). In contrast, a multisite trial completed in the United States did not find acamprosate to be effective in a primary intent-to-treat analysis, but it did find that when subjects' motivation to maintain abstinence and adhere to medication was taken into account, acamprosate was more effective than placebo in increasing the number of abstinent days (1067). The U.S. trial included outpatients who had a varied number of abstinent days before medication initiation, but, in general, overall pretreatment abstinent time was much shorter than that in the European trials. Also, in the U.S. trial, subjects received standardized medical management counseling, whereas in the European studies (1062, 1063) traditional psychosocial alcohol treatment approaches focusing on the maintenance of abstinence were generally used. It would appear that acamprosate is most effective in patients who have achieved a number of days (perhaps 7) of abstinence before starting the medication, although this theory has not been specifically studied.

+ (3) Side effects

Acamprosate at the FDA-approved dosage of two 333-mg pills three times each day (total dose 1998 mg) is well tolerated, with generally self-limiting and symptomatically treated diarrhea being the main adverse effect. Liver disease should not affect its metabolism or blood levels. Because acamprosate has minimal if any negative interaction with alcohol, it is expected to be generally safe in active or relapsed drinkers.

+ (4) Implementation issues

It is not clear if certain subgroups of alcohol-dependent individuals will benefit from acamprosate more than others. Data from European studies and a U.S. multisite trial suggest that some days of abstinence prior to initiation of acamprosate treatment might improve efficacy. In addition, a patient's strong motivation to maintain or achieve abstinence might also improve treatment response. Acamprosate can be started at a full dose immediately. It is generally taken three times per day to maintain blood levels and avoid unnecessary gastrointestinal problems. It should be well tolerated in individuals with compromised liver function because it is metabolized and excreted primarily through the kidneys. Individuals with renal impairment should be monitored carefully when taking the medication; depending on the severity of the impairment, acamprosate might need to be avoided.

Data suggest it is safe to take acamprosate with naltrexone or disulfiram, and there are no known significant medication interactions.

d) Medications acting on the serotonin system
+ (1) Goals

The use of SSRIs in the treatment of alcoholism may have two distinct goals: 1) reduce drinking or promote abstinence through these medications' effect on the serotonin system, which is hypothesized to directly effect alcohol consumption (see discussion below); and 2) treat psychiatric syndromes/symptoms in alcoholic patients through the well-documented antidepressant and antianxiety effects of these agents. To the extent that anxiety and depression underlie drinking behavior, SSRI amelioration of these symptoms should lead to reduced drinking, a reduction of relapse, and the maintenance of abstinence. The selective 5-HT3 antagonist ondansetron is thought to have effects on alcohol reward and thereby reduce alcohol consumption and promote abstinence.

+ (2) Efficacy

There is considerable evidence that serotonin modulates the behavioral effects of alcohol (479, 1073–1075). SSRIs augment brain serotonergic function and have been shown to reduce alcohol consumption in animals. Medications in this class include fluoxetine, sertraline, paroxetine, and citalopram.

Several randomized, double-blind, placebo-controlled human studies with nondepressed heavy drinkers found SSRIs reduce alcohol consumption by 15%–20% in the short term (4–6 weeks) (1076, 1077). The results of these initial studies suggested that these medications could have an effect in reducing alcohol consumption. However, the results could not be easily generalized to the more severely impaired alcohol-dependent person. Indeed, results of trials using SSRIs to treat patients diagnosed with alcohol dependence have been less consistent (1078). A double-blind study comparing fluoxetine treatment with placebo in which both groups received CBT found no overall difference in drinking between groups (1079). However, further analysis showed that type B alcoholic individuals, characterized by early onset of drinking, heavier alcohol dependence, and greater co-occurring psychopathology, showed less favorable drinking outcomes in response to treatment with fluoxetine than with placebo. In contrast, the type A alcoholic individuals, characterized by later onset of drinking, less severity of dependence, and less psychopathology, did appear to benefit from fluoxetine. These findings were confirmed in a 14-week placebo-controlled trial of sertraline (200 mg/day) in alcohol-dependent patients stratified by subtype (1080). At this time, the treatment of nondepressed alcoholic patients with SSRIs remains controversial and may worsen drinking behaviors in some individuals.

Based on animal studies (1082, 1083) and early clinical laboratory findings (1084), the 5-HT3 receptor antagonist ondansetron was thought to have promise as a therapeutic agent in alcohol dependence, which led to its clinical evaluation. In a placebo-controlled outpatient trial with 71 mildly alcohol-dependent male subjects, a low dose (0.25 mg b.i.d.) but not a high dose (2.0 mg b.i.d.) of ondansetron moderately reduced alcohol consumption (1085). Using relatively low doses of ondansetron (2–32 g/kg per day), Johnson et al. (1086) conducted a 12-week placebo-controlled clinical trial in 271 alcohol-dependent patients who also received weekly CBT. Patients with early-onset alcoholism who received ondansetron showed significant reductions in drinking (especially in the 4 g/kg b.i.d. group) compared with those who received placebo. Patients with late-onset alcoholism had higher levels of drinking across all groups but showed no significant differences between medication and placebo treatment. The efficacy of ondansetron in the treatment of alcohol dependence in patients with early-onset alcoholism has been documented in only this single-site placebo-controlled study (1086) and in an open-label study in which the same dose of ondansetron was used (1652). Replication studies have yet to be conducted, and ondansetron has not been approved by the FDA for alcoholism treatment.

+ (3) Side effects

SSRIs are generally well-tolerated medications and have many patient-years of use worldwide. They appear to have a good safety profile in heavy alcohol consumers. However, some side effects of these medications clearly overlap with alcohol-related effects. The most frequent side effects of SSRIs occur in the gastrointestinal tract (nausea, diarrhea), which is also affected by alcohol, both during withdrawal and during active drinking. SSRIs have effects on the CNS (activation/sedation and headache) that also may occur during alcohol withdrawal and active drinking. Perhaps the most difficult side effects to evaluate in the alcoholic patient are potential SSRI-induced changes in libido and sexual performance. These complaints may occur during changes in alcohol consumption under normal conditions, and the differential role of SSRIs may be particularly hard to evaluate.

Ondansetron may cause some gastrointestinal disturbances, headache, and fatigue. It has been generally well tolerated at the low doses used in alcoholism trials to date.

+ (4) Implementation issues

Because there is an overlap of signs and symptoms of alcohol withdrawal and SSRI side effects, a few days of abstinence from alcohol and monitoring of amelioration of alcohol withdrawal effects should occur before initiating an SSRI. A more gradual titration might minimize any interaction of alcohol-related and SSRI effects.

Awareness and documentation of individual differences in sleep at baseline, which may or may not be alcohol related (334), will assist the clinician in evaluating the potential SSRI effect on sleep during treatment. For example, many alcoholic individuals drink to be able to get to sleep, so if an SSRI is causing nocturnal activation, this may not only reduce its effectiveness for alcohol treatment but also may be a reason to avoid its use. If an SSRI causes sedation during the day, this effect may be augmented by alcohol if an individual relapses.

Because SSRI-related sexual complaints overlap with alcohol-related effects, a good medical and longitudinal history of these complaints is likely to be helpful. Documentation of these issues before initiating SSRI treatment should also assist the clinician in assessing these complaints should they arise during treatment.

If ondansetron is to be prescribed, the clinician and patient must be reminded that the doses used in the published clinical trials have been an order of magnitude lower than what is generally prescribed for the treatment of nausea. Ondansetron is currently available in pill form but at a higher dose strength than that found to be efficacious in early-onset alcoholic patients. It is also available in liquid form. In either case, care must be taken to provide the correct dosing.

4. Psychosocial treatments
a) Cognitive-behavioral and relapse prevention therapies

There is abundant evidence that CBT approaches aimed at improving a patient's self-control and social skills consistently lead to reduced drinking (79, 1090, 1092–1094). Subtypes of CBT strategies have been supported with generally positive but variable degrees of success. Cognitive therapy interventions that focus on identifying and modifying maladaptive thoughts but that do not include a behavioral component have not been as effective as cognitive-behavioral treatments (1090) and were found to be effective in only 4 out of 10 studies reviewed by Miller and Wilbourne (79).

Behavioral self-control training consists of cognitive and behavioral strategies, including self-monitoring, goal setting, rewards for goal attainment, functional analysis of drinking situations, and the learning of alternative coping skills (1095, 1096). Holder et al. (1090) and Hodgeson (1097) both found that cognitive-behavioral stress management interventions were effective in the majority of studies reviewed. Although some studies of behavioral self-control training have included controlled drinking as well as abstinence as a goal for treatment, behavioral self-control techniques should be used with the explicit long-term goal of abstinence.

General self-control strategies include goal setting, self-monitoring, functional analysis of drinking antecedents, and learning alternative coping skills. Miller and Wilbourne (79) found that self-control training produced better outcomes than control treatments in 17 of 35 studies.

In several studies, increases in coping responses or "self-efficacy" (1098) at the end of treatment predicted better drinking outcomes during follow-up (184, 1099, 1100). Individuals who report more frequent use of cognitive or behavioral strategies aimed at problem solving or mastery ("approach coping") typically have better drinking outcomes than those who rely on staying away from high-risk situations ("avoidant coping") (959, 1101).

Outcome studies have typically supported the efficacy of group behavioral and CBT treatments, including group marital therapy. The results of studies matching patients to treatment in which patients were randomly assigned to cognitive-behaviorally oriented treatment groups or interactional therapy groups suggest that patients with fewer antisocial personality features or with neurological impairment fare better in interactional therapy; those with higher levels of antisocial personality features and psychopathology fare better in CBT groups (1102, 1103). Litt et al. (1104), in a randomized, controlled study, also found a positive effect of matching patients to treatment strategy.

+ (1) Relapse prevention strategies

In a meta-analysis of relapse prevention therapies, Irvin et al. (1094) found an effect size of 0.27 for treatment of alcohol use derived from five studies.

+ (2) Motivational enhancement therapy

MET and motivational interviewing, brief treatments aimed at maximizing the patient's intrinsic desire to change, were found to be efficacious in 7 of 9 studies reviewed by Dunn et al. (1105) and 12 of 17 studies reviewed by Miller and Wilbourne (79). This type of treatment is typically brief, lasting 1–4 sessions, and has been frequently used to enhance adherence with more intensive or extensive subsequent treatments. The use of MET as a stand-alone treatment was most notably supported by Project MATCH (43), in which 4 sessions of MET yielded equivalent results to 12 sessions of CBT or 12 sessions of TSF as an initial treatment or care after hospitalization. Alcohol-dependent patients in all three treatments experienced substantial and enduring improvement in drinking outcomes (43, 265).

b) Behavioral therapies

Individual behavioral therapy has been found to be effective for patients with an alcohol use disorder (191, 956, 1090), particularly those treatments that emphasize positive reinforcements for targeted behaviors. Behavioral contracting was found to be effective in four of five studies reviewed by Miller and Wilbourne (79). In contrast, relaxation training, a behavioral treatment that has been widely studied, was found to be ineffective in 17 of 18 controlled trials (79).

+ (1) Community reinforcement

The most comprehensive behavioral approach to the treatment of patients with alcohol use disorders is the community reinforcement approach, which uses behavioral principles and usually includes conjoint therapy, training in finding a job, counseling focused on alcohol-free social and recreational activities, monitoring of disulfiram use, and an alcohol-free social club (1107). Using random assignment to community reinforcement treatment or standard hospital treatments, Azrin (190) found that patients in the community reinforcement group drank less, spent fewer days away from home, worked more days, and were institutionalized less over a 24-month follow-up period. A second controlled study comparing 1) the community reinforcement approach, 2) disulfiram plus a behavioral adherence program, and 3) regular outpatient treatment showed that patients treated with community reinforcement did substantially better on all outcome measures than those in the other treatment conditions (190). A meta-analytic review conducted by O'Farrell and Fals-Stewart (690) documented a medium effect size for the community reinforcement approach combined with marital therapy.

+ (2) Aversion therapy

Compared with positive reward approaches, aversion therapies have been less successful. Only a small number of studies (12 of 28) have documented the efficacy for aversion therapy using nausea or electric shock (79).

c) Social skills training

Social skills training focuses on learning skills for forming and maintaining interpersonal relationships, being assertive, and refusing alcohol. Miller and Wilbourne (79) found social skills training to be effective in 17 out of 25 studies.

d) Psychodynamic and interpersonal therapies

Holder et al. (1090) concluded that there was little empirical evidence from controlled studies that insight-oriented psychotherapy or counseling is an effective treatment for an alcohol use disorder. Individual psychotherapy produced better outcomes than a control condition in 2 of 8 studies reviewed, and psychodynamically oriented group psychotherapy produced better outcomes in 2 of 11 studies. Empirical research on the efficacy of psychodynamic treatment for substance abuse is limited by the long-term nature of this approach and difficulties in developing representative training manuals. However, there is a large body of clinical literature documenting success in individual patients in uncontrolled conditions (1653). A more recent review found support for individual therapy in 11 of 18 trials reviewed, although the quality of the studies was noted to be generally poor (79).

e) Self-help groups and 12-step-oriented treatments

The effectiveness of AA, per se, has not been evaluated in randomized studies because of a host of ethical and practical problems associated with assigning patients to a group that does not attend AA (261). However, there is growing support for the utility of AA and 12-step-oriented treatments from a range of sources (reviewed by McCrady and Irvine [1115]). Several naturalistic studies have suggested that AA can be an important support for promoting an alcohol-free lifestyle in patients who are willing to attend (956, 958, 959). A large number of studies have documented that more AA participation is linked to better drinking outcomes (260–266, 289, 1113, 1114). For example, in an evaluation of 8,087 patients in 57 programs, Hoffman and Miller (1112) reported that those attending AA at 1-year follow-up were 50% more likely to be abstinent than those not attending AA.

Professional therapies based on AA have been also found to be effective (1115). Moos and colleagues (260, 262–264, 266), in a series evaluating outcome in alcoholic veterans treated in a 12-step or CBT-oriented program, documented overall better outcome for 12-step treatments and a strong association between AA involvement and improved outcome. The findings of Project MATCH were generally positive for TSF, an individual-based, professionally delivered psychotherapy aimed at motivating patients to adhere to AA principles. Compared with MET and CBT, main effects for drinking outcomes were comparable or modestly superior for TSF (43, 265), and selected significant matching effects favored TSF for patients who had low severity of psychiatric symptoms, rated high on seeking meaning of life, and had high social support for not drinking.

Regarding patient-program matching, evidence from small-scale trials has suggested that patients with more severe drinking problems, an affective rather than a cognitive focus, a concern about purpose and meaning in life, better interpersonal skills, and a high need for affiliation are good candidates for AA (1115, 1654). In the landmark study Project MATCH, aftercare patients who rated high in the seeking meaning category fared better with TSF compared with MET and CBT at 1-year follow-up. Also, TSF was more effective than CBT for outpatients who did not show psychiatric symptoms and comparably equivalent for those with symptoms. Patients with high social support for not drinking had better drinking outcomes over the 3-year posttreatment period if they were treated with TSF rather than MET (1- and 3-year reports) (43, 265).

f) Brief interventions

Brief interventions generally delivered over one to three sessions include an abbreviated assessment of drinking severity and related problems and the provision of motivational feedback and advice. Brief therapies have typically been studied in general medical settings or school-based settings and have focused on non-treatment-seeking heavy drinkers who do not meet criteria for alcohol abuse or dependence. In 22 of 31 controlled treatment trials reviewed by Miller and Wilbourne (79), brief interventions were found to be effective. Reviews by Babor (1109) and Bien et al. (275) concluded that brief interventions 1) are typically more effective (in terms of alcohol use, general health, or social functioning) than no intervention; 2) often have efficacy comparable with that of traditional, more intense, longer-term programs; and 3) increase the effectiveness of later treatment. Even interventions that are very brief (i.e., a few hours) may have some positive effect (1110). For example, Fleming et al. (1111) conducted a trial of a brief physician advice condition plus a general health booklet versus the general health booklet alone in 774 primary care patients who screened positive for an alcohol use problem. This sample comprised 4.4% of 17,795 general medical patients screened. At 12-month follow-up, two 15–20-minute sessions of physician advice yielded significantly greater improvements in the number of alcoholic drinks taken over 7 days, the 30-day rate of binge drinking, and the rate of excessive drinking. Further research is needed to determine which patients are optimally served by receiving a brief intervention.

g) Marital and family therapy

The state of the patient's relationship with family members or significant others can be a critical factor in the posttreatment environment for patients who are married, in a committed relationship, or living with family members (1090, 1117). A meta-analytic review by O'Farrell and Fals Stewart (690) found a moderate effect size in 16 controlled trials of behavioral marital therapy. Other marital approaches with significant support were Al-Anon facilitation and disulfiram contracting (168, 248). Similarly, Miller and Wilbourne's box score review (79) found five of eight high-quality studies favored the efficacy of behavioral marital therapy; in contrast, other approaches to marital therapy were found to be effective in only three of eight trials (79). A noteworthy series of trials by O'Farrell and colleagues (225, 236, 238, 690, 961, 1118, 1119) demonstrated not only the efficacy but also the cost-effectiveness of behavioral marital therapy for patients with an alcohol use disorder.

h) Aftercare

Walker et al. (1120) found that involvement in aftercare was a stronger predictor of outcome than length of hospitalization, neuropsychological functioning, or pretreatment drinking and social stability measures. Ouimette et al. (264) found that participation in both aftercare and 12-step groups was associated with better 12-month outcomes in a sample of hospitalized male veterans. McLatchie and Lomp (1121) randomly assigned patients to mandatory, voluntary, or no aftercare for a 12-week period and found that those who completed aftercare had the lowest relapse rate, with no difference between the mandatory and voluntary groups. Gilbert (1122) randomly assigned patients to one of three aftercare conditions that varied in the degree of the therapist's efforts to maintain the patient in aftercare over 30 appointments. Patients in the maximal effort group were the most likely to complete aftercare, and all who completed aftercare, regardless of their study group, had better outcomes than those who did not. Results from studies that did not include random assignment suggest that greater participation in aftercare is generally associated with fewer drinks on drinking days, but not with diminished frequency of drinking (1655). A controlled study by O'Farrell et al. (225, 1118, 1119) demonstrated that a version of behavioral marital therapy that included relapse prevention techniques (184) and was delivered as an aftercare intervention led to better drinking outcomes. In two separate trials, Kadden and colleagues (223, 1102, 1103) compared inpatient aftercare programs consisting of 1) CBT and coping skills training and 2) insight-oriented interactional group therapy and reported similar outcomes in the two groups. Patterson et al. (1123) found nurse visits delivered over 12 months yielded better abstinence rates than clinical-based review visits. In Project MATCH (43, 265), 12 weeks of individual aftercare treatment using MET, TSF, or CBT showed comparable positive drinking outcomes in a sample of 774 alcohol-dependent patients. Also, 1 of 10 hypothesized matching effects was significant, suggesting that TSF was superior to CBT and MET for patients rated high in the seeking meaning category.


C. Marijuana-Related Disorders

1. Somatic treatments

There have been no successful controlled trials to date of pharmacotherapy for marijuana dependence. Trials with negative results have studied bupropion (1182), divalproex (1183, 1184), naltrexone, and nefazodone (1185). The main active ingredient of cannabis, -9-tetrahydrocannabinol, has been tried in a laboratory study with human research volunteers and found to reverse withdrawal-associated psychomotor performance impairment and weight loss (1184) and warrants further study. No pharmacotherapy trials to prevent marijuana reinstatement after abstinence have been reported.

2. Psychosocial treatments

Given the absence of effective pharmacotherapies for marijuana dependence, the treatment of marijuana-related psychiatric disorders has primarily focused on psychosocial approaches (1178). Controlled trials have used cognitive-behavioral relapse prevention group therapy, social support group treatment, contingency management therapies, motivational individualized assessment and intervention, and MET therapy. However, it is difficult to discuss comparative efficacy across trials because the trials differed methodologically (e.g., in the diagnostic criteria and control groups used, the length of treatment and follow-up, the use of urine toxicology screens to confirm marijuana abstinence, the way in which interventions were delivered). Study samples also differed in their size and ethnic diversity. In general, existing trials consistently support the efficacy of the active treatments being studied.

a) Cognitive-behavioral and relapse prevention therapies

A number of studies in individuals with marijuana-related problems have compared outcomes after motivational or CBT interventions (1178). Stephens et al. (1187) randomized 212 subjects to receive relapse prevention therapy or a social support discussion group intervention. Both treatment conditions were associated with significant reductions in marijuana use relative to baseline, although no significant group differences were found in abstinence rates, marijuana-related problems, or days of marijuana use. Subsequently, 291 subjects were randomized into a delayed-treatment control group, a two-session motivational treatment group, and an intensive (14-session) relapse prevention treatment group (1186). Although no significant differences were observed between the brief and the more intensive treatment, marijuana-related outcomes for the two active treatments were found to be better than those with the delayed-treatment control condition. More recently, a replication and extension of that study involving a multisite trial of 450 marijuana-dependent patients compared three approaches: 1) a delayed-treatment control, 2) a two-session motivational approach, and 3) a nine-session combined motivational and coping skills approach (276). The results suggested that both active treatments were associated with significantly greater reductions in marijuana use than the delayed-treatment control condition at 4- and 15-month follow-up. Moreover, the nine-session intervention was significantly more effective than the two-session intervention, and this effect was sustained to 15-month follow-up.

Copeland et al. (1188) also studied CBT interventions for cannabis dependence, randomly assigning 229 participants to a six-session CBT program, a single-session CBT intervention, or a delayed-treatment control group. Participants in the treatment groups were assisted in acquiring skills to promote cannabis cessation and maintain abstinence. At about 6-month follow-up, both treatment groups reported better outcomes (greater abstinence rates, fewer marijuana use–related problems, less concern about control over marijuana use) than the delayed-treatment control group.

b) Behavioral therapies

Budney et al. (201) applied contingency management to marijuana users. They randomized 60 subjects to receive 4 sessions of MET, 14 sessions of MET in combination with coping skills therapy, or 14 sessions of MET in combination with coping skills therapy and voucher incentives. Although no differences in abstinence rates were noted between individuals receiving 4 sessions of MET or 14 sessions of MET plus coping skills therapy by the end of treatment, greater rates and durations of abstinence were seen in the group receiving MET plus coping skills therapy and voucher incentives. Thus, adding voucher-based incentives to coping skills and MET appears to improve outcomes during treatment for marijuana dependence.


D. Cocaine-Related Disorders

1. Somatic treatments
a) Medications to treat cocaine dependence

More than 45 different medications have been studied in the search for an effective pharmacological treatment for cocaine dependence (1225). Although a number of studies have shown promising results with a variety of pharmacotherapeutic agents, no medication has been found to have clear-cut efficacy in the treatment of cocaine dependence (1222–1224). Most studies have been hampered by methodological problems, including lack of adequate controls and consistent outcome measures (e.g., urine tests rather than self-reports), failure to standardize the type and "dose" of the accompanying psychosocial interventions, lack of clarity about the importance of craving in the maintenance of cocaine dependence, the role of craving in the natural course of untreated cocaine abstinence syndrome, and lack of agreement as to the exact meaning of the term "craving" (1204, 1209).

+ (1) Antidepressants

Gawin et al. (1227) found desipramine to be more effective than lithium or placebo in reducing cocaine use by outpatients without a coexisting mood disorder. Other reports (1215, 1230, 1231) failed to confirm these positive findings, possibly because of differences in patient population and route of cocaine administration. A subsequent study of desipramine and placebo with and without psychotherapy showed improvement with desipramine compared with placebo in the short term (6 weeks) but not at 12 weeks or 1 year (503). In buprenorphine-treated patients, desipramine was better than placebo for cocaine use (1228), and in methadone-treated patients, contingency management with desipramine produced more cocaine abstinence than desipramine alone, contingency management alone, or no treatment (1229).

Fluoxetine (1232, 1233) and bupropion (1234) also showed some benefit in small studies but demonstrated no superiority to placebo when evaluated in larger trials (413, 1235–1238).

+ (2) Dopamine agonists

Dopamine agonists are another class of agents that have been used in the treatment of cocaine dependence. The evidence for amantadine's effectiveness is inconsistent in that efficacy was observed in one controlled trial (1240) but not in two others (1225, 1231). Another controlled trial with amantadine found no overall difference between individuals receiving amantadine and those receiving placebo (1239), although those with more severe withdrawal symptoms appeared to have a better response to amantadine (1226).

The selective MAOI selegiline showed safety and some promise in laboratory studies in humans during cocaine administration, but a recent large multisite study failed to show its superiority to placebo (1241; F. Vocci, personal communication). Evidence for the benefits of l-dopa/carbidopa is limited (1242). Pergolide has been studied in larger trials and shown to have no superiority over placebo (1220, 1243). Laboratory studies in humans have suggested that a dopamine D1 agonist (ABT-431) might reduce cocaine craving, but no oral formulations are available for outpatient trials (1656, 1657). Finally, replacement therapies using methylphenidate or sustained-release amphetamine have been superior to placebo for patient retention and reduction in cocaine use, but these studies need further replication (1244, 1245, 1247, 1658, 1659).

+ (3) Opioid-related agents

The mixed opioid agonist-antagonist buprenorphine has shown some promise in open trials in the treatment of patients dependent on both cocaine and opioids (1248, 1249), although three large-scale double-blind, clinical trials comparing patients maintained on buprenorphine with those receiving methadone showed no decrease in cocaine use among the former group (1250–1252). Although work by Schottenfeld et al. (1253) suggests that higher doses of buprenorphine (12–16 mg/day) may be effective, larger-scale clinical trials have found this medication to be effective only in combination with desipramine or, particularly, contingency management (1228, 1229, 1254). Naltrexone has also been tested and shown to be not useful for cocaine dependence (1255).

+ (4) Other medications

Anticonvulsants and agonists for GABA have been studied with some promise. Initial studies of carbamazepine in the treatment of cocaine dependence yielded some favorable results (1256), but subsequent double-blind, placebo-controlled studies failed to establish the efficacy of this drug in cocaine-dependent patients (1257–1261). A pilot study of the anticonvulsant topiramate showed promise (1262). The GABAB agonist baclofen has shown some success in treatment (1263), and a recent double-blind clinical trial of tiagabine, a GABA-reuptake blocker, was superior to placebo for reducing cocaine use (1264). Finally, modafinil shows promise for reducing cocaine abuse (1265).

b) Medications to change the subjective effects of cocaine

Attempts to find a medication that blocks or attenuates the subjective (e.g., euphorigenic) effects of cocaine have included trials of neuroleptics (1268, 1269), mazindol (1270), disulfiram, and a cocaine vaccine. At doses that can be tolerated by patients, neuroleptics and mazindol have not been effective (1271–1273). However, recent data with disulfiram have suggested that it may increase the aversive effects of cocaine and reduce its use (1277, 1660). Animal studies have demonstrated that a cocaine vaccine may form sufficient antibodies to reduce cocaine use (1278).

c) Acupuncture

Although it is not a pharmacological treatment, acupuncture is a somatic treatment that has been frequently used in the treatment of patients with a substance use disorder. Two recent randomized, controlled trials, however, one with 412 subjects (1279) and one with 620 subjects (1280), compared auricular acupuncture (which is supposed to be specifically helpful for patients with a substance use disorder) with a needle insertion control condition (sham acupuncture); the latter study also had a relaxation control condition. In both studies, acupuncture was no more effective than the sham acupuncture control or the relaxation condition in reducing cocaine use. The results of these studies, therefore, do not support the use of auricular acupuncture as a sole treatment for cocaine dependence.

d) Pharmacological treatment in individuals with a co-occurring psychiatric disorder

Several focused and reasonably well-controlled studies have examined the use of second-generation antipsychotics as anticraving agents among individuals with schizophrenia who primarily abuse cocaine. A 6-week open-label study showed that compared with patients taking first-generation antipsychotics, those receiving risperidone showed a significant reduction in cue-elicited cocaine craving, relapse to substance use, and symptom severity (401). A double-blind, randomized trial of olanzapine versus haloperidol was conducted with 31 cocaine-dependent patients with schizophrenia (1661). At the study completion, patients treated with olanzapine showed significantly less cue-elicited craving on two of four craving dimensions and fewer relapses compared with those treated with haloperidol.

2. Psychosocial treatments
a) Cognitive-behavioral therapies

A number of randomized clinical trials among several diverse cocaine-dependent populations have demonstrated that 1) compared with other commonly used psychotherapies for cocaine dependence, CBT appears to be particularly more effective with more severe cocaine users or those with co-occurring disorders (229, 452, 503, 1289, 1292, 1293); 2) CBT is significantly more effective than less intensive approaches that have been evaluated as control conditions (267, 1288); and 3) CBT is as effective or more effective than manual-guided disease-model approaches (267, 1289). Moreover, CBT appears to have a particularly durable impact, with patients continuing to reduce their cocaine use even after they leave treatment (1275, 1290, 1291, 1662).

b) Behavioral therapies
+ (1) Contingency management

Perhaps the most exciting findings pertaining to the effectiveness of behavioral treatments for cocaine dependence have been the reports by Petry et al. (188) and Higgins et al. (191–194) on the use of behavioral incentives for abstinence. The strategy of Higgins's group has four organizing features that are grounded in principles of behavioral pharmacology: 1) drug use and abstinence must be swiftly and accurately detected, 2) abstinence is positively reinforced, 3) drug use results in loss of reinforcement, and 4) emphasis is placed on the development of reinforcers to compete with drug use (1294). In this approach, urine specimens are required three times a week to systematically detect all episodes of drug use. Abstinence, verified through drug-free urine screens, is reinforced through a voucher system in which patients receive points redeemable for items consistent with a drug-free lifestyle (e.g., movie tickets, sporting goods).

In a series of well-controlled clinical trials, Higgins's group has demonstrated high rates of acceptance, retention, and abstinence in patients receiving this approach, as compared with standard counseling oriented toward 12-step programs (191, 192). Rates of abstinence do not decline substantially when less valuable incentives are substituted for the voucher system (192). The value of the voucher system itself, as opposed to other program elements, in producing good outcomes was demonstrated by comparing the behavioral system with and without the vouchers (193). Although the strong effects of this treatment declined somewhat after the contingencies were terminated, the voucher system has been shown to have durable effects (194). Moreover, the efficacy of a variety of contingency management procedures (including vouchers, direct payments, and free housing) has been replicated in other settings and samples, including cocaine-dependent individuals within methadone maintenance (195, 1295), substance-abusing homeless individuals (1297), freebase cocaine users (1298), and pregnant drug users (1299).

These findings are of great importance because contingency management procedures are potentially applicable to a wide range of target behaviors and problems, including treatment retention and adherence with pharmacotherapy (e.g., retroviral therapies for individuals with HIV). For example, Iguchi et al. (202) showed that contingency management can be effective in reinforcing desired treatment goals (e.g., looking for a job) in addition to abstinence.

In a line of research attempting to reduce the costs of contingency management, Petry and colleagues (204, 205) demonstrated that a variable ratio schedule of reinforcement that provides access to large reinforcers but at lower probabilities is effective in retaining subjects in treatment and reducing substance use. Rather than earning vouchers, subjects earn the chance to draw from a bowl and win prizes of varying magnitudes. The prizes range from small $1 prizes (e.g., bus tokens, McDonald's coupons) to large $20 prizes (e.g., portable radios, watches, and phone cards) to jumbo $100 prizes (e.g., small televisions). This system is far less expensive than the standard voucher system because only a proportion of behaviors are reinforced with a prize. In a study of 42 alcohol-dependent veterans who were randomly assigned to standard treatment or standard treatment plus contingency management, 84% of the contingency management subjects were retained in treatment throughout an 8-week period compared with 22% of standard treatment subjects. By the end of the treatment period, 69% of those receiving contingency management had not experienced a relapse to alcohol use, but only 39% of those receiving standard treatment were abstinent (205); there were similar findings among cocaine abusers (204).

+ (2) Cue exposure treatment

Cue exposure therapy has also been used in the treatment of cocaine use disorders, with equivocal results (189).

c) Psychodynamic and interpersonal therapies

No randomized clinical trials have been conducted for psychodynamically oriented treatments for cocaine abuse or dependence. A case series of patients successfully treated with individual psychodynamically oriented psychotherapy was reported by Schiffer (1300), and there is a preliminary report revealing a high rate of retention with modified psychodynamically oriented group psychotherapy (1301). Spitz (1302) has also described the use of group therapy for this population.

d) Self-help groups and 12-step-oriented treatments

Several trials have evaluated manual-guided professionally delivered treatments, including TSF (268) and individual drug counseling (269), that enhance a patient's motivation to participate in 12-step programs. As part of these professional treatments, patients are actively encouraged (but not required) to attend Narcotics Anonymous or Cocaine Anonymous meetings, become involved in traditional fellowship activities, and maintain journals of their self-help group attendance and participation.

In a comparison of TSF, CBT, and clinical management (a supportive approach in which patients received comparable empathy, support, and other "common elements" of psychotherapy but none of the unique "active ingredients" of TSF or CBT) for alcoholic cocaine-dependent individuals, TSF was found to be significantly more effective than clinical management and was comparable with CBT in reducing cocaine use (267). In addition, results at 1-year follow-up suggested that gains from treatment were maintained for subjects who received TSF or CBT; these subjects reported continuing to reduce their cocaine use throughout the follow-up period compared with subjects who received clinical management. Moreover, there was a strong association between the attainment of significant periods of abstinence during treatment and abstinence during follow-up, which emphasizes that the inception of abstinence, even for comparatively brief periods, is an important goal of treatment (194, 1275).

The results of the NIDA Collaborative Cocaine Treatment Study, a multisite randomized trial of psychotherapeutic treatments for cocaine dependence (219, 1305), suggest the effectiveness of a similar approach: 12-step-oriented individual drug counseling (269). In that study, 487 cocaine-dependent participants in four sites were randomly assigned to one of four conditions: 1) cognitive therapy (1306) plus group drug counseling; 2) supportive-expressive therapy, a psychodynamically oriented approach (217) plus group drug counseling; 3) 12-step-based individual drug counseling plus group drug counseling; or 4) group drug counseling alone. The treatments offered were intensive (36 individual and 24 group sessions over 24 weeks, for a total of 60 sessions) (219). On the whole, outcomes were good, with all groups significantly reducing their cocaine use from baseline; however, the best outcomes were seen for subjects who received individual drug counseling.

McKay et al. (229) compared 12-step-oriented standard group counseling to relapse prevention aftercare in 98 cocaine-dependent patients. Rates of complete abstinence in the 6-month study period were higher in the standard group counseling group than in the relapse prevention group, whereas relapse prevention was more effective in limiting the extent of cocaine use in those who currently used cocaine. Overall effectiveness was judged to be equivalent for the two treatments.

The studies above evaluated professional treatments based on 12-step concepts and not participation in self-help groups, per se. The literature on participation in Cocaine Anonymous or other 12-step self-help groups is more limited than that for patients with an alcohol use disorder participating in AA. In one study of day hospital rehabilitation for patients with a cocaine use disorder (1304), greater participation in self-help programs 3 months after treatment predicted less cocaine use 6 months after treatment, even after pretreatment patient characteristics and degree of success in the day hospital program were controlled for in the study.

Self-help groups have not been shown to be a sufficient alternative to professional treatment. For example, a large randomized trial that directly compared referral to self-help with professional treatments found poorer outcomes, with high rates of treatment utilization for the patients referred to self-help compared with inpatient treatment (1303).


E. Opioid-Related Disorders

1. Somatic treatments
a) Opioid agonist therapies
+ (1) Methadone

Methadone is the most thoroughly studied and widely used pharmacological treatment for opioid dependence. Studies of its efficacy and safety have focused on its use as a maintenance medication and a medication for the treatment of opioid withdrawal. This section will review studies of methadone's efficacy and safety for these two purposes.

+ a. Use of methadone as a maintenance agent

Studies of methadone's efficacy and safety fall into two general categories: controlled clinical trials and naturalistic/survey studies. When designed and conducted properly under controlled conditions that are related to but distinct from routine clinical practice, clinical trials provide evidence of methadone's efficacy. The scientific value inherent in the methods of such studies (e.g., double-blind dosing) is offset by the somewhat unusual experimental conditions of a clinical trial (e.g., intensive monitoring of participants). Naturalistic survey studies of methadone provide complementary evidence to clinical trials and typically report results for larger populations that have been treated in routine settings. There is little control for factors such as expectancy, but these studies do provide data that are more closely tied to real-world clinical settings.

+ 1) Naturalistic/survey studies of methadone treatment

Examples of naturalistic/survey studies of methadone efficacy include the Drug Abuse Reporting Program (61), the Treatment Outcome Prospective Study (1399), the Drug Abuse Treatment Outcome Study (45, 73, 1663), and the Effectiveness of Methadone Maintenance Treatment Study (169). The first three of these survey studies assessed methadone treatment as one of several substance abuse treatment modalities. Participants in these projects were not randomized to a treatment modality, and services were given in routine clinic settings and were not delivered in a blinded fashion. Results from such studies generally showed that methadone is effective when post- and pretreatment functioning are compared and that better outcomes are associated with longer periods of treatment. However, the relative efficacy of different doses of methadone has generally not been addressed in such survey studies.

The Effectiveness of Methadone Maintenance Treatment Study (169) was somewhat different from these national survey studies, as it assessed methadone treatment in a relatively restricted geographic region—six clinics located on the East Coast (two each in New York, Philadelphia, and Baltimore). However, the study did provide intensive evaluation of a large number of patients treated specifically with methadone (versus a more heterogeneous population of patients in the other surveys). A total of 617 patients were initially assessed; of these, 126 were new methadone admissions, 346 were in treatment for <4.5 years, and 145 were in treatment for 4.5 years.

At the 1-year follow-up, the methadone dose was inversely related to self-reported heroin use in the 30 days prior to the interview. The study found that longer time in treatment was associated with decreased rates of intravenous drug use. For patients in treatment for 4 years, self-reported use declined from 81% at the time of admission to 29% after 4 years. Among patients who had been in treatment for 4.5 years, 92% reported no heroin use and 83% reported no cocaine use in the 30 days before the study assessment. Other factors besides methadone dose were also found to be related to treatment and outcome in this study, such as the level of involvement of the clinic director.

Other naturalistic studies of methadone treatment provide further evidence of methadone's efficacy, especially with regard to methadone dose. For example, one study of 652 methadone maintenance patients found that heroin use was greatest among those with daily methadone maintenance doses <70 mg/day and that, independent of dose, time in treatment was associated with less heroin use (1664). Similarly, a review of methadone dosing for 62 patients treated in an Australian clinic found that higher doses were associated with less heroin use (1665). This study also concluded that the relative odds of heroin use were reduced by 2% for every 1 mg increase in maintenance dose.

It is important to note that an inherent weakness to survey studies is that patients are not randomly assigned to the conditions being compared. There may be other patient or program characteristics besides methadone dose or time in treatment that may account for the differences seen in such studies. In addition, these studies often rely on patient self-reports of substance use, including retrospective assessments of past use that are then used to calculate changes over time. Expectancy effects may also play an important role in outcomes seen. Thus, although survey studies can be helpful and informative, these and other liabilities compromise the degree of certainty that can be assigned to their conclusions.

+ 2) Clinical trials of methadone dose and treatment

Numerous clinical trials have tested methadone for the treatment of opioid dependence. In general, these studies have found that methadone medication is a safe and effective pharmacotherapy for the treatment of opioid dependence and that methadone combined with other services such as counseling, behavioral interventions, and urine monitoring can broaden the efficacy of this intervention to include a greater range of treatment outcomes such as increased prosocial behavior and decreased nonopioid substance use.

There have been two double-blind, placebo-controlled studies of methadone for the treatment of opioid dependence. The first was conducted in Hong Kong in the early 1970s and enrolled 100 male opioid-dependent patients who were initially hospitalized for 2 weeks and treated with 60 mg/day of methadone (1666). Participants were then randomly assigned to 1) the placebo condition, in which they underwent 1-mg dose reductions of methadone each day on an outpatient basis and then were maintained on placebo after 60 days, or 2) the maintenance condition, in which they could have outpatient dose adjustments with a maximum possible dose of 130 mg/day. For this latter group, the average dose was 97 mg/day (range 30–130 mg) after 1 year. Subjects were withdrawn from the study if they missed 6 consecutive weeks of treatment or had six consecutive opioid-positive urine samples. At the 32-week evaluation point, 10% of placebo-treated patients remained in treatment (most of the rest having been withdrawn because of urine evidence of persisting illicit opioid use) compared with 76% of maintenance patients.

The second placebo-controlled study of methadone treatment was conducted in the late 1980s in Baltimore, Maryland, and enrolled 247 opioid-dependent outpatient participants (1350, 1351). The study began by randomly assigning subjects to one of three fixed doses of oral methadone (0, 20, or 50 mg/day). Participants were required to attend the clinic daily for supervised dose ingestion and were discharged for missing 3 consecutive days of treatment. Primary outcomes were treatment retention and urine test results. At the end of the 20-week study, significant differences were found among the three groups for treatment retention (primarily between the 50- and 0-mg groups), with the 20-mg group generally doing better than the 0-mg group. In addition, there were significantly lower rates of opioid-positive urine test results for the 50-mg group compared with the other two groups. A variety of secondary outcome measures, such as self-reported illicit opioid use, also showed dose-related effects. In addition to the demonstrated dose-related efficacy of methadone, these study results also indicated that the 20-mg dose of methadone might keep some patients in treatment but was not as effective in decreasing illicit opioid use as the 50-mg dose. Study limitations included the use of fixed doses of methadone and the absence of a dose condition 50 mg/day.

Further controlled clinical trials of methadone have tested higher versus lower doses of methadone. For example, a 40-week multicenter VA outpatient study of LAAM for the treatment of 430 opioid-dependent males included two fixed methadone dose control conditions of 50 and 100 mg/day (1349). The induction procedure for the 288 methadone-treated subjects was relatively slow; subjects were started on 30 mg of methadone per day and received 10-mg dose increases once per week until the target dose was achieved. Induction, therefore, lasted 5 weeks longer for patients in the 100-mg versus the 50-mg methadone group. Urine samples were collected and tested once per week, and the results were summarized using a set of rules that weighted results based on when the sample was collected and how missing values were handled. Outcomes for the methadone-treated subjects showed higher opioid urine scores (poorer outcomes) for the 50-mg versus the 100-mg group. Although this study provided evidence of methadone's dose-related efficacy on illicit opioid use, its results are limited by its use of an all male population, a slow induction procedure, fixed doses, and a somewhat unusual method for summarizing urine test results.

Another outpatient study compared a moderate dose (40–50 mg/day; N=97) with a higher dose (80–100 mg/day; N=95) of methadone for the treatment of opioid dependence (1352). This 40-week double-blind, randomized trial used a flexible dosing procedure in which participants could receive dose increases based on evidence of continued illicit opioid use. Primary outcome measures were treatment retention, the results of twice-weekly urinalyses, and self-reported illicit opioid use. The results showed no significant difference in treatment retention for the two groups but found a significantly lower rate of opioid-positive urine samples for the higher-dose condition. Both groups had marked declines in self-reported illicit opioid use, with significantly less use by the high-dose versus the moderate-dose group. Although significant effects were found on some outcomes in this study, both doses produced clinically meaningful decreases in illicit opioid use. The lack of difference between the study groups for treatment retention suggests that there may be a plateau in the dose-related efficacy of methadone in maintaining patients in treatment but not in decreasing illicit opioid use for the doses tested. However, the schedule of twice-weekly urinalyses used in this study may have failed to capture all illicit opioid use occurring in the study population.

Other controlled trials of methadone treatment and methadone dosing have also been conducted (1250, 1251, 1667–1670). In general, these studies have shown that methadone has dose-related efficacy, although it is important to note that not all randomized double-blind methadone studies have shown such an effect. However, it is also important to note that no double-blind, randomized, controlled clinical trials have tested daily doses of methadone 100 mg/day. There have been single-blind and open studies of higher doses of methadone that were conducted primarily in the early years of methadone treatment (1671–1673), and reports from clinical practice in both the United States and other countries indicate higher doses of methadone are used by some clinicians (1342–1346). Currently, there is no research database that provides information about the relative efficacy and safety of higher doses (i.e., 100 mg/day) of methadone.

+ b. Use of methadone as a withdrawal (detoxification) agent

The number of studies examining methadone for treating opioid withdrawal is more limited than the number examining methadone in maintenance treatment of opioid dependence. Outcomes from methadone withdrawal are generally poor (1674–1676), especially when compared with the success associated with methadone maintenance treatment. These studies have examined the various parameters under which methadone tapering can occur in an effort to determine optimal withdrawal schedules.

An early double-blind, randomized, outpatient study of methadone withdrawal by Senay et al. (1677) used a four-group design: dose reductions of 10% per week for 10 weeks, reductions of 3% per week for 30 weeks, stable dosing with patients knowing their dose would not change, and double-blind stable dosing. The 127 study participants were in methadone maintenance treatment, with an average dose of 31 mg/day. They could request a 1-week dose halt or increase during their withdrawal. Results from the study showed the poorest outcomes occurred for patients in the rapid dose-reduction (10% per week) group as measured by taper interruptions, positive urine sample rates, and withdrawal symptom complaints. As a group, patients undergoing the rapid withdrawal essentially stabilized around an average of 10 mg/day of methadone due to their requests for dose halts and temporary dose increases. These results suggested that a more gradual methadone taper (3% per week) leads to better outcomes than a more rapid taper (10% per week), although methadone maintenance treatment is even more effective.

Another randomized clinical trial compared methadone withdrawal—120 days of methadone induction/stabilization followed by a 60-day withdrawal and then 8 months of nonmethadone treatment—to 14 months of maintenance treatment in 179 opioid-dependent patients (1678). The study was not conducted in a blinded fashion, and the withdrawal group had more nonpharmacological services available to them. Results from the study showed significantly better treatment retention for the maintenance group but similar rates of illicit opioid use for the two groups until month 5, when patients withdrawn ftom methadone began to have higher rates of illicit opioid use. These study results support the value of methadone for maintenance treatment compared with withdrawal from methadone, although certain qualifiers to the study should be noted: expectancy effects could contribute to the outcomes shown, the requirement that withdrawn patients attend more groups and counseling may have contributed to the high dropout rate, and the length of the withdrawal (60 days) may have been too quick, as suggested by the results from the Senay et al. study (1677) reviewed above.

Other studies of methadone withdrawal have been conducted but generally with smaller sample sizes or in atypical treatment settings (e.g., inpatient units) that make their results limited in their applicability to current treatment circumstances. For example, an inpatient study of a 10-day methadone withdrawal comparing a linear schedule of daily dose reductions (based on a fixed percentage of the starting dose) to an exponential schedule of daily dose reductions (based on a percentage of the previous day's dose) found no differences in the time course of withdrawal symptoms, peak symptom severity, and patient adherence (1679). Other studies (1675, 1677, 1680) examining whether or not informing patients about their methadone dose-reduction schedule influences treatment outcome have concluded that informed patients have better outcomes. There is some evidence that patients do not have better outcomes, however, if they, rather than clinicians, are allowed to control their methadone withdrawal schedule (1675, 1681). Finally, one study found that lower rates of illicit opioid use occurred when patients received voucher incentives for opioid-negative urine samples during methadone withdrawal, although it appears this intervention primarily delayed relapse to illicit opioid use (1682).

+ c. Safety and side effects of methadone

Like all mu agonist opioids, overdose with methadone can produce respiratory depression and death. When used under physician supervision, methadone is a safe medication with limited side effects. Constipation, sweating, sexual dysfunction, and sedation are the most commonly reported side effects. When patients first start on methadone, they may experience some side effects (e.g., sedation) to which they will develop a tolerance. The two most common side effects reported by patients, constipation and sweating, may persist despite chronic treatment on a stable dose. Small effects that are not clinically significant have been noted on some physiological measures assessed during chronic methadone dosing (e.g., heart rate, pupil diameter) (1359, 1683). Studies of chronic dosing with methadone have shown no evidence of organ damage, such as hepatotoxic effects (1684–1686). There have been case reports of high-dose methadone treatment associated with torsade de pointes and a prolonged corrected QT interval (1687–1689). Methadone-related increases in the corrected QT interval appear to be related to dose (1690–1692), and patients experiencing this problem often have had other risk factors for torsade de pointes.

Early studies of psychomotor and cognitive performance tasks in methadone-maintained patients showed no significant impairments (1356–1358), but more recent studies have shown significantly poorer performance on these tasks in methadone-maintained patients versus matched control subjects (1359, 1360). Whether such effects are due to the acute or chronic effects of methadone or to other factors associated with a long history of substance use is not entirely clear despite efforts in these more recent studies to carefully match patients.

+ (2) LAAM

LAAM is an opioid agonist medication that has effects similar to those of methadone but a longer duration of action. LAAM has demonstrated efficacy as a maintenance agent in large multisite, randomized, double-blind studies (1349, 1693); smaller randomized, double-blind studies (1667, 1694–1696); and nonblind clinical trials (1697–1705). It also appears to be effective as a withdrawal (detoxification) agent (1674, 1706). Clinical trials with LAAM reported typical opioid agonist side effects (1349, 1707) but no cases of deaths or other significant adverse events that seemed linked to LAAM use. However, later case reports of corrected QT interval (QTc) prolongation and torsade de pointes (1337) led to LAAM's withdrawal from the U.S. and European markets.

+ (3) Buprenorphine

Clinical studies of buprenorphine have used two different formulations: solution and tablets. Because of its poor oral bioavailability, most clinical trials of buprenorphine for the treatment of opioid dependence have used a sublingual, often alcohol-based, solution. A liquid sublingual form of buprenorphine, however, was felt to be an impractical long-term treatment option, and as the evidence for its efficacy and safety in the treatment of opioid dependence grew, a water-soluble sublingual tablet form of buprenorphine was developed. However, the tablet form of buprenorphine appears to have lower bioavailability compared with the alcohol-based solution form (1708–1710), suggesting dosing parameters from clinical trials should be interpreted with caution if the solution form of buprenorphine is used.

Because buprenorphine can be abused (1711–1716), it is possible for the sublingual tablet to be dissolved and used parenterally. In France, where buprenorphine is the primary medication used for the treatment of opioid dependence, there have been reports of abuse of buprenorphine tablets (1713, 1717–1719). For this reason, a formulation that combines buprenorphine with naloxone was developed for the U.S. market. Naloxone has poor sublingual bioavailability (1361), so use of this combination tablet sublingually would produce a predominant buprenorphine effect. However, if an opioid-dependent individual dissolved and injected the combination tablet, he or she would risk precipitating a withdrawal syndrome due to naloxone's good parenteral bioavailability. Studies of the combination buprenorphine-naloxone tablet have confirmed such differential effects to be a function of the route of administration (1720–1724). Although the combination tablet is the more commonly used form of this medication, many clinical trials have used a monotherapy tablet, and the monotherapy tablet is the primary form of sublingual buprenorphine used outside the United States. Under chronic dosing conditions, there may be slightly better bioavailability for the combination versus the monotherapy tablets (1710). However, for a given dose of buprenorphine, there is considerable between-subject variability in buprenorphine blood levels.

+ a. Use of buprenorphine on a daily basis as a maintenance agent

Numerous randomized, double-blind clinical trials have studied the efficacy and safety of sublingual buprenorphine for the outpatient treatment of opioid dependence. This section reviews the three representative studies that compared buprenorphine with placebo and then provides a more limited review of the many studies that compared buprenorphine with methadone.

The first study was a 16-week multisite double-blind, randomized outpatient clinical trial using four different doses of sublingual buprenorphine solution: 1, 4, 8, and 16 mg/day (1725). The primary goal was to compare the 8- and 1-mg doses, with the 1-mg dose serving as the placebo condition. The study enrolled 736 opioid-dependent patients (about 33% female); primary outcome measures were treatment retention, opioid urinalysis results, craving, and global ratings by staff and patients. Results from the study showed significantly better outcomes for the 8- versus the 1-mg groups on all the primary outcome measures. In general, there was no clear pattern of increased side effects or adverse events for the 8-mg group, with the exception of ratings of constipation (but multiple comparisons were made). Although not the primary purpose of the study, the outcomes also showed dose-related effects for buprenorphine, where 4 mg of buprenorphine was better than 1 mg, 8 mg was better than 4 mg, and 16 mg was better than 8 mg, although these differences were not always robust in their magnitude.

The second study was a double-blind, placebo-controlled study that used a somewhat novel clinical trial design that lasted only 2 weeks (1726). In this study, the 150 male and female participants were fast-tracked into treatment and randomly assigned to 0 (N=60), 2 (N=60), or 8 (N=30) mg/day of sublingual buprenorphine solution. Subjects were informed that they could receive placebo or one of two buprenorphine doses and that after 6 days, they could request to have a blind change to one of the other two conditions. Primary outcomes were the percentage of patients in each condition who remained on their original dose and the percentage of patients in each condition who requested a dose change. Other study outcomes such as opioid urine test results and self-reports of drug use were also provided. This study found that, regardless of dose, a significantly higher percentage of patients in the two active conditions remained on their doses compared with the placebo group. Similarly, a significantly higher percentage of patients in the placebo condition requested a dose change compared with the other two groups, but, once again, there was no significant difference between the two active buprenorphine groups. In an interesting finding, male subjects in the two active buprenorphine groups had a significantly lower rate of opioid-positive urine samples, but there was no difference across the three conditions for female subjects. This study provides an alternative demonstration of the efficacy of buprenorphine compared with placebo.

The third study was an office-based protocol that compared sublingual placebo tablets to active buprenorphine/naloxone (16 mg/4 mg) and buprenorphine alone (16 mg) tablets (1727). This multicenter study enrolled 326 opioid-dependent individuals, and study participation lasted 4 weeks for each volunteer. Subjects received supervised dose administration on weekdays and take-home doses of tablets on weekends. Urine samples were collected thrice weekly. Study enrollment was discontinued early because significant differences between the two active conditions and placebo were found on an interim analysis. For example, rates of opioid-negative urine samples were significantly higher for both active treatment groups compared with the placebo group. After the 4-week period, there was a further period of open-label treatment for purposes of safety assessment; the results from this phase showed that buprenorphine was safe and well tolerated. The results from this study provide further evidence of buprenorphine's efficacy compared with placebo, especially when the tablet form is studied (rather than the solution form used in the previous two studies described).

In addition to these three studies, two others have compared buprenorphine with placebo (1394, 1728). Like the studies previously described, these reports also showed that buprenorphine maintenance is superior to placebo treatment as measured by treatment retention, opioid urine test results, and mortality.

Numerous studies have compared buprenorphine's efficacy and, to a more limited extent, its safety to methadone. These studies have generally been randomized, double-blind, clinical trials conducted at a single site with fixed doses of sublingual buprenorphine solution and oral methadone (137, 1251, 1362, 1668, 1729). Variations on this basic methodological approach include studies that have compared methadone with buprenorphine tablets rather than solution (1730–1732), have used double-blind, flexible-dosing schedules rather than fixed doses of methadone and buprenorphine (74, 1363, 1364), or have compared thrice-weekly buprenorphine to daily methadone (74). Some of these studies had relatively small groups of subjects (1364, 1730), although other features of these studies may represent important methodological advances in this line of research.

In general, these studies comparing buprenorphine with methadone demonstrated that outcomes for the two medications can be very similar. An important qualification to this conclusion is that it is not always clear that the studies used comparable doses of the two medications. For example, a study comparing a very low fixed methadone dose with an average fixed buprenorphine dose could lead to the conclusion that buprenorphine is more effective than methadone, but such a study would be a comparison of dose efficacy rather than medication efficacy. When studies have used flexible doses, compared doses of methadone and buprenorphine that are thought to be approximately equivalent, or used more than one fixed dose of each medication for comparison, they have generally found similar outcomes on measures of treatment retention and rates of opioid-positive urine samples for buprenorphine and methadone. However, it appears that based on these studies, buprenorphine's maximal therapeutic effect occurs in the range of 8–16 mg/day of sublingual tablets, which is equivalent to outcomes seen with daily doses of about 50–60 mg of methadone.

+ b. Use of buprenorphine on a less than daily basis as a maintenance agent

Buprenorphine's long duration of action suggests that it may be effective on a less than daily basis. Clinical trials of intermittent buprenorphine dosing have typically stabilized patients on a daily dose and then switched them to a less than daily schedule where buprenorphine was administered on active dosing days and placebo was given on the other days. In some studies, buprenorphine doses were not increased (1733, 1734), whereas in other studies the dose was doubled (1367, 1735) on active dosing days. Although there is some evidence that significant withdrawal can occur if the daily dose is not increased when switching to intermittent dosing (1733), increasing the dose to compensate for the 48-hour interdose period generally provides adequate effects for patients and is preferred to daily dosing (1367).

There are several studies of intervals between active doses that are 48 hours. For example, a study where triple the daily buprenorphine maintenance dose was administered to patients every 72 hours found some mild increase in opioid agonist effects in the first 24 hours and some mild withdrawal at 72 hours, but neither effect was robust or clinically significant (1368). Similar results were reported in another study testing buprenorphine's effects over a 72-hour time interval (1366). A double-blind study examining a quadruple daily maintenance dose of buprenorphine given at up to 96-hour intervals (1369) did not find excessive opioid agonist effects 24 hours after the quadrupled dose and only mild withdrawal effects at 96 hours. A follow-up open report on double, triple, and quadruple buprenorphine doses for 48-, 72-, and 96-hour intervals, respectively, found similar tolerability of the 96-hour interval as well as a preference by patients for all intermittent, rather than daily, dosing schedules of buprenorphine (1368). These results suggest dosing with buprenorphine on every fourth day is clinically possible, with no significant adverse effects from the higher doses of buprenorphine and comparable efficacy with daily dosing. However, findings from studies that have tested even longer intervals between active buprenorphine doses suggest that 96 hours may be the upper limit of the interdose interval (1736, 1737). In these studies, significant but clinically mild opioid agonist effects were noted 24 hours after the quintuple or sextuple dose, with clinically mild withdrawal effects appearing after 96 hours. Studies have not tested if higher doses of buprenorphine, such as seven times a daily dose, might be effective and safe for shorter periods such as 5 days.

One of the largest controlled clinical trials of intermittent buprenorphine dosing was a double-blind, randomized study comparing thrice-weekly to daily buprenorphine dosing (1738). This study assigned 92 opioid-dependent outpatients to receive sublingual buprenorphine solution daily or sublingual buprenorphine solution on Tuesdays, Fridays, and Sundays and placebo on the other four days. The daily buprenorphine dose was 16 mg/70 kg, whereas buprenorphine dosing in the intermittent condition was 34 mg/70 kg on Friday and Sunday and 44 mg/70 kg on Tuesday. The 72-hour dosing interval was from Tuesday to Friday to avoid potential study confounders associated with the weekend. There were no differences in rates of opioid-positive urine tests, self-reports of illicit opioid use, treatment retention, and medication adherence between the two groups and no significant adverse events noted for either group. The results from this large controlled study suggest that intermittent buprenorphine can be as equally effective as daily buprenorphine.

Although the studies reviewed here have all used the solution form of buprenorphine, there have also been studies of intermittent dosing with buprenorphine tablets. One double-blind outpatient clinical trial compared daily dosing of 8-mg buprenorphine tablets to thrice-weekly dosing with 16-mg tablets on Mondays and Wednesdays, 24-mg tablets on Fridays, and placebo tablets on the four nonactive dosing days. The study found equivalent retention for the two groups but a higher rate of opioid-positive urine samples in the intermittent dosing group (1739). Using the buprenorphine/naloxone combination tablets, Another study showed that patients prefer thrice-weekly take-home doses of the medication to both daily and thrice-weekly medication visits at a clinic (1740). These two studies provide evidence of the practicality of using buprenorphine tablets on an intermittent basis.

+ c. Use of buprenorphine as a withdrawal (detoxification) agent

Controlled clinical trials and reports of buprenorphine for the treatment of opioid withdrawal have varied in several ways. Some withdrawal studies have compared buprenorphine with methadone (1741), whereas others have compared it with clonidine (1393, 1742, 1743). Different formulations of buprenorphine (1384, 1744) and rapid or prolonged withdrawals (1392, 1745, 1746) have also been used in these studies. There has been some interest in and research on using buprenorphine as a bridging agent to treatment with naltrexone (1745, 1747). Several reports of buprenorphine's use in opioid withdrawal are open studies describing clinicians' experience with buprenorphine (1384, 1748–1752). Although the outcomes noted in these reports are confounded by the lack of important features found in appropriately conducted clinical trials, they do provide important clinical evidence of buprenorphine's acceptability as a withdrawal medication.

An early double-blind, double-dummy, random-assignment, outpatient clinical trial compared fixed doses of sublingual buprenorphine solution with oral methadone in the treatment of opioid withdrawal (1741). A total of 45 heroin-dependent male subjects were stabilized on 30 mg of methadone or 2 mg of buprenorphine for 3 weeks, underwent a dose taper for 4 weeks and then received placebo for 6 weeks. A novel feature to this study's design was that subjects underwent a 6-mg hydromorphone challenge session during the second week of stabilization (to test the relative blockade efficacy of the two treatment medications). Results from the withdrawal assessments showed the two medications were quite similar on measures of treatment retention, drug use, and rating of withdrawal. However, results from the hydromorphone challenge sessions showed that methadone produced significantly greater blockade than buprenorphine. The results from this early buprenorphine withdrawal study demonstrate that similar outcomes can be achieved with these two medications, although the study used relatively low doses of both medications and the overall results showed poor outcomes for both groups.

An inpatient, randomized, double-blind clinical trial compared sublingual buprenorphine solution to oral clonidine in a relatively short opioid withdrawal procedure (1742). In this study of 25 opioid-dependent male and female patients, participants received 3 days of a fixed- dose schedule of buprenorphine or a 5-day fixed-dose schedule of clonidine. Because this was an inpatient study, reports of withdrawal symptoms were not confounded by illicit substance use. Overall, patients treated with buprenorphine were found to have less opioid withdrawal than those treated with clonidine, whereas there was more hypotension in patients treated with clonidine. This study demonstrated that sublingual buprenorphine was more effective than clonidine in the inpatient treatment of opioid withdrawal; other studies and reports comparing buprenorphine with clonidine have shown similar results (1384, 1391–1393).

A large double-blind, randomized, outpatient clinical trial compared withdrawal using buprenorphine with clonidine and clonidine plus naltrexone in an outpatient primary care clinic setting (1747). This study randomly assigned 162 opioid-dependent male and female patients to one of three conditions: sublingual buprenorphine for 3 days, followed by clonidine and naltrexone; 7 days of clonidine; or 7 days of clonidine plus naltrexone. Results from the study showed that treatment retention was not significantly different for the three groups. However, there were significantly less opioid withdrawal symptoms (both overall withdrawal and peak effects) for buprenorphine-treated patients compared with the other two groups. These study results give further evidence of the clinical value of buprenorphine compared with clonidine provided on an outpatient basis.

+ d. Safety and side effects of buprenorphine

Buprenorphine has been extensively tested in a variety of outpatient clinical trials, with no reports of significant adverse events from these studies. In addition, it has been used extensively in other countries, especially France, where it is estimated that there are over 70,000 buprenorphine-treated patients.

There is some evidence that buprenorphine may have effects on hepatic function. One report, based on a retrospective review of 120 patients treated with sublingual buprenorphine, suggested that buprenorphine is associated with elevated results on liver function tests for some patients with a history of hepatitis (1753). Although these elevations were relatively mild, there is also evidence that intravenous use of buprenorphine can produce marked increases in liver function test values (1754, 1755). One inpatient study of buprenorphine also found mild increases in liver transaminases over time, although the lack of a control group, the nonspecificity of these laboratory results, and the relatively mild effects seen make interpretation of such findings difficult (1756).

Systematic evaluations of buprenorphine's other side effects have been relatively scarce. One report by Lange et al. (1756) on safety and side effect measures collected during a 12-week inpatient study of buprenorphine induction as well as a study on daily versus alternate-day dosing (1733, 1757) found that no symptoms were rated as definitely related to buprenorphine; the only two symptoms rated as probably related to buprenorphine were drowsiness/sedation and, most frequently, constipation. Nausea/vomiting and headache, which occurred in the first 1–2 weeks of treatment, were rated as possibly related to buprenorphine. A second report that compared safety and side effect measures from a clinical trial comparing daily buprenorphine solution to daily oral methadone found that there were few significant differences in side effect reports from the two medications (1758). However, as was found in the Lange et al. (1756) study, ratings of constipation were higher for the buprenorphine versus the methadone group, although they were relatively low for both groups.

Available evidence varies with respect to mortality and serious adverse events such as respiratory depression associated with buprenorphine. Because of buprenorphine's unusual pharmacological profile, which includes a bell-shaped dose-response curve observed in preclinical studies (1759–1761), it has been thought that overdose with buprenorphine would have a low likelihood of producing significant respiratory depression. Indeed, there have been at least two case reports of buprenorphine overdose in which patients did not experience respiratory depression (1762, 1763). However, there have also been case reports from France of deaths associated with buprenorphine use, typically when buprenorphine was injected in combination with a benzodiazepine, most typically flunitrazepam, which is not available by prescription in the United States (1764, 1765). There is also preclinical evidence that the combination of buprenorphine and a benzodiazepine can cause respiratory depression (1766) and evidence that suggests the interaction of buprenorphine with flunitrazepam is due to pharmacodynamic rather than pharmacokinetic effects (1767). Finally, benzodiazepine abuse is not uncommon in opioid-dependent patients in the United States; in one outpatient clinical trial of buprenorphine-treated patients, 6.2% of the patients' urine samples were positive for benzodiazepines (1363). However, buprenorphine-related deaths in U.S. clinical trials were not found, suggesting that use of buprenorphine as prescribed, even when combined with illicit benzodiazepine use, is typically not associated with fatalities. However, more liberal availability of buprenorphine in the United States could lead to deaths as has been seen in France.

b) Opioid antagonist therapy: naltrexone

Naltrexone is the only opioid antagonist medication currently approved in the United States for the maintenance treatment of opioid dependence.

+ (1) Efficacy of naltrexone as a maintenance agent

Naltrexone's efficacy in the treatment of opioid dependence has been studied under two general experimental designs: the inpatient human laboratory and outpatient clinical trials.

+ a. Naltrexone's efficacy as demonstrated in human laboratory studies

Studies of naltrexone in the human laboratory setting have typically withdrawn opioid-dependent volunteers from opioids and then tested the acute effects of opioids while maintaining the subjects on naltrexone. Under such conditions, naltrexone has been shown to be effective in blocking the effects of acute opioid use (145, 1768–1772).

In one study, for example, 12 heroin-dependent inpatients were withdrawn from opiates and then maintained on 50 mg/day of naltrexone (N=3) or daily placebo (N=9) (1372). The subjects were then allowed to self-administer four doses of heroin each day over the next 10 days. At the end of the study, the placebo group self-administered significantly more doses of heroin than did the naltrexone group. The naltrexone group self-administered heroin six times over the 10 days, whereas every placebo-maintained subject took heroin at least twice per day and three of the nine took all available heroin over the 10-day period. Results such as these demonstrate that naltrexone can be highly effective in blocking the effects of short-acting opioids such as heroin.

+ b. Naltrexone's efficacy as demonstrated in clinical trials

There have been numerous reports on clinician experience with the use of naltrexone in the outpatient treatment of opioid dependence (1773–1779). Large, double-blind, placebo-controlled studies of naltrexone are more uncommon, partly because maintaining the blind in an outpatient study of naltrexone is virtually impossible. Participants can easily guess their condition assignment if they use opiates and feel or do not feel an effect.

One of the earliest and still one of the largest double-blind studies comparing naltrexone with placebo for the treatment of opioid dependence illustrates the difficulties in such clinical trials (1375). This multisite study enrolled 735 patients who were recently withdrawn from illicit opioids, maintained on but withdrawn from methadone prior to receiving the study medication, or withdrawn from opioids and participating in drug-free treatment. The attrition rate for this study was high, with 543 (74%) participants dropping out before receiving any study medication. Of the 192 who received medication, 170 (89%) dropped out before completing the 9-month study, with >50% dropping out in the first 2 months of the study. For participants who submitted at least five urine samples, there was no significant difference in rates of opioid-positive urine samples. However, an analysis of a subgroup of patients who had at least one opioid-positive urine sample, implying they would have experienced naltrexone's beneficial effect or placebo's lack of effect, found a significantly lower rate of opioid-positive urine samples in the naltrexone-treated (N=17) versus the placebo-treated (N=18) group. This study, which sought to test the overall acceptability and efficacy of naltrexone treatment, showed there could be some possible efficacy in a small group of patients and concluded that this group is most likely to be those who are already in drug-free counseling and who have a high level of motivation.

Voucher incentives have been shown to improve patient adherence with naltrexone treatment (1406, 1407), and improved adherence implies that there should be less illicit opioid use. In the study by Preston et al. (1406), for example, 58 naltrexone-treated patients were randomly assigned to a contingent group where voucher incentives were given for each naltrexone dose taken (N=19), a noncontingent group where voucher incentives were given unsystematically and independent of medication adherence (N=19), or a no-voucher group (N=20). Comparing illicit opioid use among the three groups was extremely difficult because the group receiving vouchers contingent on adherence had substantially greater rates of treatment completion and mean time remaining in treatment. Nevertheless, there was a tendency for the contingent group to have the lowest rate of opioid-positive urine samples. The study by Carroll et al. (1407) also found voucher incentives enhanced naltrexone adherence, which in turn was associated with less illicit opioid use.

The legal system can also provide an external motivation for adherence with naltrexone treatment. A random assignment study of 51 individuals on federal probation compared probation plus naltrexone and counseling with probation and counseling alone (1375). At the end of the 6-month study, no significant differences in treatment retention were found between the two groups, but the group that also received naltrexone had significantly lower rates of opioid-positive urine samples and reincarceration. Other studies of individuals with high levels of motivation to remain abstinent, such as business executives and physicians (1374), provide further evidence that this medication can be useful under the proper clinical circumstances.

+ (2) Safety and side effects of naltrexone

In general, naltrexone is a safe medication with few side effects. There has been concern that naltrexone can produce elevations in liver function tests, but these effects were observed in only four studies where patients were treated for atypical indications (e.g., obesity, dementia) with naltrexone doses higher than those used for the treatment of opioid dependence (e.g., up to 300 mg/day). These elevations in liver function tests occurred only with the 300 mg/day dose of naltrexone, were seen in only 1 out of 40 (2.5%) subjects under age 40 years compared with 18.5% of individuals older than age 40, and resolved with naltrexone discontinuation (1374). However, there is one letter reporting that the risk of increased liver function tests may occur if naltrexone is taken with NSAIDs (1047). Although the approved label for naltrexone cautions about its potential effects on liver function tests, the likelihood that these hepatic effects will occur in patients treated within the usual dose range (50–150 mg/day) and without preexisting liver disease is minimal.

c) Clonidine and other nonopioid medications used for the treatment of opioid withdrawal

Clonidine is an 2-adrenergic agonist approved in the United States for the treatment of hypertension but not for the treatment of opioid withdrawal. Clonidine has been used as an aid in controlling opioid withdrawal symptoms; unlike methadone, LAAM, buprenorphine, and naltrexone, it is not used as a maintenance treatment agent.

A related medication is lofexidine, another 2-adrenergic agonist approved for the treatment of opioid withdrawal in the United Kingdom but not currently available in the United States. There is evidence that lofexidine can be as effective as clonidine in the treatment of opioid withdrawal and that it produces less hypotension than does clonidine (1780–1782). However, at least one controlled inpatient study did not show that lofexidine markedly attenuated opioid withdrawal symptoms that were produced by a naloxone challenge (1783), and another study found buprenorphine was more effective than lofexidine when used in the treatment of opioid withdrawal (1784). The remainder of this section focuses on clonidine; no further review of lofexidine is provided here.

+ (1) Use of clonidine as a withdrawal (detoxification) agent

Several early studies tested the efficacy of clonidine when used on an inpatient basis for the treatment of opioid withdrawal (1388, 1785–1787). For example, a study that sought to determine optimal clonidine dosing parameters used a 2-week inpatient design and enrolled 25 methadone-maintained male and female patients (1380). Subjects had a double-blind abrupt discontinuation of their methadone dose and then started receiving thrice-daily clonidine 24 or 48 hours later, with flexible dosing based on signs and symptoms of withdrawal. Clonidine was found to reduce opioid withdrawal symptoms (compared with patient perception of previous withdrawal attempts), although symptoms of anxiety, restlessness, insomnia, and muscle aches were not relieved completely.

Another inpatient study used a classic substitution design to test the efficacy of acute doses of clonidine to suppress spontaneous opioid withdrawal symptoms in 10 male patients maintained on daily subcutaneous doses of morphine (1383). In this study, clonidine was found to be more effective than morphine in suppressing opioid withdrawal signs, although it was less effective than morphine in attenuating subjective complaints of opioid withdrawal.

Finally, a double-blind study from Spain compared the inpatient use of clonidine to a relatively low dose of methadone for short-term (8- to 10-day) withdrawal from heroin (1382). There were initially 45 subjects in the study; most of those who left the study early were taking clonidine. For those who remained (N=30), clonidine and methadone were found to be equally effective on measures of withdrawal. In an interesting finding, it appears that there was actually less residual opioid withdrawal at the end of this inpatient study for the clonidine patients compared with the methadone patients. However, the relatively short period of treatment and the low dose of methadone may have contributed to the finding of comparable efficacy for the two medications among patients who remained in the study.

Clonidine has also been examined for its efficacy in the outpatient treatment of opioid withdrawal (1384–1386). For example, a double-blind outpatient study with 49 opioid-dependent male and female participants compared methadone and clonidine in the treatment of withdrawal (1381). Participants in the study had relatively low levels of physical dependence (20 mg of daily methadone). In the study, participants underwent a double-blind 20-day methadone withdrawal or a 15-day clonidine withdrawal. There was no difference in the two groups among those successfully completing the withdrawal, although the clonidine-treated subjects tended to have greater withdrawal symptoms and to drop out earlier compared with the methadone-treated patients.

+ (2) Safety and side effects of clonidine

The most significant side effect associated with the use of clonidine in the treatment of opioid withdrawal is hypotension; the lowering of blood pressure by clonidine can be clinically significant for some opioid-dependent patients. This effect has been noted in many of the clinical studies of clonidine (1380, 1382, 1383, 1392, 1742, 1787) and has led to the common recommendation that when clonidine is used for opioid withdrawal, the treating physician regularly check the patient's blood pressure and hold the dose if hypotensive effects are noted.

Other side effects noted with clonidine when used for the treatment of opioid withdrawal have included sedation and other sleep difficulties, dry mouth, and constipation. In some cases, it can be difficult to know if these more uncommon symptoms are side effects of clonidine or opioid withdrawal effects that are not fully treated by clonidine.

2. Psychosocial treatments

As noted previously, psychosocial treatments for opioid-related disorders have been studied only in programs that also provide maintenance treatment with either opioid agonists (e.g., methadone) or antagonists.

a) Cognitive-behavioral therapies

Woody et al. (177, 531, 1400, 1401) randomly assigned methadone maintenance patients to one of three groups: 1) drug counseling alone, 2) drug counseling plus supportive-expressive psychotherapy, or 3) drug counseling plus CBT. Outcomes were determined at 7 and 12 months. In patients with a moderate to high degree of depression or other psychiatric symptoms, drug counseling with supportive-expressive therapy or CBT was much more effective than drug counseling alone; for patients with low levels of psychiatric symptoms, all three treatment conditions were equally effective. These findings were essentially replicated in three community-based methadone maintenance clinics (218). O'Neill et al. (1402) evaluated a CBT intervention added to a methadone program for pregnant, injecting drug users at risk for HIV infection. CBT resulted in significantly greater reductions in HIV risk behaviors.

+ (1) Relapse prevention strategies

McAuliffe (1403) reported that when combined with participation in a self-help group, group relapse prevention based on a conditioning model of addiction was more effective than no treatment in reducing opioid use, unemployment, and criminal activities in recently detoxified patients.

b) Behavioral therapies
+ (1) Contingency management

In a study evaluating the intensity of behavioral treatments for methadone patients, McLellan et al. (170) determined whether the addition of contingency-based counseling, general medical care, and psychosocial services improved the efficacy of methadone maintenance treatment in a study of newly admitted opioid patients randomly assigned to three levels of care. Patients who received counseling and contingencies based on urine test results, in addition to methadone, had better drug use outcomes than those who received methadone only. Patients who in addition received on-site general medical and psychiatric care, employment services, and family therapy had the best outcomes of all three conditions. Methadone alone was an effective treatment for only a small percentage of patients.

Several studies have evaluated the use of contingency management in reducing the use of illicit drugs in opioid-dependent individuals who are maintained on methadone. In these studies, a reinforcer (reward) is provided to patients who demonstrate specified target behaviors such as providing drug-free urine specimens, accomplishing specific treatment goals, or attending treatment sessions. For example, offering methadone take-home privileges contingent on reduced drug use is an approach that capitalizes on an inexpensive reinforcer that is potentially available in all methadone maintenance programs. Stitzer and colleagues (197–199) have done extensive work in evaluating methadone take-home privileges as a reward for decreased illicit drug use. In a series of well-controlled trials, these researchers have demonstrated 1) the relative benefits of positive (e.g., rewarding desired behaviors such as abstinence) compared with negative (e.g., punishing undesired behaviors such as continued drug use through discharges or dose reductions) contingencies (197), 2) the attractiveness of take-home privileges over other incentives available within methadone maintenance clinics (198), and 3) the relative effectiveness of rewarding drug-free urine specimens compared with other target behaviors (202).

Silverman and colleagues (195, 1295), drawing on the compelling work of Higgins and colleagues (described below), evaluated in a series of studies the efficacy of a voucher-based contingency management system to address concurrent illicit drug use (typically cocaine) among methadone-maintained opioid-dependent individuals. In this approach, urine specimens were required three times a week to systematically detect all episodes of drug use. Abstinence, verified through urine screens, was reinforced through a voucher system in which patients received points redeemable for items consistent with a drug-free lifestyle that were intended to help the patient develop alternate reinforcers to drug use (e.g., movie tickets, sporting goods). Silverman and colleagues (195, 1295) demonstrated the efficacy of this approach in reducing illicit opioid and cocaine use.

Opioid antagonist treatment (naltrexone) offers many advantages over methadone maintenance, including the fact that it is nonaddicting and can be prescribed without concerns about diversion, has a benign side effect profile, and can be less costly in terms of demands on professional time and patient time than the daily or near-daily clinic visits required for methadone maintenance (165). Most important are the behavioral aspects of treatment, as unreinforced opiate use allows the extinction of the association between cues and drug use. Although naltrexone treatment is likely to be attractive only to a small number of opioid-dependent individuals (166), naltrexone's unique properties make it an important alternative to methadone maintenance and other agonist approaches.

However, despite its many advantages, naltrexone has not fulfilled its promise. Naltrexone treatment programs remain comparatively rare and underutilized as compared with methadone maintenance programs (165), largely because of problems with retention, particularly during the induction phase; an average of 40% of patients drop out during the first month of treatment and 60% drop out by 3 months (166). In the 1970s, several preliminary evaluations identified the promise of using behavioral interventions to address naltrexone's weaknesses, including providing incentives for adherence with naltrexone treatment (1404, 1405) and the addition of family therapy to naltrexone treatment (1788). However, the interventions were not widely adopted, adherence remained a major problem, and naltrexone treatment and research dropped off considerably until the past few years, when the need for alternatives to methadone maintenance stimulated a modest revival of interest in naltrexone.

Some of the most recent promising data about strategies to enhance retention and outcome in naltrexone treatment have come from investigations of contingency management approaches. Preston et al. (1406) found improved retention and naltrexone treatment adherence with an approach that provided vouchers for adherence as compared with standard naltrexone treatment that did not provide vouchers. Carroll and colleagues (167, 1407) found that reinforcement of naltrexone treatment adherence and drug-free urine specimens, alone or in combination with family involvement in treatment, improved retention rates and reduced drug use among recently detoxified opioid-dependent individuals.

+ (2) Cue exposure treatment

Cue exposure treatment has been demonstrated to be effective in reducing classically conditioned responses to substance-related cues in a small group of patients with opioid use disorders (189).

c) Psychodynamic and interpersonal therapies

The effectiveness of adding a psychodynamic therapy to a program of methadone maintenance has been investigated. Woody et al. (177, 218) found that supportive-expressive therapy was more effective than drug counseling alone for patients with high levels of other psychiatric symptoms. Rounsaville et al. (1789) attempted to compare the efficacy of a 6-month course of weekly individual IPT with a low-contact comparison condition for individuals in a full-service methadone maintenance program that included weekly group psychotherapy. Patients with opioid dependence who met the inclusion criteria (including the presence of an additional nonpsychotic psychiatric diagnosis) were randomly assigned to the two groups. However, only 5% of the eligible patients agreed to participate (compared with 60% in the Woody et al. study), and only about 50% completed the trial. The highly selective nature of the participants (i.e., 95% of eligible patients refused), the high attrition rate, and the lack of significant outcome differences between the two groups led to the conclusions that it is very difficult to engage opioid-dependent patients in individual IPT and that the potential benefit of such treatment is unclear for those who do participate.

Psychodynamically oriented group therapy modified for substance-dependent patients appears to be effective in promoting abstinence when combined with behavioral monitoring and individual supportive psychotherapy (1301).

Although some follow-up studies of naturalistic treatment have found equivalent efficacy for methadone maintenance and outpatient drug-free programs for heroin users (61, 1396–1398), early attempts at providing psychotherapy alone have yielded unacceptably high attrition rates (1399).


Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).
Related Content
Manual of Clinical Psychopharmacology, 7th Edition > Chapter 3.  >
Manual of Clinical Psychopharmacology, 7th Edition > Chapter 4.  >
The American Psychiatric Publishing Textbook of Psychiatry, 5th Edition > Chapter 10.  >
The American Psychiatric Publishing Textbook of Psychiatry, 5th Edition > Chapter 26.  >
The American Psychiatric Publishing Textbook of Psychiatry, 5th Edition > Chapter 31.  >
Psychiatric News
PubMed Articles
  • Print
  • PDF
  • E-mail
  • Chapter Alerts
  • Get Citation