Anticonvulsants for alcohol withdrawal

Abstract

Background

Alcohol abuse and dependence represents a most serious health problem worldwide with major social, interpersonal and legal interpolations. Besides benzodiazepines, anticonvulsants are often used for the treatment of alcohol withdrawal symptoms. Anticonvulsants drugs are indicated for the treatment of alcohol withdrawal syndrome, alone or in combination with benzodiazepine treatments. In spite of the wide use, the exact role of the anticonvulsants for the treatment of alcohol withdrawal has not yet bee adequately assessed.

Objectives

To evaluate the effectiveness and safety of anticonvulsants in the treatment of alcohol withdrawal.

Search methods

We searched Cochrane Drugs and Alcohol Group' Register of Trials (December 2009), PubMed,  EMBASE,  CINAHL (1966 to December 2009), EconLIT (1969 to December 2009). Parallel searches on web sites of health technology assessment and related agencies, and their databases.

Selection criteria

Randomized controlled trials (RCTs) examining the effectiveness, safety and overall risk‐benefit of anticonvulsants in comparison with a placebo or other pharmacological treatment. All patients were included regardless of age, gender, nationality, and outpatient or inpatient therapy.

Data collection and analysis

Two authors independently screened and extracted data from studies.

Main results

Fifty‐six studies, with a total of 4076 participants, met the inclusion criteria. Comparing anticonvulsants with placebo, no statistically significant differences for the six outcomes considered.

Comparing anticonvulsant versus other drug, 19 outcomes considered, results favour anticonvulsants only in the comparison carbamazepine versus benzodiazepine (oxazepam and lorazepam) for  alcohol withdrawal symptoms (CIWA‐Ar score): 3 studies, 262 participants, MD ‐1.04 (‐1.89 to ‐0.20),  none of the other comparisons reached statistical significance.

Comparing different anticonvulsants no statistically significant differences in the two outcomes considered.

Comparing anticonvulsants plus other drugs versus other drugs (3 outcomes considered), results from one study, 72 participants, favour paraldehyde plus chloral hydrate versus chlordiazepoxide, for the severe‐life threatening side effects, RR 0.12 (0.03 to 0.44).

Authors' conclusions

Results of this review do not provide sufficient evidence in favour of anticonvulsants for the treatment of AWS. There are some suggestions that carbamazepine may actually be more effective in treating some aspects of alcohol withdrawal when compared to benzodiazepines, the current first‐line regimen for alcohol withdrawal syndrome. Anticonvulsants seem to have limited side effects, although adverse effects are not rigorously reported in the analysed trials.

Plain language summary

Anticonvulsants for alcohol withdrawal syndrome

There are limited data on anticonvulsants versus placebo for alcohol withdrawal syndrome, while comparisons with other drugs show no clear differences.

This Cochrane review summarizes evidence from forty‐eight randomised controlled trials evaluating the effectiveness and safety of anticonvulsants in the treatment of alcohol withdrawal symptoms. There are limited data comparing anticonvulsants versus placebo and no clear differences between anticonvulsants and other drugs in the rates of therapeutic success. Data on safety outcomes are sparse and fragmented. There is a need for larger, well‐designed studies in this field.

Summary of findings

Summary of findings for the main comparison. Anticonvulsant versus Placebo for alcohol withdrawal.

Anticonvulsant versus Placebo for alcohol withdrawal
Patient or population: patients with alcohol withdrawal Settings: Intervention: Anticonvulsant versus Placebo
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	Anticonvulsant versus Placebo
Alcohol Withdrawal Seizures post treatment ‐ Any Anticonvulsant vs. Placebo	Study population		RR 0.61 (0.31 to 1.2)	883 (9 studies)	⊕⊕⊕⊝ moderate1
	110 per 1000	67 per 1000 (34 to 132)
	Medium risk population
	167 per 1000	102 per 1000 (52 to 200)
Alcohol Withdrawal Seizures post treatment ‐ Phenytoin vs. Placebo	Study population		RR 0.78 (0.35 to 1.77)	381 (4 studies)	⊕⊕⊕⊝ moderate2
	180 per 1000	140 per 1000 (63 to 319)
	Medium risk population
	176 per 1000	137 per 1000 (62 to 312)
Adverse events	Study population		RR 1.56 (0.74 to 3.31)	663 (7 studies)	⊕⊕⊕⊝ moderate3
	50 per 1000	78 per 1000 (37 to 165)
	Medium risk population
	34 per 1000	53 per 1000 (25 to 113)
Dropout ‐ Any Anticonvulsant vs. Placebo	Study population		RR 0.82 (0.5 to 1.34)	344 (7 studies)	⊕⊕⊕⊝ moderate4
	208 per 1000	171 per 1000 (104 to 279)
	Medium risk population
	182 per 1000	149 per 1000 (91 to 244)
Dropout ‐ Chlormethiazole vs. Placebo	Study population		RR 1.05 (0.22 to 5.11)	140 (3 studies)	⊕⊕⊕⊝ moderate5
	44 per 1000	46 per 1000 (10 to 225)
	Medium risk population
	21 per 1000	22 per 1000 (5 to 107)
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RR: Risk ratio;
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.

¹ Allocation concealmnt adequate only in 1 on 9 included studies 
 ² Allcation concealment adequate in 1/4 studies 
 ³ In 3 studies results were not estimable due no adverse events occurred 
 ⁴ Allocation concealment adequate in 2 on 7 included studies 
 ⁵ only 3 studies considered

Summary of findings 2. Anticonvulsant versus Other Drugs for alcohol withdrawal.

Anticonvulsant versus Other Drugs for alcohol withdrawal
Patient or population: patients with alcohol withdrawal Settings: Intervention: Anticonvulsant versus Other Drugs
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	Anticonvulsant versus Other Drugs
Alcohol withdrawal seizures ‐ Any Anticonvulsant vs any Other	Study population		RR 0.58 (0.22 to 1.58)	880 (12 studies)	⊕⊕⊝⊝ low1,2
	27 per 1000	16 per 1000 (6 to 43)
	Medium risk population
	19 per 1000	11 per 1000 (4 to 30)
Adverse events ‐ Any Anticonvulsant vs any Other	Study population		RR 0.71 (0.45 to 1.12)	726 (14 studies)	⊕⊕⊕⊝ moderate3
	287 per 1000	204 per 1000 (129 to 321)
	Medium risk population
	188 per 1000	133 per 1000 (85 to 211)
Dropout ‐ Any Anticonvulsant vs any Other	Study population		RR 0.92 (0.67 to 1.26)	1359 (20 studies)	⊕⊕⊕⊝ moderate4
	114 per 1000	105 per 1000 (76 to 144)
	Medium risk population
	99 per 1000	91 per 1000 (66 to 125)
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RR: Risk ratio;
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.

¹ Allocation concealment adequate in 2 on 12 included studies 
 ² In 5 studies no events and results not estimable 
 ³ Allocation concealment adeqate in 4/14 included studies 
 ⁴ Allocation concealment adequate in 5 on 20 included studies

Background

Description of the condition

Alcohol abuse and dependence represents a most serious health problem worldwide with major social, interpersonal and legal interpolations. Dependence on alcohol is associated with both physiological symptoms such as tolerance and withdrawal, and behavioural symptoms such as impaired control over drinking (Hasin 1990). Alcohol withdrawal syndrome is a cluster of symptoms that occurs in alcohol‐dependent people after cessation or reduction in alcohol use that has been heavy or prolonged. The clinical presentation varies from mild to serious and the onset of symptoms typically occurs a few hours after the last alcohol intake. The most common manifestations are tremor, restlessness, insomnia, nightmares, paroxysmal sweats, tachycardia, fever, nausea, vomiting, seizures, hallucinations (auditory, visual, tactile), increased agitation, tremulousness and delirium. These symptoms involve a wide range of neurotransmitter circuits that are implicated in alcohol tolerance and reflect a homeostatic readjustment of the central nervous system (De Witte 2003; Koob 1997; Nutt 1999; Slawecki 1999). Long‐term alcohol consumption affects brain receptors that undergo adaptive changes in an attempt to maintain normal function. Some of the key changes involve reduced brain gamma‐amino butyric acid (GABA) levels and GABA‐ receptor sensitivity (Dodd 2000; Gilman 1996; Kohl 1998; Petty 1993) and activation of glutamate systems (Tsai 1995), which lead to nervous system hyperactivity in the absence of alcohol. The advances in knowledge of neurobiology and neurochemistry have prompted the use of drugs in the treatment of alcohol dependence and withdrawal that act through these GABA pathways.

Description of the intervention

Besides benzodiazepines, anticonvulsants are often used for the treatment of alcohol withdrawal symptoms. A meta‐analysis of studies concerning pharmacological therapies of alcohol withdrawal (Mayo‐Smith 1997) has suggested that carbamazepine, an anticonvulsant agent that is widely used in particular in Europe, may have modest beneficial effects on selected signs and symptoms of withdrawal. Carbamazepine may also be considered as adjunctive therapy to benzodiazepines, the classic treatment for alcohol withdrawal.

How the intervention might work

Anticonvulsants drugs are indicated for the treatment of alcohol withdrawal syndrome, alone or in combination with benzodiazepine treatments. In spite of the wide use, the exact role of the anticonvulsants for the treatment of alcohol withdrawal has not yet been adequately assessed. Moreover not all patients may need pharmacological treatment and it is unknown whether different anticonvulsants and different regimens of administration (e.g. symptom‐triggered versus fixed schedule) may have the same merits.

Why it is important to do this review

The need to research non‐benzodiazepines effective in Alcohol Withdrawal is related to the risks of side‐effects and the potential of abuse of benzodiazepine (Leggio 2008).

Since there are several anticonvulsant agents and a large amount of evidence of their use in the management of alcohol withdrawal has been published during the last years, an updated systematic is needed in order to clarify the exact role of anticonvulsants in the management of alcohol withdrawal.

The purpose of this systematic review is to examine the evidence in the effectiveness and safety of anticonvulsants in the treatment of alcohol withdrawal symptoms. Results of a previous version of a Cochrane Systematic review (Polycarpou 2005) on anticonvulsants efficacy and safety are not conclusive. New trials have been published and the review needs update.

This review has a parallel one on benzodiazepines for alcohol withdrawal (Amato 2010) and together they are part of a series of reviews and protocols on the efficacy of pharmacological treatment (Acamprosate GHB, nitrous oxide, magnesium) for alcohol withdrawal (Gillman 2007; Leone 2010; Fox 2003; Tejani 2010).

Objectives

The objectives of this systematic review are:

1. To evaluate the effectiveness of anticonvulsants in the treatment of alcohol withdrawal.

2. To evaluate the safety of anticonvulsants in the treatment of the alcohol withdrawal symptoms (AWS).

Methods

Criteria for considering studies for this review

Types of studies

Randomized Controlled Trials (RCT) and Controlled Clinical Trials (CCT) evaluating the efficacy, safety and overall risk‐benefit of anticonvulsants for the treatment of alcohol withdrawal.

Types of participants

Alcohol dependent patients diagnosed in accordance with appropriate standardized criteria (e.g., criteria of Diagnostic and Statistical Manual of Mental Disorders (DSM‐IV‐R) or ICD) who experienced alcohol withdrawal symptoms regardless of the severity of the withdrawal manifestations.  All patients were included regardless of age, gender, nationality, and outpatient or inpatient therapy. The history of previous treatments was considered, but it was not an eligibility criterion.

Types of interventions

‐ Experimental intervention

Anticonvulsants drugs alone or in combination with other drugs

‐ Control Intervention

Placebo; Other pharmacological interventions

Types of comparisons

1. anticonvulsant versus placebo;

2. anticonvulsant versus other drugs;

3. different anticonvulsants between themselves;

4. anticonvulsant combined with other drugs versus other drugs.

5. anticonvulsant 1 combined with other drugs versus anticonvulsant 2

Types of outcome measures

Primary outcomes

Efficacy outcomes

1. Alcohol withdrawal seizures as number of subjects experiencing seizures

2. Alcohol withdrawal delirium as number of subjects experiencing delirium

3. Alcohol withdrawal symptoms as measured by prespecified scales(as the CIWA‐Ar score)

4. Global improvement of overall alcohol withdrawal syndrome as measured in pre‐specified scales ( as number of patients with global improvement, global doctors assessment of efficacy, Patients assessment of efficacy)

5. Craving as measured by prespecified scales

Safety outcomes

1. Adverse events as number of subjects experiencing at least one adverse event

2. Severe, life‐threatening adverse events as measured by number of subjects experiencing severe, life threatening adverse events

Acceptability outcomes

1. Dropout

2. Dropout due to adverse events

Secondary outcomes

1. Additional medication needed

2. Length of stay in intensive therapy

3. Mortality

4. Quality of life

Search methods for identification of studies

Electronic searches

Relevant trials were obtained from the following sources:

1. Cochrane Drugs and Alcohol Group' Register of Trials (December 2009)

2. PubMed (January 1966– December 2009)

3. EMBASE (January 1988– December 2009)

4. CINAHL (January 1982– December 2009)

5. EconLIT (1969 to December 2009)

We compiled detailed search strategies for each database searched, for detail see Appendix 1; Appendix 2; Appendix 3; Appendix 4

Searching other resources

We also searched:

1. the reference lists of all relevant papers to identify further studies.

2. conference proceedings likely to contain trials relevant to the review.

We contacted investigators seeking information about unpublished or incomplete trials.All searches included non‐English language literature and studies with English abstracts were assessed for inclusion. When considered likely to meet inclusion criteria, studies were translated.

Data collection and analysis

Selection of studies

Two authors independently screened the titles and abstracts of all publications, obtained through the search strategy. All potentially eligible studies were obtained as full articles and two authors independently assessed these for inclusion. In doubtful or controversial cases, all identified discrepancies were discussed and reached consensus on all items.

Data extraction and management

Two authors independently extracted data from published sources, where differences in data extracted occurred this was resolved through discussion. Where required additional information was obtained through collaboration with the original authors.

Assessment of risk of bias in included studies

The risk of bias assessment  for RCTs and CCTs in this review was performed using four out of the six criteria recommended by the Cochrane Handbbok (Higgins 2008). The recommended approach for assessing risk of bias in studies included in Cochrane Review is a two‐part tool, addressing four specific domains (namely sequence generation, allocation concealment, blinding, incomplete outcome data). The first part of the tool involves describing what was reported to have happened in the study. The second part of the tool involves assigning a judgement relating to the risk of bias for that entry. This is achieved by answering a pre‐specified question about the adequacy of the study in relation to the entry, such that a judgement of "Yes" indicates low risk of bias, "No" indicates high risk of bias, and "Unclear" indicates unclear or unknown risk of bias. To make these judgments we will use the criteria indicated by the handbook adapted to  the addiction field. See Appendix 5 for details.

The domains of sequence generation and  allocation concealment (avoidance of selection bias) will be addressed in the tool by a single entry for each study.

Blinding of participants, personnel and outcome assessor (avoidance of performance bias and detection bias) was considered separately for objective outcomes (e.g. drop out, drop out due to adverse events, seizures, delirium, adverse events) and subjective outcomes (e.g. duration and severity of signs and symptoms of withdrawal, craving, psychiatric symptoms; improvements assessed by doctors and patients).

Incomplete outcome data (avoidance of attrition bias) was considered for all outcomes except for the drop out from the treatment, which is very often the primary outcome measure in trials on addiction.

Measures of treatment effect

Dichotomous outcomes were analysed calculating the Relative risk (RR) for each trial with the uncertainty in each result being expressed by their confidence intervals. Continuous outcomes were analysed calculating the MD or the SMD with 95%CI. In case of missing standard deviation of the differences from baseline to the end of treatment, the standard deviation were imputed using the standard deviation of the mean at the end of treatment for each group.

Assessment of heterogeneity

Statistically significant heterogeneity among primary outcome studies will be assessed with Chi‐squared (Q) test and I‐squared (Higgins 2008). A significant Q ( P<.05) and I‐squared of at least 50% will be considered as statistical heterogeneity

Assessment of reporting biases

Funnel plot (plot of the effect estimate from each study against the sample size or effect standard error) was not used to assess the potential for bias related to the size of the trials, because all the included studies had small sample size and not statistically significant results.

Data synthesis

The outcomes from the individual trials have been combined through meta‐analysis where possible (comparability of intervention and outcomes between trials) using a fixed effect model unless there was significant heterogeneity, in which case a random effect model have been used.

If all arms in a multi‐arm trial are to be included in the meta‐analysis and one treatment arm is to be included in more than one of the treatment comparisons, then we divided the number of events and the number of participants in that arm by the number of treatment comparisons made. This method avoid the multiple use of participants in the pooled estimate of treatment effect while retaining information from each arm of the trial. It compromise the precision of the pooled estimate slightly.

Sensitivity analysis

To assess the effect of methodological quality on the results, we first performed a graphical inspection of any effect sorting the results on the forest plots according to risk of bias for sequence generation, allocation concealment, blinding (subjective outcomes) ; if we found a difference in the results between studies at low, unclear, high risk of bias, we performed a sensitivity analysis excluding studies at high risk of bias

Results

Description of studies

Results of the search

We identified 993 reports from all electronic databases searched excluding duplicate, 899 were excluded on basis of title and abstract; 91 articles were retrieved in full text for more detailed evaluation, 35 of which were excluded, 56 satisfied all the criteria to be included in the review. Three studies are awaiting assessment because we are trying to find the full text. SeeFigure 1 to see the Flow chart showing the identification of included trials.

1.

Flow chart showing identification of studies

Included studies

56 studies met the inclusion criteria, with a total of 4076 participants. For a description of the characteristics of the included studies, see Characteristics of included studies table

Country of origin of the included studies

33 studies were conducted in Europe, 18 in North America, 4 in Australia/New Zealand, and one in Asia. 

Number of studies per type of comparison

1. Anticonvulsant versus placebo (No. = 17 studies, 17 comparisons) (Alldredge 1989; Bjorkqvist 1976; Blanchard 1985; Bonnet 2003; Burroughs 1985a; Chance 1991; Gann 2004; Glatt 1966; Golbert 1967; Koethe 2007; Krupitsky 2007; Lambie 1980; Murphy 1983; Rathlev 1994; Reoux 2001; Sampliner 1974; Stanhope 1989)

2. Anticonvulsant versus other drug (No. = 32 studies, 36 comparisons) (Agricola 1982; Borg 1986; Borg 1986; Burroughs 1985a; Burroughs 1985a; Burroughs 1985b; Burroughs 1985b; Choi 2005; Dencker 1978; Elsing 1996; Elsing 2009; Golbert 1967; Kaim 1972; Kaim 1972; Kalyoncu 1996; Koppi 1987; Kramp 1978; Krupitsky 2007; Lapierre 1983; Longo 2002; Lucht 2003; Madden 1969; Malcolm 1989; Malcolm 2002; Malcolm 2007; Manhem 1985; McGrath 1975; Murphy 1983; Nimmerichter 2002; Radouco‐Thomas 1989; Robinson 1989; Santo 1985; Stuppaeck 1992; Stuppaeck 1998; Thompson 1975; Tubridy 1988)

3. Different anticonvulsants between themselves (No.= 10 studies, 11 comparisons) (Flygering 1984; Golbert 1967; Kaim 1972; Krupitsky 2007; Krupitsky 2007; Mariani 2006; Ritola 1981; Rosenthal 1998; Schik 2005; Seifert 2004; Teijeiro 1975)

4. Anticonvulsant combined with other drug versus other drug (No. = 6 studies, 7 comparisons) (Balldin 1986; Golbert 1967; Lucht 2003; Myrick 2000; Spies 1996; Spies 1996; Rothstein 1973)

5. Anticonvulanst combined with other drug vs different anticonvulsant (No=1 study) (Croissant 2009)

For a more detailed information about the comparisons considered in the studies, see Additional Table 3; Table 4; Table 5; Table 6; Table 7

1. Comparisons Anticonvulsants versus placebo.

Author	Treatment (Anticonvulsant)	Control
Alldredge 1989	Phenytoin	Placebo
Bjorkvist 1976	Carbamazepine  (34)	Placebo
Blanchard 1985	Phenobarbital	Placebo
Bonnet 2003	Gabapentin	Placebo
Burroughs 1985 a	Chlormethiazole	Placebo
Chance 1991	Phenytoin	Placebo
Gann 2004	Chlormethiazole	Placebo
Glatt 1966	Chlormethiazole	Placebo
Golbert 1967	Promazine	Placebo
Koethe 2007	Oxcarbazepine	Placebo
Krupitsky 2007	Topiramate	Placebo
Lambie 1980	Valproate	Placebo
Murphy 1983	Chlormethiazole	Placebo
Rathlev 1994	Phenytoin	Placebo
Reoux 2001	Divalproex	Placebo
Sampliner 1974	Phenytoin	Placebo
Stanhope 1989	Carbamazepine	Placebo

2. Comparisons Anticonvulsants versus Other.

Author	Treatment (Anticonvulsant)	Control (Other)
Borg 1986	Amobarbital	Oxazepam  (benzodiazepine)
Borg 1986	Amobarbital	Melperone (antipsychotic)
Kramp 1978	Barbital	Diazepam (benzodiazepine)
Agricola 1982	Carbamazepine	Tiapride (antipsychotic)
Kalyoncu 1996	Carbamazepine	Diazepam (benzodiazepine)
Malcom 1989	Carbamazepine	Oxazepam  (benzodiazepine)
Malcom 2002	Carbamazepine	Lorazepam (benzodiazepine)
Stuppaeck 1992	Carbamazepine	Oxazepam  (benzodiazepine)
Burroughs 1985 a	Chlormethiazole	Bromocriptine (dopamine agonist)
Burroughs 1985 a	Chlormethiazole	Chlordiazepoxide  (benzodiazepine)
Burroughs 1985  b	Chlormethiazole	Bromocriptine (dopamine agonist)
Burroughs 1985  b	Chlormethiazole	Chlordiazepoxide (benzodiazepine)
Dencker 1978	Chlormethiazole	Piracetam (CNS stimulant)
Elsing 1996	Chlormethiazole	GHB (gamma‐Hydroxybutyric acid)
Elsing 2009	Chlormethiazole	GHB (gamma‐Hydroxybutyric acid)
Lapierre 1983	Chlormethiazole	Chlordiazepoxide  (benzodiazepine)
Lucht 2003	Chlormethiazole	Diazepam (benzodiazepine)
Madden 1969	Chlormethiazole	Trifluoperazine (antipsychotic)
Manhem 1985	Chlormethiazole	Clonidine (adrenergic agonist)
McGrath 1975	Chlormethiazole	Chlordiazepoxide (benzodiazepine)
Murphy 1983	Chlormethiazole	Tiapride (antipsychotic)
Nimmerichter 2002	Chlormethiazole	GHB (gamma‐Hydroxybutyric acid)
Robinson 1989	Chlormethiazole	Clonidine (adrenergic agonist)
Tubridy 1988	Chlormethiazole	Alprazolam (benzodiazepine)
Longo 2002	Depakote	Chlordiazepoxide or Loranzepam (benzodiazepine)
Malcom 2007	Gabapentin	Lorazepam (benzodiazepine)
Koppi 1987	Meprobamate	Caroverine (spasmolytic)
Kaim 1972	Paraldehyde	Perhenazine  (antipsychotic)
Kaim 1972	Paraldehyde	Chlordiazepoxide (benzodiazepine)
Thompson 1975	Paraldehyde	Diazepam (benzodiazepine)
Golbert 1967 a	Promazine	Chlordiazepoxide (benzodiazepine)
Raduco‐Thomas 1989	Tetrabamate	Chlordiazepoxide  (benzodiazepine)
Santo 1985	Tetrabamate	Tiapride (Antipsychotic)
Choi 2005	Topiramate	Lorazepam  (benzodiazepine)
Krupitsky 2007	Topiramate	Diazepam (benzodiazepine)
Author	Treatment (Anticonvulsant)	Control
Stuppaeck 1998	Vigabatrin	Oxazepam  (benzodiazepine)

3. Comparisons Anticonvulsant 1 versus Anticonvulsant 2.

Author	Treatment (Anticonvulsant 1)	Control (Anticonvulsant 2)
Flygering 1984	Carbamazepine	Barbital
Scik 2005	Carbamazepine	Oxcarbazepine
Mariani 2006	Chlormethiazole	Phenobarbital
Ritola 1981	Chlormethiazole	Carbamazepine
Seifert 2004	Chlormethiazole	Carbamazepine
Teijeiro 1975	Heminiurine	Phenobarbital+ Ferbamate (tranquillizes)
Kaim 1972	Paraldehyde	Pentobarbital
Golbert 1967  b	Promazine	Paraldehyde and Chloral hydrate (sedative)
Krupitsky 2007	Topiramate	Memantine
Krupitsky 2007	Topiramate	Lamotrigine
Rosenthal 1998	Valproate	Phenobarbital

4. Anticonvulsant + Other versus Other.

Author	Treatment (Anticonvulsant+Other)	Control
Balldin 1986	Carbamazepine +Chlorprothixen (antipsychotic)	Clonidine (adrenergic agonist)
Lucht 2003	Carbamazepine+Tiapride (antipsychotic)	Diazepam (benzodiazepine)
Spies 1996	Chlormethiazole+Haloperidol (antipsychotic)	Clonidine (adrenergic agonist)+fFunitrazepam (benzodiazepine)
Spies 1996	Chlormethiazole+Haloperidol (antipsychotic)	Funitrazepam (benzodiazepine)+ Haloperidol (antipsychotic)
Rothstein 1973	Diphenylhydantoin	Tiamine (antipsychotic)+ Chlordiazepoxide (benzodiazepine)
Myrick 2000	Divalproex+Lorazepam (benzodiazepine)	Lorazepam (benzodiazepine)
Golbert 1967	Paraldehyde and Chloral hydrate (sedative)	Chlordiazepoxide (benzodiazepine)

5. Comparison Anticonvulsant 1 + Other versus Anticonvulsant 2.

Author	Treatment (Anticonvulsant 1+other)	Control (Anticonvulsant 2)
Croissant 2009	Oxcarbazepine + Tiapride (antipsychotic)	Chlormethiazole

Excluded studies

35 studies did not meet the criteria for inclusion in this review. The grounds for exclusion were: type of intervention not in the inclusion criteria: 13 studies; study design not in the inclusion criteria: 12 studies; type of outcomes measures not in the inclusion criteria: 6 studies; duplicate publication: 2 studies, outcome measures presented in a way not suitable for meta‐analysis: 2 studies. SeeExcluded studies Table

Risk of bias in included studies

All the studies were randomised controlled trials.

Allocation

The sequence generation was adequate in 17  studies, unclear in 34 and inadequate in 5 studies; the allocation concealment was adequate in 9 studies, unclear in 42 and  inadequate in 5 studies

Blinding

Blinding for subjective outcomes was adequate in 33 studies, it was unclear in 10 and inadequate in 13

Blinding for objective outcomes was adequate in 52 studies and unclear in 4 studies

Incomplete outcome data

Incomplete outcome data were addressed in 48 studies, it was unclear in 6 studies and were not addressed in 2 studies.

See Included studies Table and Figure 2 and Figure 3 for a more detailed description of risk of bias across the studies.

2.

Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.

3.

Methodological quality summary: review authors' judgements about each methodological quality item for each included study.

With a graphical inspection of the forest plots sorting studies according to the risk of bias, we didn't find any systematic difference in the results between studies at high risk of bias and studies at low or unclear risk of bias. So sensitivity analysis excluding studies at high risk of bias was not performed

Effects of interventions

See: Table 1; Table 2

We only performed meta‐analysis for the studies that had directly comparable interventions and used exactly the same rating scales for continuous outcome measures or had the same binary outcomes. The rest of the data retrieved from the studies (single comparison data) were not synthesized quantitatively. The following results refer to the cases where quantitative synthesis was performed.

The Results are split into four sections referring to the four main comparisons:

1. Anticonvulsant versus Placebo,

2. Anticonvulsant versus Other Drug,

3. Anticonvulsant 1 versus Anticonvulsant 2,

4. (Anticonvulsant + Other drug) versus Other Drug.

5. (Anticonvlusant 1 + other drug) versus Anticonvulsant 2

The outcomes are categorized as primary efficacy outcomes and secondary efficacy outcomes, according to the protocol. We dived them according to efficacy, safety and acceptability.

We decided to not consider separately the comparison between anticonvulsants and benzodiazepines because this is due in the parallel review (Amato 2010) on benzodiazepines for alcohol withdrawal

For a summary of results of some important outcomes see Summary of findings table 1 and Summary of findings table 2

Comparison 1 Anticonvulsant versus placebo:

Efficacy

1.1 Alcohol withdrawal seizures

1.1.1 Any Anticonvulsants versus placebo, 9 studies (Alldredge 1989; Bonnet 2003; Chance 1991; Koethe 2007; Lambie 1980; Murphy 1983; Rathlev 1994; Sampliner 1974; Stanhope 1989), 883 participants, RR 0.61 (0.31 to 1.20), the result is not statistically significant, see Analysis 1.1 or Figure 4

1.1. Analysis.

Comparison 1 Anticonvulsant versus Placebo, Outcome 1 Alcohol Withdrawal Seizures post treatment.

4.

Forest plot of comparison: 1 Anticonvulsant versus Placebo, outcome: 1.1 Alcohol Withdrawal Seizures post treatment.

1.1.2 Phenytoin versus placebo, 4 studies, (Alldredge 1989; Chance 1991; Rathlev 1994; Sampliner 1974), 381 participants, RR 0.78 (0.35 to 1.77), the result is not statistically significant, see Analysis 1.1 or Figure 4

Safety

1.2 Adverse events as number of participants with at least one adverse event

7 studies (Bjorkqvist 1976; Bonnet 2003; Burroughs 1985b; Chance 1991; Krupitsky 2007; Lambie 1980; Stanhope 1989), 516 participants, RR 1.56 (0.74, 3.31), the result is not statistically significant, see Analysis 1.2 or Figure 5

1.2. Analysis.

Comparison 1 Anticonvulsant versus Placebo, Outcome 2 Adverse events.

5.

Forest plot of comparison: 1 Anticonvulsant versus Placebo, outcome: 1.4 Adverse events (N. of patients with at least one adverse event).

Acceptability

1.3 Dropout

1.5.1 Any Anticonvulsants versus placebo , 7 studies (Bjorkqvist 1976; Blanchard 1985; Bonnet 2003; Burroughs 1985a; Gann 2004; Glatt 1966; Reoux 2001), 344 participants, RR 0.82 (0.50 to 1.34), the result is not statistically significant, see Analysis 1.3 or Figure 6

1.3. Analysis.

Comparison 1 Anticonvulsant versus Placebo, Outcome 3 Dropout.

6.

Forest plot of comparison: 1 Anticonvulsant versus Placebo, outcome: 1.5 Dropout.

1.1.2 Chlormethiazole versus placebo , 3 studies (Burroughs 1985a; Gann 2004; Glatt 1966), 140 participants, RR 1.05 (0.22 to 5.11), the result is not statistically significant, see Analysis 1.3 or Figure 6

1.4 Dropout due to adverse events

8 studies (Bjorkqvist 1976; Blanchard 1985; Bonnet 2003; Burroughs 1985a; Chance 1991; Koethe 2007; Lambie 1980; Stanhope 1989), 623 participants, 0.67 (0.13, 3.36), the result is not statistically significant, see Analysis 1.4 or Figure 7

1.4. Analysis.

Comparison 1 Anticonvulsant versus Placebo, Outcome 4 Dropout due to adverse events.

7.

Forest plot of comparison: 1 Anticonvulsant versus Placebo, outcome: 1.6 Dropout due to adverse events

Comparison 2 Anticonvulsant versus Other:

Efficacy

2.1 Alcohol withdrawal seizures

2.1.1 Any Anticonvulsants versus any Other , 12 studies (Agricola 1982; Borg 1986; Kaim 1972; Koppi 1987; Kramp 1978; Lucht 2003; Murphy 1983; Nimmerichter 2002;Radouco‐Thomas 1989 ;Robinson 1989; Stuppaeck 1992; Tubridy 1988), 880 participants, RR 0.58 (0.22 to 1.58), the result is not statistically significant, see Analysis 2.1 or Figure 8

2.1. Analysis.

Comparison 2 Anticonvulsant versus Other Drugs, Outcome 1 Alcohol withdrawal seizures.

8.

Forest plot of comparison: 2 Anticonvulsant versus Other, outcome: 2.1 Alcohol Withdrawal Seizures

2.1.2 Chlormethiazole versus Benzodiazepine, 2 studies (Lucht 2003; Tubridy 1988), 155 participants, RR 0.35 (0.01 to 8.33), the result is not statistically significant, see or Figure 8

2.2 Alcohol withdrawal delirium

2.2.1 Any Anticonvulsants versus any Other , 6 studies (Kalyoncu 1996; Lucht 2003; Manhem 1985;McGrath 1975; Murphy 1983; Stuppaeck 1992; ), 394 participants, RR 0.88 (0.23, 3.42), the result is not statistically significant, see Analysis 2.2 or Figure 9

2.2. Analysis.

Comparison 2 Anticonvulsant versus Other Drugs, Outcome 2 Alcohol withdrawal delirium.

9.

Forest plot of comparison: 2 Anticonvulsant versus Other, outcome: 2.2 Alcohol Withdrawal Delirium

2.2.2 Carbamazepine versus Benzodiazepine, 2 studies (Kalyoncu 1996; Stuppaeck 1992), 125 participants, RR 1.01 (0.04 to 24.43), the result is not statistically significant, see Analysis 2.2 or Figure 9

2.3 CIWA‐Ar score at 48 hours

2.3.1 Any Anticonvulsants versus any Other , 4 studies (Malcolm 1989; Malcolm 2002; Nimmerichter 2002; Stuppaeck 1992), 358 participants, MD ‐0.21 (‐0.95 to 0.53), the result is not statistically significant, see Analysis 2.3

2.3. Analysis.

Comparison 2 Anticonvulsant versus Other Drugs, Outcome 3 CIWA‐Ar score (48 hrs).

2.3.2 Carbamazepine versus Benzodiazepine, 3 studies (Malcolm 1989; Malcolm 2002; Stuppaeck 1992), 260 participants MD ‐0.60 (‐1.88 to 0.67), the result is not statistically significant, see Analysis 2.3

2.4 CIWA‐Ar score at the end of treatment

2.4.1 Any Anticonvulsants versus any Other , 4 studies (Malcolm 1989; Malcolm 2002; Nimmerichter 2002; Stuppaeck 1992), 358 participants, MD ‐0.73 (‐1.76 to 0.31), the result is not statistically significant, see Analysis 2.4

2.4. Analysis.

Comparison 2 Anticonvulsant versus Other Drugs, Outcome 4 CIWA‐Ar score (end of treatment).

2.4.2 Carbamazepine versus Benzodiazepine , 3 studies (Malcolm 1989; Malcolm 2002; Stuppaeck 1992), 260 participants MD ‐1.04 (‐1.89 to ‐0.20), the result is statistically significant in favour of carbamazepine, see Analysis 2.4

2.5 Global Doctor’s Assessment of Efficacy

2 studies (Kramp 1978; Tubridy 1988), 181 participants, RR 0.97 (0.88 to 1.08), the result is not statistically significant, see Analysis 2.5

2.5. Analysis.

Comparison 2 Anticonvulsant versus Other Drugs, Outcome 5 Global Doctor's Assessment of Efficacy.

Safety

2.6 Adverse events as number of participants with at least one adverse event

2.6.1 Any Anticonvulsants versus any Other (Agricola 1982;Burroughs 1985a; Burroughs 1985b;Elsing 2009; Krupitsky 2007; Lapierre 1983; Longo 2002; Lucht 2003; Nimmerichter 2002; Radouco‐Thomas 1989 ;Robinson 1989; Santo 1985; Stuppaeck 1992; Tubridy 1988), 14 studies, 726 participants, RR 0.71(0.45 to 1.12), the result is not statistically significant, see Analysis 2.6 or Figure 10

2.6. Analysis.

Comparison 2 Anticonvulsant versus Other Drugs, Outcome 6 Adverse events.

10.

Forest plot of comparison: 2 Anticonvulsant versus Other, outcome: 2.14 Adverse events (number of patient with at least one adverse event)

2.6.2 Chlormethiazole versus Benzodiazepine , (Burroughs 1985a; Burroughs 1985b; Lapierre 1983; Lucht 2003; Tubridy 1988), 5 studies, 235 participants, RR 0.75 (0.35 to 1.59), the result is not statistically significant, see Analysis 2.6 or Figure 10

2.7 Severe, life‐treating adverse events

2.17.1 Any Anticonvulsants versus any Other , 12 studies (Agricola 1982; Burroughs 1985a; Burroughs 1985b; Dencker 1978;Elsing 2009; Koppi 1987; Krupitsky 2007; Lapierre 1983; Radouco‐Thomas 1989; Santo 1985; Thompson 1975; Tubridy 1988), 578 participants, RR 2.38 (0.33 to 17.24), the result is not statistically significant, see Analysis 2.7 or Figure 11

2.7. Analysis.

Comparison 2 Anticonvulsant versus Other Drugs, Outcome 7 Severe, life‐treatening adverse events.

11.

Forest plot of comparison: 2 Anticonvulsant versus Other, outcome: 2.15 Severe, life‐treating adverse events

2.7.2 Chlormethiazole versus Benzodiazepine , (Burroughs 1985a; Burroughs 1985b; Lapierre 1983; Tubridy 1988), 4 studies, 170 participants, RR 0.69 (95% CI, 0.09 to 5.33), the result is not statistically significant, see Analysis 2.7 or Figure 11

Acceptability

2.8 Dropout

2.8.1 Any Anticonvulsants versus any Other , 20 studies (Agricola 1982; Borg 1986; Burroughs 1985a; Burroughs 1985b; Dencker 1978; Elsing 2009; Kaim 1972; Kalyoncu 1996; Koppi 1987; Kramp 1978; Lucht 2003; Manhem 1985;McGrath 1975; Murphy 1983; Nimmerichter 2002; Radouco‐Thomas 1989 ;Robinson 1989; Santo 1985; Stuppaeck 1992; Tubridy 1988), 1359 participants, RR 0.92 (0.67, 1.26), the result is not statistically significant, see Analysis 2.8 or Figure 12

2.8. Analysis.

Comparison 2 Anticonvulsant versus Other Drugs, Outcome 8 Dropout.

12.

Forest plot of comparison: 2 Anticonvulsant versus Other, outcome: 2.18 Dropout.

2.8.2 Chlormethiazole versus Benzodiazepine, 5 studies (Burroughs 1985a; Burroughs 1985b; Lucht 2003;McGrath 1975; Tubridy 1988), 311 participants, RR 0.68 (0.37, 1.24), the result is not statistically significant, see Analysis 2.8 or Figure 12

2.9 Dropout due to adverse events

2.9.1 Any Anticonvulsants versus any Other , 10 studies (Agricola 1982; Burroughs 1985a; Burroughs 1985b; Dencker 1978; Elsing 2009; Kaim 1972; Koppi 1987; Lapierre 1983; Stuppaeck 1992; Tubridy 1988), 596 participants, RR 2.28 (0.67 to 7.70), the result is not statistically significant, see Analysis 2.9 or Figure 13

2.9. Analysis.

Comparison 2 Anticonvulsant versus Other Drugs, Outcome 9 Dropout due to adverse events.

13.

Forest plot of comparison: 2 Anticonvulsant versus Other, outcome: 2.19 Dropout due to adverse events.

2.9.2 Chlormethiazole versus Benzodiazepine, 4 studies (Burroughs 1985a; Burroughs 1985b; Lapierre 1983; Tubridy 1988), 170 participants, RR 1.09 (0.12 to 9.97), the result is not statistically significant, see Analysis 2.9 or Figure 13

Comparison 3 Anticonvulsant 1 versus Anticonvulsant 2:

Safety

3.1 Adverse events as number of participants with at least one adverse event

3.1.1 Carbamazepine versus Chlormethiazole , 2 studies (Lucht 2003; Ritola 1981), 121 participants, RR 3.12 (0.50 to 19.27), the result is not statistically significant

3.1.2 Carbamazepine versus Barbital , 1 study (Flygering 1984), 61 participants, RR 1.81 (0.70 to 4.68), the result is not statistically significant

3.1.3 Carbamazepine versus Oxcarbazepine , 1 study (Schik 2005), 29 participants, the result is not estimable because there were no side effects in both groups

3.1.4 Chlormethiazole versus Pentobarbital , 1 study (Mariani 2006), 27 participants, RR 2.80 (0.12 to 63.20), the result is not statistically significant

See Analysis 3.1 or Figure 14

3.1. Analysis.

Comparison 3 Anticonvulsant 1 versus Anticonvulsant 2, Outcome 1 Adverse events.

14.

Forest plot of comparison: 3 Anticonvulsant 1 versus Anticonvulsant 2, outcome: 3.1 Adverse events (number of patients with at least one adverse event).

Acceptability

3.2 Dropout

3.2.1 Carbamazepine versus Chlormethiazole , 2 studies (Lucht 2003; Ritola 1981), 121 participants, RR 0.51 (0.08 to 3.11), the result is not statistically significant

3.2.2 Carbamazepine versus Barbital , 1 study (Flygering 1984), 60 participants, RR 0.94 (0.34 to 2.57), the result is not statistically significant

3.2.3 Carbamazepine versus Oxcarbazepine , 1 study (Schik 2005), 29 participants, RR 3.20 (0.14 to 72.62), the result is not statistically significant

3.2.4 Pentobarbital  versus Paraldehyde , 1 study (Kaim 1972), 96 participants, RR 0.37 (0.03 to 3.97), the result is not statistically significant

3.1.4 Chlormethiazole versus Pentobarbital , 1 study (Mariani 2006), 27 participants, RR 1.39 (0.28 to 7.05), the result is not statistically significant

See Analysis 3.2 or Figure 15

3.2. Analysis.

Comparison 3 Anticonvulsant 1 versus Anticonvulsant 2, Outcome 2 Dropout.

15.

Forest plot of comparison: 3 Anticonvulsant 1 versus Anticonvulsant 2, outcome: 3.2 Dropout.

Comparison 4 (Anticonvulsant + Other drug) versus Other Drug:

Efficacy

4.1Alcohol withdrawal delirium

3 studies (Golbert 1967; Lucht 2003; Rothstein 1973), 311 participants, RR 0.79 (0.18 to 3.52), the result is not statistically significant, see Analysis 4.1 or Figure 16

4.1. Analysis.

Comparison 4 (Anticonvulsant + Other) versus Other, Outcome 1 Alcohol withdrawal delirium.

16.

Forest plot of comparison: 4 (Anticonvulsant + Other) versus Other, outcome: 4.1 Alcohol Withdrawal Delirium post treatment.

Safety

4.2 Severe, life‐threatening adverse events

1 study, (Golbert 1967; ), 49 participants, RR 0.13 (0.02 to 0.89), the result is statistically significant in favour of the association Paraldehyde+ chloral hydrate  in respect of chlordiazepoxide, see Analysis 4.2

4.2. Analysis.

Comparison 4 (Anticonvulsant + Other) versus Other, Outcome 2 Severe, life‐threatening adverse events.

Acceptability

4.3 Dropout

Three studies (Golbert 1967; Lucht 2003; Spies 1996), 267 participants, RR 0.56 (0.18 to 1.72), the result is not statistically significant, see Analysis 4.3 or Figure 17

4.3. Analysis.

Comparison 4 (Anticonvulsant + Other) versus Other, Outcome 3 Dropout.

17.

Forest plot of comparison: 4 (Anticonvulsant + Other) versus Other, outcome: 4.3 Dropout.

Comparison 5 (Anticonvulsant 1 + Other drug) versus Anticonvulsant 2

Efficacy

1 study( Croissant 2009), 56 participants, compared Oxcarbaxepine plus Tiapride versus Chlormethiazole. No alcohol withdrawal seizures or delirium happened in both groups. See Analysis 5.1 and Analysis 5.2

5.1. Analysis.

Comparison 5 Anticonvulsant1 + other vs anticonvulsant 2, Outcome 1 alcohol withdrawal seizures.

5.2. Analysis.

Comparison 5 Anticonvulsant1 + other vs anticonvulsant 2, Outcome 2 alcohol withdrawal delirium.

Safety

5.3. Adverse events

1 study (Croissant 2009) 56 participants, compared Oxcarbaxepine plus Tiapride versus Chlormethiazole, RR 0.68 (0.45 to 1.04) , the result is not statistically significant, see Analysis 5.3

5.3. Analysis.

Comparison 5 Anticonvulsant1 + other vs anticonvulsant 2, Outcome 3 adverse events.

5.4Severe, life‐threatening adverse events

1 study ( Croissant 2009), 56 participants, compared Oxcarbaxepine plus Tiapride versus Chlormethiazole. No severe adverse events happened in both groups.Analysis 5.4

5.4. Analysis.

Comparison 5 Anticonvulsant1 + other vs anticonvulsant 2, Outcome 4 severe, life threatening adverse events.

Acceptability

5.5 Drop out

1 study (Croissant 2009) 56 participants, compared Oxcarbaxepine plus Tiapride versus Chlormethiazole, RR 5.36 (0.27 to 106.78) the result is not statistically significant. See Analysis 5.5

5.5. Analysis.

Comparison 5 Anticonvulsant1 + other vs anticonvulsant 2, Outcome 5 drop out.

Discussion

Summary of main results

Fifty‐six studies, with a total of 4076 participants, met the inclusion criteria for this review. However, despite the considerable number of randomised controlled trials, the large variety of outcomes and rating scales considerably limited a quantitative synthesis of data. A large chunk of information could not be synthesized.

‐ No statistically significant differences were found in the four outcomes considered in the comparisons of anticonvulsants versus placebo. Anticonvulsant showed a potentially protective benefit against seizures. The adverse events were non‐significantly more common among the anticonvulsant‐treated patients, but the discontinuations due to adverse events tended to be more common in the placebo‐group. None of these trends, however, reached statistical significance.

‐ For the anticonvulsant versus other drug, results favour anticonvulsants only in one of the nine outcomes considered: comparing carbamazepine with benzodiazepine (oxazepam and lorazepam), results favour carbamazepine for alcohol withdrawal symptoms rated with CIWA‐Ar score.  None of the other comparisons reached statistical significance. This can suggests that anticonvulsants and specifically carbamazepine may actually be more effective in treating some aspects of alcohol withdrawal when compared to benzodiazepines, the current first‐line regimen for alcohol withdrawal syndrome. The incidence of seizures tended to be more common among the participants that were treated with other drugs (e.g. benzodiazepines) than anticonvulsants, but delirium tremens favoured the other‐treatment participants. Adverse events showed a potentially better profile for the anticonvulsants.

‐ Comparing different anticonvulsants (2 outcomes considered), when carbamazepine was compared to Chlormethiazole the incidence of adverse events tended to be more common among the carbamazepine‐treated participants, whereas withdrawals tended to be more common among the Chlormethiazole participants. When carbamazepine was compared to barbital, the side effects were not statistically significant but more common among the carbamazepine patients, and no difference was found regarding the withdrawals between the compared drugs. Out of eleven studies comparing different anticonvulsant agents, no participant died during the treatment period.

‐ Comparing anticonvulsants plus other drugs versus other drugs (3 outcomes considered), results favour paraldehyde plus chloral hydrate versus chlordiazepoxide, for the severe‐life threatening side effects, but only one study, with four arms, and 72 participants was included in this comparison.

‐ Comparing anticonvulsant plus other drug versus another anticonvulsant, no significant difference were found in all the outcome considered (seizures, delirium, adverse events and drop out). But only one study with 56 participants, which compared oxcarbazepine plus Tiapride versus Chlormethiazolewas included in this comparison.

Overall completeness and applicability of evidence

Our review has some important limitations. First, most studies involve a very small sample size and have been conducted in different years, and in diverse patient populations.

Another potential limitation is the unavoidable grouping of drugs. All anticonvulsants were analysed together compared to the other drugs. Differences between specific anticonvulsant agents could not be seen, because of the limited number of studies comparing different anticonvulsants between themselves. Other drugs included a large variety of medicines such as benzodiazepines, neuroleptics, hypnotics, etc.  These drugs do not belong to the same class and do not share the same mechanism of action. The grouping of all these drugs together, could have led to the loss of possibly important between‐drugs differences concerning effectiveness or safety. While we made an effort to address drug‐specific comparisons as well, these data are definitely more limited and even more inconclusive. It was also not possible to examine dose‐response effects, since participants were not treated with even similar doses of various anticonvulsants across RCTs. Moreover, data on side effects should be interpreted cautiously, as they were derived from participants with different co‐morbidity. Many trials tended to exclude participants in severe, medical conditions such as hepatic, heart or lung disease. However, these participants may be more sensitive to various adverse events of anticonvulsants or comparator drugs.

Quality of the evidence

Although randomisation was an inclusion criterion, indicating a degree of methodological quality for these studies, many of them are quite old, the mode of randomisation is not described, allocation of concealment is often unclear and information on follow‐up is frequently missing. On the other hand, this makes the meta‐analysis potentially more useful, since small trends that could not reach statistical significance due to the small number of participants in these trials, could now be revealed after the data synthesis. Also highlighting reporting deficits is important for improving future research.

Authors' conclusions

Implications for practice.

Results of this review do not provide sufficient evidence in favour of anticonvulsants for the treatment of AWS. Anticonvulsants seem to have limited side effects, although adverse effects are not rigorously reported in the analysed trials.

Implications for research.

Although a significant number of trends has emerged, most of these were small and the data for most outcomes did not reach statistical significance; the need for  further studies should be carefully evaluated on the basis of these findings.  If these studies are going to be carried out, they should be limited to few important efficacy and safety measures such as the severity of the alcohol withdrawal syndrome, the incidence of seizures and delirium tremens, side effects, withdrawals and mortality. In the case of continuous outcomes the same rating scale (e.g. CIWA‐Ar) should preferably be used, in order to have more comparable information across studies and to be able to estimate the real effect of all these agents.

What's new

Date	Event	Description
7 January 2010	New citation required and conclusions have changed	new studies founded
7 January 2010	New search has been performed	substantially updated; authors changed

History

Protocol first published: Issue 1, 2005
 Review first published: Issue 3, 2005

Date	Event	Description
21 March 2008	Amended	Converted to new review format.
13 April 2005	New citation required and conclusions have changed	Substantive amendment

Notes

This review is a substantial update of a previous version (Polycarpou 2005). The authors of the first version are not yet interested in updating this review that for that change the authorship.

Appendices

Appendix 1. CENTRAL search strategy

Free text: (((alcohol*) AND (withdraw* or detox* or abstinen* or abstain*)) AND (anticonvulsant*)))

Appendix 2. PubMed search strategy

1. Alcohol‐related disorders [mesh]

2. ((alcohol*) and (disorder* or withdr* or abstinen* or abstain* or detox* or neuropathy)) [tiab]

3. Alcohol‐Induced Disorders, Nervous System [mesh]

4. 1 or 2 or 3

5. ANTICONVULSANTS [Mesh]

6. anticonvulsant* [tiab]

7. (ACTH or carbamazepine or clorazepate or clobazam or clonazepam or chlordiazepoxide or divalproex or ethosuximide or ethosuccimide or ethotoin or felbamate or fosphenytoin or gabapentin or lignocaine or lamotrigine or Levetiracetam or lidocaine or hydantoins or levetiracetam or mephobarbital or  methsuximide or oxcarbazepine or paraldehyde or phenacemide or phenytoin or pregabalin or primidone or succinimide or tiagabine or topiramate or valproate or vigabatrin or zonisamide) [tw]

8. 5 or 6 or 7

9. randomized controlled trial[pt]

10. controlled clinical trial[pt]

11. random*[tiab]

12. placebo[tiab]

13. drug therapy[mesh]

14. trial[tiab]

15. groups[tiab]

16. 9 or 10 or 11 or 12 or 13 or 14 or 15

17. animals [mesh]

18. humans [mesh]

19. animals NOT (17 and 18)

20. 16 NOT 19

21. 4 AND 8 AND 20

22. Anticonvulsants/adverse effects [mesh] or Anticonvulsants/poisoning [mesh] or anticonvulsants/toxicity [mesh] or Anticonvulsants/contraindications [mesh]

23. adverse effects [subheadings]

24. Drug toxicity [mesh]

25. “adverse effect*”[tiab]

26. “side effect* [tiab]

27. serious or safety or surveillance or adverse or toxicity or complication or tolerability [tiab]

28. 22 or 23 or 24 or 25 or 26 or 27

29. 4 AND 8 AND 28

30. 29 NOT 19

Appendix 3. EMBASE search strategy

1. exp alcohol withdrawal

2. exp withdrawal syndrome

3. ((alcohol*) and (disorder* or withdr* or abstinen* or abstain* or detox* or neuropathy)).ti,ab

4. 1 or 2 or 3

5. exp anticonvulsive agent/

6. (ACTH or carbamazepine or clorazepate or clobazam or clonazepam or chlordiazepoxide or divalproex or ethosuximide or ethosuccimide or ethotoin or felbamate or fosphenytoin or gabapentin or lignocaine or lamotrigine or Levetiracetam or lidocaine or hydantoins or levetiracetam or methsuximide or oxcarbazepine or paraldehyde or phenacemide or phenytoin or pregabalin or primidone or succinimide or tiagabine or topiramate or valproate or vigabatrin or zonisamide)

7. 5 or 6

8. random*.ti,ab

9. placebo. Ti,ab

10. (singl* or doubl* or trebl* or tripl*) and (blind* or mask*)).ti,ab

11. crossover*.ti,ab

12. exp randomized controlled trial/

13. exp double blind procedure/

14. exp single blind procedure/

15. exp triple blind procedure/

16. exp crossover procedure/

17. exp Latin square design/

18. exp placebos/

19. exp multicenter study/

20. 8/19

21. 4 AND 7 AND 20

22. limit 21 to human

Appendix 4. CINAHL search strategy

1. MESH alcohol related disorders

2. MESH alcohol withdrawal delirium

3. TX ((alcohol*) and (disorder* or withdr* or abstinen* or abstain* or detox* or neuropathy))

4. 1 or 2 or 3

5. MH ANTICONVULSANTS

6. MH Dipotassium

7. Valproic acid:ME

8. TX Acetazolamide or carbamazepine or Chlormethiazole or Clorazepate or Clorazepate or  divalproex or or Ethosuximide or felbamate or fosphenytoin or gabapentin or lamotrigine or Levetiracetam or Metaclazepam or lidocaine or mephobarbital or lignocaine or methsuximide or oxcarbazepine or paraldehyde or pentobarbital or phenytoin or primidone or tiagabine or topiramate or valproate or vigabatrin or zonisamide

9. 5 or 6 or 7 or 8

10. MH Random Assignment/

11. MH Clinical Trials/

12. TW random*

13. TW placebo*

14. TW group*

15. TW (singl* or doubl* or tripl* or trebl*) and (mask* or blind*)

16. MH crossover design

17. TW crossover*

18. TW allocate*

19. TW assign*

20. 10/19 OR

21. 4 and 9 and 20

Appendix 5. Criteria for risk of bias in RCTs and CCTs

	Item	Judgment	Description
1	Was the method of randomization adequate?	Yes	The investigators describe a random component in the sequence generation process such as: random number table; computer random number generator; coin tossing; shuffling cards or envelopes; throwing dice; drawing of lots; minimization
		No	The investigators describe a non‐random component in the sequence generation process such as: odd or even date of birth; date (or day) of admission; hospital or clinic record number; alternation; judgement of the clinician; results of a laboratory test or a series of tests;  availability of the intervention
		Unclear	Insufficient information about the sequence generation process to permit judgement of ‘Yes’ or ‘No’.
2	Was the treatment allocation concealed?	Yes	Investigators enrolling participants could not foresee assignment because one of the following, or an equivalent method, was used to conceal allocation: central allocation (including telephone, web‐based, and pharmacy‐controlled, randomization); sequentially numbered drug containers of identical appearance; sequentially numbered, opaque, sealed envelopes.
		No	Investigators enrolling participants could possibly foresee assignments because one of the following method was used: open random allocation schedule (e.g. a list of random numbers); assignment envelopes without appropriate safeguards (e.g. if envelopes were unsealed or non ­opaque or not sequentially numbered); alternation or rotation; date of birth; case record number; any other explicitly unconcealed procedure.
		Unclear	Insufficient information to permit judgement of ‘Yes’ or ‘No’. This is usually the case if the method of concealment is not described or not described in sufficient detail to allow a definite judgement
3/4	Was knowledge of the allocated interventions adequately prevented during the study? (blinding of patients, provider, outcome assessor) Objective outcomes Subjective outcomes	Yes	Blinding of participants, providers and outcome assessor and unlikely that the blinding could have been broken; Either participants or providers were not blinded, but outcome assessment was blinded and the non‐blinding of others unlikely to introduce bias. No blinding, but the objective  outcome measurement are not likely to be influenced by lack of blinding
		No	No blinding or incomplete blinding, and the outcome or outcome measurement is likely to be influenced by lack of blinding; Blinding of key study participants and personnel attempted, but likely that the blinding could have been broken; Either participants or outcome assessor were not blinded, and the non‐blinding of others likely to introduce bias
		Unclear	Insufficient information to permit judgement of ‘Yes’ or ‘No’;
5	Were incomplete outcome data adequately addressed? For all outcomes except retention in treatment or drop out	Yes	No missing outcome data; Reasons for missing outcome data unlikely to be related to true outcome (for survival data, censoring unlikely to be introducing bias); Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups; For dichotomous outcome data, the proportion of missing outcomes compared with observed event risk not enough to have a clinically relevant impact on the intervention effect estimate; For continuous outcome data, plausible effect size (difference in means or standardized difference in means) among missing outcomes not enough to have a clinically relevant impact on observed effect size; Missing data have been imputed using appropriate methods All randomized patients are reported/analyzed in the group they were allocated to by randomization irrespective of non‐compliance and co‐interventions (intention to treat)
		No	Reason for missing outcome data likely to be related to true outcome, with either imbalance in numbers or reasons for missing data across intervention groups; For dichotomous outcome data, the proportion of missing outcomes compared with observed event risk enough to induce clinically relevant bias in intervention effect estimate; For continuous outcome data, plausible effect size (difference in means or standardized difference in means) among missing outcomes enough to induce clinically relevant bias in observed effect size; ‘As‐treated’ analysis done with substantial departure of the intervention received from that assigned at randomization;
		Unclear	Insufficient reporting of attrition/exclusions to permit judgement of ‘Yes’ or ‘No’ (e.g. number randomized not stated, no reasons for missing data provided; number of drop out not reported for each group);

Data and analyses

Comparison 1. Anticonvulsant versus Placebo.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Alcohol Withdrawal Seizures post treatment	9		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.1 Any Anticonvulsant vs. Placebo	9	883	Risk Ratio (M‐H, Random, 95% CI)	0.61 [0.31, 1.20]
1.2 Phenytoin vs. Placebo	4	381	Risk Ratio (M‐H, Random, 95% CI)	0.78 [0.35, 1.77]
2 Adverse events	7	663	Risk Ratio (M‐H, Random, 95% CI)	1.56 [0.74, 3.31]
3 Dropout	7		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
3.1 Any Anticonvulsant vs. Placebo	7	344	Risk Ratio (M‐H, Random, 95% CI)	0.82 [0.50, 1.34]
3.2 Chlormethiazole vs. Placebo	3	140	Risk Ratio (M‐H, Random, 95% CI)	1.05 [0.22, 5.11]
4 Dropout due to adverse events	8	649	Risk Ratio (M‐H, Random, 95% CI)	0.67 [0.13, 3.36]

Comparison 2. Anticonvulsant versus Other Drugs.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Alcohol withdrawal seizures	12		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.1 Any Anticonvulsant vs any Other	12	880	Risk Ratio (M‐H, Random, 95% CI)	0.58 [0.22, 1.58]
1.2 Chlormethiazole vs Benzodiazepine	2	155	Risk Ratio (M‐H, Random, 95% CI)	0.35 [0.01, 8.33]
2 Alcohol withdrawal delirium	6		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
2.1 Any Anticonvulsant vs any Other	6	394	Risk Ratio (M‐H, Random, 95% CI)	0.88 [0.23, 3.42]
2.2 Carbamazepine vs Benzodiazepine	2	125	Risk Ratio (M‐H, Random, 95% CI)	1.01 [0.04, 24.43]
3 CIWA‐Ar score (48 hrs)	4		Mean Difference (IV, Random, 95% CI)	Subtotals only
3.1 Any Anticonvulsant vs Other	4	358	Mean Difference (IV, Random, 95% CI)	‐0.21 [‐0.95, 0.53]
3.2 Carbamazepine vs. Benzodiazepine	3	260	Mean Difference (IV, Random, 95% CI)	‐0.60 [‐1.88, 0.67]
4 CIWA‐Ar score (end of treatment)	4		Mean Difference (IV, Random, 95% CI)	Subtotals only
4.1 Any Anticonvulsant vs any Other	4	358	Mean Difference (IV, Random, 95% CI)	‐0.73 [‐1.76, 0.31]
4.2 Carbamazepine vs Benzodiazepine	3	260	Mean Difference (IV, Random, 95% CI)	‐1.04 [‐1.89, ‐0.20]
5 Global Doctor's Assessment of Efficacy	2	181	Risk Ratio (M‐H, Random, 95% CI)	0.97 [0.88, 1.08]
6 Adverse events	14		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
6.1 Any Anticonvulsant vs any Other	14	726	Risk Ratio (M‐H, Random, 95% CI)	0.71 [0.45, 1.12]
6.2 Chlormethiazole vs Benzodiazepine	5	235	Risk Ratio (M‐H, Random, 95% CI)	0.75 [0.35, 1.59]
7 Severe, life‐treatening adverse events	12		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
7.1 Any Anticonvulsant vs anyOther	12	578	Risk Ratio (M‐H, Random, 95% CI)	2.38 [0.33, 17.24]
7.2 Chlormethiazole vs. Benzodiazepine	4	170	Risk Ratio (M‐H, Random, 95% CI)	0.69 [0.09, 5.33]
8 Dropout	20		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
8.1 Any Anticonvulsant vs any Other	20	1359	Risk Ratio (M‐H, Random, 95% CI)	0.92 [0.67, 1.26]
8.2 Chlormethiazole vs Benzodiazepine	5	311	Risk Ratio (M‐H, Random, 95% CI)	0.68 [0.37, 1.24]
9 Dropout due to adverse events	10		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
9.1 Any Anticonvulsant vs any Other	10	596	Risk Ratio (M‐H, Random, 95% CI)	2.28 [0.67, 7.70]
9.2 Chlormethiazole vs Benzodiazepine	4	170	Risk Ratio (M‐H, Random, 95% CI)	1.09 [0.12, 9.97]

Comparison 3. Anticonvulsant 1 versus Anticonvulsant 2.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Adverse events	5		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.1 Carbamazepine vs. Chlormethiazole	2	121	Risk Ratio (M‐H, Random, 95% CI)	3.12 [0.50, 19.27]
1.2 Carbamazepine vs. Barbital	1	61	Risk Ratio (M‐H, Random, 95% CI)	1.81 [0.70, 4.68]
1.3 Carbamazepine vs Oxcarbazepine	1	29	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
1.4 Chlormethiazole vs. Pentobarbital	1	27	Risk Ratio (M‐H, Random, 95% CI)	2.8 [0.12, 63.20]
2 Dropout	6		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
2.1 Carbamazepine vs. Chlormethiazole	2	121	Risk Ratio (M‐H, Random, 95% CI)	0.51 [0.08, 3.11]
2.2 Carbamazepine vs. Barbital	1	60	Risk Ratio (M‐H, Random, 95% CI)	0.94 [0.34, 2.57]
2.3 Carbamazepine vs Oxcarbazepine	1	29	Risk Ratio (M‐H, Random, 95% CI)	3.2 [0.14, 72.62]
2.4 Pentobarbital vs. Paraldehyde	1	96	Risk Ratio (M‐H, Random, 95% CI)	0.37 [0.03, 3.97]
2.5 Chlormethiazole vs. Pentobarbital	1	27	Risk Ratio (M‐H, Random, 95% CI)	1.39 [0.28, 7.05]

Comparison 4. (Anticonvulsant + Other) versus Other.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Alcohol withdrawal delirium	3	311	Risk Ratio (M‐H, Random, 95% CI)	0.79 [0.18, 3.52]
2 Severe, life‐threatening adverse events	1	49	Risk Ratio (M‐H, Random, 95% CI)	0.13 [0.02, 0.89]
3 Dropout	3	267	Risk Ratio (M‐H, Random, 95% CI)	0.56 [0.18, 1.72]

Comparison 5. Anticonvulsant1 + other vs anticonvulsant 2.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 alcohol withdrawal seizures	1	56	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
1.1 Oxcarbaxepine+tiapride vs chlometiazole	1	56	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
2 alcohol withdrawal delirium	1	56	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
2.1 Oxcarbaxepine+tiapride vs chlometiazole	1	56	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
3 adverse events	1	56	Risk Ratio (M‐H, Random, 95% CI)	0.68 [0.45, 1.04]
3.1 Oxcarbaxepine+tiapride vs chlometiazole	1	56	Risk Ratio (M‐H, Random, 95% CI)	0.68 [0.45, 1.04]
4 severe, life threatening adverse events	1	56	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
4.1 Oxcarbaxepine+tiapride vs chlometiazole	1	56	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
5 drop out	1		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
5.1 Oxcarbaxepine+tiapride vs chlometiazole	1	56	Risk Ratio (M‐H, Random, 95% CI)	5.36 [0.27, 106.78]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Agricola 1982.

Methods	Randomized controlled trial
Participants	No. = 60; age = 20‐64 years (mean: 44.28) Inclusion criteria: alcoholic hospitalised patients with a severe withdrawal syndrome proceeding delirium tremens. Exclusion criteria: severe heart, liver or kidney diseases; use of psychotropic drugs; addicted or chronic abusers of other gender = 53 males; 7 females
Interventions	Group A (30) carbamazepine 200mg tid (no. = 30); Group B (30) Tiapride 200mg tid
Outcomes	Withdrawal symptoms; overall evaluation of the clinical condition; assessment of therapeutic effectiveness both by the doctor and the patient; blood pressure and heart rate; SGOT, SGPT; tolerability; seizures
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	Unclear
Allocation concealment?	Unclear risk	Unclear
Blinding? subjective outcomes	Unclear risk	Unclear
Blinding? Objective outcomes	Unclear risk	Unclear
Incomplete outcome data addressed? All outcomes except withdrawal	Unclear risk	Unclear

Alldredge 1989.

Methods	Randomized controlled trial
Participants	No. = 90; gender = 85 males, 5 females; age = 30.3‐50.9. Inclusion criteria: patients admitted to the Emergency Department for a recent seizure occurred in the setting of acute withdrawal from alcohol. Exclusion criteria: history of seizures unrelated to alcohol use, severe head trauma, sedative or anticonvulsant use within 14 days, stimulant drug use within 3 days, significant electrolyte abnormalities, second‐or third‐degree atrioventricular heart block, bradycardia, or known hypersensitivity to hydantoin derivatives.
Interventions	Group A (45) diphenylhydantoin sodium 1000mg intravenously over 20 minutes; Group B (45) equivalent volume of 0.9% sodium chloride
Outcomes	Efficacy : Seizure Recurrence;
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"patients were randomly assigned to receive either diphenylhydantoin sodium injection or infusion of an equivalent volume of 0.9% sodium chloride"
Allocation concealment?	Unclear risk	"patients were randomly assigned to receive either diphenylhydantoin sodium injection or infusion of an equivalent volume of 0.9% sodium chloride"
Blinding? subjective outcomes	Low risk	study described as "double blind placebo controlled trial". "Both diphenylhydantoin and sodium chloride had an identical appearance" For patients with recurrent seizures, the treating physician could request that the study code be broken and subsequent treatment administered. Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	study described as "double blind placebo controlled trial". "Both diphenylhydantoin and sodium chloride had an identical appearance" For patients with recurrent seizures, the treating physician could request that the study code be broken and subsequent treatment administered. Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Unclear risk	"Ninety eligible patients completed the study between October 1982 and June 1988" COMMENT: it is not reported if more than 90 patients were randomised during the study period but they did not complete the study

Balldin 1986.

Methods	Randomized controlled trial
Participants	No. = 38; Age = 28‐57 Inclusion criteria:  Patients alcohol dependent (DMS‐III) Exclusion Criteria: somatic diseases; psychotic symptoms; use of alcohol together with sedatives and anxiolytics.
Interventions	Group A chlorprothixene 50mg x 3 daily+ Carbamazepin 200mg x 2 daily; Group B clonidine 300ìg x 2 daily
Outcomes	Efficacy: Psychiatric rating CPRS; mood: (pleasantness‐unpleasantness) and somatic symptoms (including symptoms in the abstinence situation but also side‐effects from the medication) (self‐rating scales developed by Sjoberg and Persson & Sjoberg were used)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"patients were treated either with alpha2 agonist clonidine or chlorprothixene" COMMENT: it was not stated if the study was randomised
Allocation concealment?	Unclear risk	"patients were treated either with alpha2 agonist clonidine or chlorprothixene" COMMENT: it was not state if the study was randomised
Blinding? subjective outcomes	Low risk	"the design was double blind and double dummy". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	"the design was double blind and double dummy". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"two patients in the chlorprothixene group were excluded from day two and four in the clonidine group ". reason fro drop out given

Bjorkqvist 1976.

Methods	Randomized controlled trial
Participants	No. = 105; Gender: all male.; age 20‐60 years Inclusion criteria: out‐patients volunteers who sought treatment for alcohol withdrawal symptoms at five clinics;  drinking for at least three days;  cooperative;  not in need of hospitalisation. Exclusion criteria: heart, liver, renal insufficiency;  antihypertensive treatment; chronic narcotics; use of drugs
Interventions	Group A (34) carbamazepine; Group B (35)  placebo
Outcomes	Efficacy: Symptoms related to abstinence; evaluation of the patient's global feelings;
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Patients were randomly allocated to two treatments"
Allocation concealment?	Unclear risk	"Patients were randomly allocated to two treatments"
Blinding? subjective outcomes	Low risk	"a double blind technique was used". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	"a double blind technique was used". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	18 patients dropped out from the carbamazepine group and 18 from the the placebo group. Reason fro drop out given

Blanchard 1985.

Methods	Randomized controlled trial
Participants	No. = 38; Gender:29 males/9 females;  Age: 38 years; Inclusion criteria: Hospitalized patients for alcohol withdrawal symptoms; severity score for AWS minimum of 2.
Interventions	Group A Atrium 300 mg; Group B (18)  placebo
Outcomes	Efficacy: Evaluation of global efficacy; Safety: side effects
Notes	Study in French
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"patients were randomised in two parallel group"
Allocation concealment?	Unclear risk	"patients were randomised in two parallel group"
Blinding? subjective outcomes	Low risk	study described as "double blind". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	study described as "double blind". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	1 subject interrupted the treatment in the experimental group and 5 in the placebo group. reason for interruption given

Bonnet 2003.

Methods	Randomized controlled trial
Participants	No. = 61; Age = 36.8‐51.8;  Gender: 43 males/18 females; All Caucasian Inclusion criteria: Inpatients suffering from alcohol dependence (ICD‐10) with moderate or severe AWS (MAWS = or > 4) Exclusion Criteria: psychiatric condition requiring medication; abuse or dependence of other substances;
Interventions	Group A (32) gabapentin 400mg qid for 7 days; Group B (29) placebo.
Outcomes	Efficacy: Amount of CLO required (first 24 h); course of MAWS (first 48 h); Accecptability: premature discontinuations (first 48h);drop out: Safety: adverse events
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"patients were randomised to receive either P or GP"
Allocation concealment?	Unclear risk	"patients were randomised to receive either P or GP"
Blinding? subjective outcomes	Low risk	study described as "double blind". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	study described as "double blind". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"efficacy was analysed for the intention to treat sample, including all subject randomised. Missing data were substituted by an estimated value, derived from regressing the number of CLO capsule on baseline MAWS (for primary effectiveness measure) or by carrying last observation forward"

Borg 1986.

Methods	Randomised controlled trial
Participants	No. = 45; Gender: 100% male; age range: 29‐73 years Inclusion criteria: WHO criteria for alcohol dependence. Exclusion criteria: history of liver cirrhosis or psychotic disorders
Interventions	Group A (15) amobarbital; initial dose=800 mg/day reduced in steps of 100 mg/day during the week.Group B (15) oxazepam; initial dose=120 mg/day reduced in steps of 10 mg/day during the week, Group C (15) Melperone; 200 mg/day throughout the week
Outcomes	Efficacy: seizures, CPRS scores ;  Safety: mortality; Acceptability: dropouts
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"the patients were randomly assigned to treatment with either amobarbital, oxazepam or Melperone"
Allocation concealment?	Unclear risk	"the patients were randomly assigned to treatment with either amobarbital, oxazepam or Melperone"
Blinding? subjective outcomes	Low risk	"patients were treated with amobarbital, oxazepam and Meplerone in a double blind design". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	"patients were treated with amobarbital, oxazepam and Meplerone in a double blind design". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"four patients were excluded before the study was finished from the Melperone group because of epileptic fits"

Burroughs 1985a.

Methods	Randomised controlled trial
Participants	No. = 44 Inclusion criteria: history of alcohol drinking in excess of 80 g/day for five or more years; history of previous alcohol withdrawal. Patient with minimal withdrawal syndrome Exclusion criteria: participants who had taken psychotropic drugs within 48h of hospital administration
Interventions	Group A (12) Chlormethiazole, Group B (11) placebo, Group C (10) chlordiazepoxide, Group D (11) bromocriptine
Outcomes	Efficacy: global improvement, Changes in Gross scale and Borg scale; Safety: adverse events, severe life‐treating adverse events, mortality; Acceptability: dropouts , dropouts due to adverse events
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"the patients in each group were randomised to treatment using a pre‐fixed code devised by the hospital pharmacy from a random number table, with a blocked design to ensure roughly equal numbers in the different treatment groups"
Allocation concealment?	Low risk	"the patients in each group were randomised to treatment using a pre‐fixed code devised by the hospital pharmacy from a random number table, with a blocked design to ensure roughly equal numbers in the different treatment groups"
Blinding? subjective outcomes	Low risk	"the patients in each group were randomised to treatment in double blind fashion using a pre‐fixed code devised by the hospital pharmacy" "The drugs were masked in the same size capsule and were pre‐packaged into daily dosage containers". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	"the patients in each group were randomised to treatment in double blind fashion using a pre‐fixed code devised by the hospital pharmacy" "The drugs were masked in the same size capsule and were pre‐packaged into daily dosage containers". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"three patients withdrawn from the study: Two in the minor withdrawal group from the placebo group. One in the major withdrawal group taking chlordiazepoxide". " Analysis of difference between groups was based on intent to treat regardless of subsequent withdrawal from the study"

Burroughs 1985b.

Methods	Randomised controlled trial
Participants	No. = 27 Inclusion criteria: history of alcohol drinking in excess of 80 g/day for five or more years; history of previous alcohol withdrawal. Patients with intense withdrawal syndrome. Exclusion criteria: participants who had taken psychotropic drugs within 48h of hospital administration
Interventions	Group A (8) Chlormethiazole, Group B (10) chlordiazepoxide, Group C (9) bromocriptine
Outcomes	Efficacy: global improvement, Changes in Gross scale and Borg scale; Safety: adverse events, severe life‐treating adverse events, mortality; Acceptability: dropouts, dropouts due to adverse events
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"the patients in each group were randomised to treatment using a pre‐fixed code devised by the hospital pharmacy from a random number table, with a blocked design to ensure roughly equal numbers in the different treatment groups"
Allocation concealment?	Low risk	"the patients in each group were randomised to treatment using a pre‐fixed code devised by the hospital pharmacy from a random number table, with a blocked design to ensure roughly equal numbers in the different treatment groups"
Blinding? subjective outcomes	Low risk	"the patients in each group were randomised to treatment in double blind fashion using a pre‐fixed code devised by the hospital pharmacy" "The drugs were masked in the same size capsule and were pre‐packaged into daily dosage containers". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	"the patients in each group were randomised to treatment in double blind fashion using a pre‐fixed code devised by the hospital pharmacy" "The drugs were masked in the same size capsule and were pre‐packaged into daily dosage containers". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"three patients withdrawn from the study: Two in the minor withdrawal group from the placebo group. One in the major withdrawal group taking chlordiazepoxide". " Analysis of difference between groups was based on intent to treat regardless of subsequent withdrawal from the study"

Chance 1991.

Methods	Randomized controlled trial
Participants	No. = 55. Inclusion criteria: patients with a complaint of seizure due exclusively to alcohol withdrawal Exclusion criteria: patients with seizure unrelated to alcohol withdrawal;  detectable phenytoin levels (> 2.5 ìg/ml) or allergies or contraindications to hydantoin derivatives; women due the inability to detect early pregnancy and the teratogenic risk of phenytoin.
Interventions	Group A (27) phenytoin 50 mg/ml; Group B (27) 0.9% normal saline
Outcomes	Efficacy: Seizure
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"Patients were randomly assigned to phenytoin or normal saline. The drugs solutions were in identical vials prepared by the pharmacy in an order determined by random‐number tables and sent to the ED labelled only by test solution numbers"
Allocation concealment?	Low risk	"Patients were randomly assigned to phenytoin or normal saline. The drugs solutions were in identical vials prepared by the pharmacy in an order determined by random‐number tables and sent to the ED labelled only by test solution numbers"
Blinding? subjective outcomes	Low risk	"double blind study". "The drugs solutions were in identical vials prepared by the pharmacy". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment.
Blinding? Objective outcomes	Low risk	"double blind study". "The drugs solutions were in identical vials prepared by the pharmacy". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment.
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no withdrawn from the study

Choi 2005.

Methods	Randomized controlled trial
Participants	No: 52. . Gender: male:93%; mean age: 46 Inclusion criteria: meet DSM‐IV criteria for alcohol dependence
Interventions	Group A (25) Topiramate fixed, single dose, 50mg; Group B (27): lorazepam 4mg
Outcomes	Efficacy: Withdrawal symptoms (CIWA‐Ar scale)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Fifty‐two patients after providing written informed consent are randomised to either lorazepam (N=27) or Topiramate (N=25) groups"
Allocation concealment?	Unclear risk	"Fifty‐two patients after providing written informed consent are randomised to either lorazepam (N=27) or Topiramate (N=25) groups"
Blinding? subjective outcomes	Unclear risk	no information about blindness
Blinding? Objective outcomes	Low risk	no information about blindness COMMENT: objective outcomes are unlikely to be influenced by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no withdrawn from the study

Croissant 2009.

Methods	Randomized controlled trial
Participants	No: 56; Gender: male 82,25%; mean age: 45 Inclusion criteria: patients with diagnosis of alcohol dependence according to both ICD‐10 and DSM‐IV, 18 ‐ 65 years of age , who knew enough German to fill in the questionnaires provided and who declared their commitment to the goal of total abstinence. Exclusion criteria: patients with positive screening results for opiates, amphetamine and cocaine ; severe somatic, psychiatric, or terminal diseases ; pregnancy, lactation period, suicidal tendencies, and legal or illegal drug addiction (except nicotine dependence and infrequent THC consumption) within the previous six months. . Patients taking anti‐psychotic drugs or antidepressants, AEDs or benzodiazepines, and patients for whom OXC, Tiaprid or CLO were contraindicated
Interventions	Group A (27) Oxcarbazepine + Tiapride Group B (29): Chlormethiazole
Outcomes	Efficacy: Ciwa‐Ar score;delirium, seizures Safety: adverse events, severe adverse events; Acceptability: drop out
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"We used a table of randomisation to assign subjects to the OXC / TIA or to the CLO‐group. Randomization was effected block by block by five patients, i.e. every five patients the treatment regime changed from OXC / TIA to CLO and vice versa."
Allocation concealment?	High risk	Information on allocation concealment not reported COMMENT: the concealment is judged inadequate given the type of block randomisation used
Blinding? subjective outcomes	High risk	Open label
Blinding? Objective outcomes	Low risk	Open label COMMENT: objective outcomes unlikely to be biased by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	analysis on an intention‐to‐treat basis (ITT), including all patients who received at least one dose of medication.

Dencker 1978.

Methods	Randomized controlled trial
Participants	No. = 63; Gender: 59 males/1 female; Age (mean): Chlormethiazole, 37 years, Piracetam, 39 years; Inclusion criteria: In‐patients, with diagnosis of chronic alcoholism and a previous history of at least 10 years alcohol abuse; had been drinking for at least 7 days; in an acute alcohol withdrawal state. Exclusion criteria: history of heavy drug addiction
Interventions	Group A  (29) Chlormethiazole; Group B (34) Piracetam 1600 mg three times daily
Outcomes	Efficacy: Comprehensive Psychiatric Rating Scale (CPRS); Adjective Check‐List (ACL); Estimation Scale (DES); Memory Test;  laboratory examination; physical, including neurological examination; Safety: Clinical Side‐Effect Scale (CSE)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"patients were randomly allocated to either Piracetam or Chlormerhiazole"
Allocation concealment?	Unclear risk	"patients were randomly allocated to either Piracetam or Chlormerhiazole"
Blinding? subjective outcomes	Low risk	study described as "double blind". Blinding of outcome assessor: "Al the rating were performed by two psychologists who did not communicate with the nursing staff"
Blinding? Objective outcomes	Low risk	study described as "double blind". Blinding of outcome assessor: "Al the rating were performed by two psychologists who did not communicate with the nursing staff"
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"three patients dropped out from trial". One from the Piracetam group, one from the Chlormethiazole group and the third od not stated from which group. Reason fro drop out given

Elsing 1996.

Methods	Randomized controlled trial
Participants	No. = 21; Gender: 18 males/3 females; mean age: GHB, 43.5; Clomethiazole, 42.2 Inclusion criteria: Patients with severe alcohol withdrawal syndrome.
Interventions	Group A (10) Chlomethiazole 250 mg every 4 hours as a mixture ; Group B (11)  GHB initially 30 mg/kg BW followed by 15 mg/kg BW i.v.
Outcomes	Efficacy: Tremor; sweating; nausea; restlessness;
Notes	Meeting Abstract
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"randomised prospective study"
Allocation concealment?	Unclear risk	"randomised prospective study"
Blinding? subjective outcomes	High risk	Blindness: not mentioned. COMMENT: we judged that the study was not blinded because the frequency of drug administration was different between the two groups
Blinding? Objective outcomes	Low risk	Blindness: not mentioned COMMENT: objective outcomes unlikely to be influenced by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no withdrawn from the study

Elsing 2009.

Methods	Randomised controle trail
Participants	No: 26; Gender: 22 male, 4 female. Age: mean 45 Inclusion criteria: Alcohol addicted patients admitted to the intermediate and intensive care unit of the medical department either because of severe withdrawal syndrome or because of additional concomitant life threatening diseases Exclusion criteria: not reported
Interventions	Group a: (12) Chlormethiazole; Group B:(14) Gamma‐Hydroxybutyric acid (GHB)
Outcomes	Efficacy: Ciwa‐Ar score; Safety: side effects, severe side effects Acceptability:. drop out; drop out due to side effect. Other: mortality
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Patients were randomised to receive either GHB (initially 30 mg/kg BWfollowed by 15 mg/kg BW i. v.) or CLO (250 mg every 4 hours as a liquid)"
Allocation concealment?	Unclear risk	"Patients were randomised to receive either GHB (initially 30 mg/kg BWfollowed by 15 mg/kg BW i. v.) or CLO (250 mg every 4 hours as a liquid)"
Blinding? subjective outcomes	High risk	Open label
Blinding? Objective outcomes	Low risk	Open label COMMENT: objective outcome unlikely to be biased by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	1 patient dropped out. Reason given

Flygering 1984.

Methods	Randomized controlled trial
Participants	No. = 72; Age = 19‐62; Gender = 65 males/7 females Inclusion criteria: Patients requiring hospitalisation for alcohol withdrawal symptoms Exclusion Criteria: Concomitant abuse of drugs; full‐blown or incipient delirium tremens; concomitant treatment with antiepileptics, barbital or other psychoactive drugs
Interventions	Group A (37) carbamazepine; Group B (35)  barbital The daily dose individual
Outcomes	Efficacy: Alcohol withdrawal symptoms; visual analogue scale expressing the patient's subjective feeling; consumption of trial medication and duration of treatment; cure rate during treatment; Safety: adverse drug reactions
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Patients were randomly allocated to treatment"
Allocation concealment?	Unclear risk	"Patients were randomly allocated to treatment"
Blinding? subjective outcomes	Low risk	a double blind technique were used". Tablets had identical appearance". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	a double blind technique were used". Tablets had identical appearance". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	6 patients dropped out from treatment from each group. Reason fro drop out given.

Gann 2004.

Methods	Randomized controlled trial
Participants	No.= 20 Gender: not reported; mean age: 41. Inclusion criteria: meet diagnosis of alcohol dependence according to DSM‐IV criteria Exclusion criteria: psychotic features; clinically significant cognitive impairment; abuse of other substances other than alcohol; significant medical problem
Interventions	Group A (11) Chlormethiazole capsule of 384mg; 3 cps on day one, 4 cps on day two, 3 cps on day three, 2 cps on day four, 1 cps on day five; Group B (9): placebo
Outcomes	Efficacy: Withdrawal symptoms (CIWA‐Ar scale); Sleep (polysomnography, total sleep time, sleep efficiency, latency to stage 2, latency to REM, number of wake period, PSQI);
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"patients were randomised to group"
Allocation concealment?	Low risk	the randomisation list was prepared by the pharmacy. Investigators were unaware of the randomisation code"
Blinding? subjective outcomes	Low risk	double blind placebo controlled study
Blinding? Objective outcomes	Low risk	double blind placebo controlled study
Incomplete outcome data addressed? All outcomes except withdrawal	High risk	two patients (15%) in the experimental group and four in the placebo group (31%) excluded before the first night because the withdrawal symptoms reached an extent when medication was considered necessary

Glatt 1966.

Methods	Randomized controlled trial
Participants	No. = 97; Gender=53 males/44 females Inclusion criteria: Alcoholics admitted to the St. Bernard's Hospital in varying states of alcoholic intoxication.
Interventions	Group A (49) Chlormethiazole; Group B (48)  placebo
Outcomes	Efficacy: Individual psychiatric symptoms; prevalence of sedation; over‐all effect of treatment assessed by the doctor and by the patient
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"The test tablets for each patients were provided with a serial number and distributed within the series by mean of random numbers"
Allocation concealment?	Unclear risk	"The test tablets for each patients were provided with a serial number and distributed within the series by mean of random numbers"
Blinding? subjective outcomes	Low risk	"The investigation was carried out as a double blind trial.". To render the placebo indistinguishable in taste from the active tablet, 10 mg of Chlormetiazole was added to the coating of the placebo tablet, an amount that is too small to have any clinical effect". Blindness of outcome assessor: "the outcomes were assessed by a doctor; at the end of the trial he was asked whether he thought the patients had been receiving the active tablet or the placebo"
Blinding? Objective outcomes	Low risk	"The investigation was carried out as a double blind trial.". To render the placebo indistinguishable in taste from the active tablet, 10 mg of Chlormetiazole was added to the coating of the placebo tablet, an amount that is too small to have any clinical effect". Blindness of outcome assessor: "the outcomes were assessed by a doctor; at the end of the trial he was asked whether he thought the patients had been receiving the active tablet or the placebo"
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"three patients did not complete the trial, two from the Chlormethiazole and one from the placebo group"

Golbert 1967.

Methods	Randomised controlled trial
Participants	N=49, Gender: 100% male; age range: 31‐71 years Inclusion criteria: patients admitted to the Veterans Administration Hospital, and in whom alcohol withdrawal syndromes subsequently developed. 47 were classified as in the "tremulous state" and 2 patients were classified as in "acute hallucination"
Interventions	Group A(12) paraldehyde and chloral hydrate, Group B (12) chlordiazepoxide, Group C (13) promazine. Group D (12) alcohol.
Outcomes	Efficacy: delirium, global improvement; Safety: mortality
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	High risk	"Random selection was achieved by assigning each patients medication by rotation"
Allocation concealment?	High risk	"Random selection was achieved by assigning each patients medication by rotation"
Blinding? subjective outcomes	High risk	Blinding: Not mentioned. COMMENT:We judged that the study was not blind because the route of administration of treatment was different (alcohol administered orally vs drug administered intramuscularly)
Blinding? Objective outcomes	Low risk	Blinding: Not mentioned. COMMENT: objective outcomes unlikely to be influenced by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no drop out from the study

Kaim 1972.

Methods	Randomised controlled trial
Participants	No. = 202, Gender: 100% male Inclusion criteria: patients with a clearly established history of alcoholism and manifested all three of the cardinal symptoms of delirium tremens ‐ disorientation, tremor, and hallucinations ‐ during the episode that led to their hospitalisation. Exclusion criteria: frank schizophrenic reaction; obvious chronic brain syndrome; serious medical or surgical conditions; diabetes mellitus; or a diagnosis of epilepsy
Interventions	Group A (46) chlordiazepoxide, Group B (14) placebo, Group C (55) paraldehyde, Group D (41) pentobarbital, Group E (46) perphenazine,
Outcomes	Efficacy: seizures; Nurse Rating Scale; Physicians Symptom Recor d;symptoms checklist; global rating; Safety: mortality; Acceptability: dropouts, dropouts due to adverse events; Other: Treatment Booklet
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"The random assignment of patients to treatments was controlled by a study code number;; patients were assigned consecutive numbers. Prepackaged box of study medication corresponding to the patient's code number." COMMENT: it is not clear how the code numbers were generated
Allocation concealment?	Low risk	"The random assignment of patients to treatments was controlled by a study code number;; patients were assigned consecutive numbers. Prepackaged box of study medication corresponding to the patient's code number."
Blinding? subjective outcomes	High risk	"The parenteral form of chlordiazepoxide is straw‐coloured and required to be mixed with an intramuscular diluent; The IM form of prephenazine is a clear fluid supplied in 2cc. ampoules: IM sodium pentobarbital is colourless fluid, supplied in 5cc ampules.To maintain a partial double blind all IM medication was prepared with matching placebo and each patients was supplied with two injections of equivalent amount, one of which was placebo" "Oral medication other than paraldehyde were in capsule of identical appearance" .Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, COMMENT: the authors described the study as partial double blind; we judged that physicians could easily understand the type of treatment given
Blinding? Objective outcomes	Low risk	"The parenteral form of chlordiazepoxide is straw‐coloured and required to be mixed with an intramuscular diluent; The IM form of perphenazine is a clear fluid supplied in 2cc. ampoules: IM sodium pentobarbital is colourless fluid, supplied in 5cc ampules.To maintain a partial double blind all IM medication was prepared with matching placebo and each patients was supplied with two injections of equivalent amount, one of which was placebo" "Oral medication other than paraldehyde were in capsule of identical appearance" COMMENT: objective outcomes unlikely to be influenced by lack of physician's blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"There were 21 early terminators. 13 for administrative reason and they were not included in the analysis, 8 for treatment failure and they were included in the analysis"

Kalyoncu 1996.

Methods	Randomised controlled trial
Participants	No. = 83, Gender: 100% male; age range: 18‐65 years
Interventions	Group A (34) diazepam; max. 80 mg/day for 7 days, Group B (33) carbamazepine; max. 800mg/day for 7 days.
Outcomes	Efficacy: delirium, MMSE; CIWA score; SCL‐90‐R; Beck depression inventory; global pathology assessment; Safety: mortality; Acceptability: dropouts
Notes	Meeting Abstract
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"patients were randomly assigned to either carbamazepine or diazepam"
Allocation concealment?	Unclear risk	"patients were randomly assigned to either carbamazepine or diazepam"
Blinding? subjective outcomes	Unclear risk	blinding not mentioned COMMENT: there are no sufficient information about treatments to judge if the study could be blinded
Blinding? Objective outcomes	Unclear risk	blinding not mentioned COMMENT: objective outcomes unlikely to be influenced by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	16 patients dropped out from studies. reason given

Koethe 2007.

Methods	RCT
Participants	No.= 50 Gender: 80 % male; mean age: 46. Inclusion criteria: meet diagnosis of alcohol dependence according to DSM‐IV criteria; age between 18 and 60 years; symptoms of alcohol addiction persisting for a period of six months; willing to be hospitalised for alcohol withdrawal; having a history of severe AWS during former detoxification treatments; mentally competent to give written informed consent. Esxclusion criteria: patients showing severe alcohol withdrawal symptoms requiring the administration of CLO  to avoid withdrawal seizures; using illicit drugs, except nicotine; history of drug abuse within the past year; psychiatric disorders or mental impairment limiting the ability to comply with study; acute illness of the respiratory system or reduced breath function or with a known hypersensitivity to CBZ, OXC, or CLO; history of a laboratory abnormality consistent with,or other evidence of a severe medical or neurological disease or malignancy; showing signs of renal or liver insufficiency; hyponatraemia; breath alcohol concentration more than 1.5%;  positive urine drug screening at baseline. pregnancy; breast feeding
Interventions	Group A (24) Oxcarbazepine on titration dosage starting with 600 mg the first day. Reaching the final dose of 900 mg OXC at day 2, the medication was applied over 2 days and tapered off the following 2 days (days 4 and 5).; Group B (26) placebo CLO was permitted (192 mg CLO per capsule, up to 18 capsules a day), and administered to the subjects in a symptom‐triggered manner according to the SAB score
Outcomes	Efficacy:Amount of CLO capsules taken to limit the alcohol withdrawal symptoms; intensity of withdrawal syndrome (AES); anxiety (STAI);  depression (BDI); sleep (PSQI); Safety: adverse effects
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"After admission, subjects were randomly assigned to the 2 treatment groups"
Allocation concealment?	Unclear risk	"After admission, subjects were randomly assigned to the 2 treatment groups"
Blinding? subjective outcomes	Low risk	double blind placebo controlled trial Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	double blind placebo controlled trial Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	3 patients were excluded from the studies from each group; reason given

Koppi 1987.

Methods	Randomized controlled trial
Participants	No. = 20; Gender: 19 males/1 female; Age: 21‐65. Inclusion criteria: Patients with chronic alcoholism presenting mild to moderate withdrawal symptoms (DSM III criteria), Exclusion criteria: other forms of addiction; psychoses; dementias; known epilepsy; severe medical or neurological illnesses; pregnancy.
Interventions	Group A (10) meprobamate mean 2400 mg/24h; Group B (10) Caroverine mean 120 mg/24h
Outcomes	Efficacy: Intensity of clouded consciousness in organic brain syndromes (SKT)‐; Webster rating scale; nurse's observation scale for inpatient evaluation (NOSIE); severity of illness (NGI)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"On a random number basis patients were either assigned to Caroverine or meprobamate"
Allocation concealment?	Unclear risk	"On a random number basis patients were either assigned to Caroverine or meprobamate"
Blinding? subjective outcomes	Low risk	"double blind". "identical capsules were provided". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	"double blind". "identical capsules were provided". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	only one patient dropped out from treatment

Kramp 1978.

Methods	Randomised controlled trial
Participants	No. = 91, Gender: 89% male; age range: 21‐62 years Inclusion criteria: delirium tremens; history of alcohol abuse; actual condition was related to alcohol abuse; severity of the symptoms permitted admission and treatment according to the general routine of the department; intense gross tremor of the extremities and intense perspiration; duration of the symptoms should be at least some hours. Exclusion criteria: intake of psychoactive drugs during the last 24 hours before treatment; alcohol in the blood at the time of treatment; acute event in chronic alcoholic hallucinosis
Interventions	Group A (44) diazepam 20 mg i.m. plus placebo p.os; Group B (47) barbital 500 mg p.os plus placebo i.m.
Outcomes	Efficacy: seizures, global improvement, doctor's assessment of efficacy; Safety: mortality; Acceptability: dropouts; Other: Physical status; mental condition
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"The patients were allocated double blind to treatment" COMMENT: it was not state if the study was randomised
Allocation concealment?	Unclear risk	"The patients were allocated double blind to treatment"
Blinding? subjective outcomes	Low risk	"The patients were allocated double blind to treatment with either barbital (by oral route ) or diazepam (by intramuscular route). Patients received tablets as well as injections when medication was given(active tables plus placebo injection or active injection plus placebo tablet) ". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	"The patients were allocated double blind to treatment with either barbital (by oral route ) or diazepam (by intramuscular route). Patients received tablets as well as injections when medication was given(active tables plus placebo injection or active injection plus placebo tablet) ". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"18 patients were excluded for reasons unrelated to te treatment, 10 in the diazepam group and 8 in the barbital group". Reason fro exclusion reported

Krupitsky 2007.

Methods	Randomised controlled trial
Participants	No.= 127 Gender: 100 % male; mean age: 43. Inclusion criteria: meet DSM‐IV criteria for alcohol dependence; history of most recent alcohol consumption between 8 and 48 hours before study entry; clinically significant alcohol withdrawal symptoms on the basis of the CIWA‐Ar. Exclusion criteria : use of  psychoactive or anticonvulsant medications other than those prescribed in the study; opiate dependence; need of urgent treatment for other symptoms; to be at high risk for untoward side effects from study medications
Interventions	Group A (26) Topiramate 25mg every 6 hours for a total of 100 mg/d, Group B (26) Memantine 10 mg every 8 hours for a total of 30 mg/d, Group C (25) lamotrigine 25 mg every 6 hours for a total daily dose of 100 mg/d, Group D (25) diazepam 10 mg every 8 hours for a total daily dose of 30 mg/d, Group E (25) placebo,
Outcomes	Efficacy: Withdrawal symptom (CIVA‐Ar observer and self rated); Dysphoric mood (MADRS scale); Safety: adverse events; Acceptability: dropouts
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	Block randomisation was used with block sizes of 15, 20, and 25 randomly varied
Allocation concealment?	Low risk	patients randomly chose from blocks of 15, 20, or 25 envelopes, each envelope containing 1 treatment assignment. Assessors were unaware of block sizes and were not involved with any aspect of the randomisation. Patient treatment assignment was kept under lock and key throughout the study
Blinding? subjective outcomes	Low risk	placebo‐controlled randomised single‐blinded Blinding of outcome assessor: "raters blind to the treatment assignment administered an alcohol withdrawal severity scale ". Patients blinded to treatment completed a self reported withdrawal symptoms checklist "Because study medications were not encapsulated, there was a potential for subjects to learn their group assignment by studying their medications if they were aware of distinctive markings associated with each study medication. We suspect that the blind was largely intact in this patient group based on informal clinical interactions with patients, although the integrity of the blind was not formally assessed. To promote the integrity of the blind, no medication could be identified by its administration schedule
Blinding? Objective outcomes	Low risk	placebo‐controlled randomised single‐blinded Blinding of outcome assessor: "raters blind to the treatment assignment"
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"All analyses were performed on an intent‐to‐treat basis" 133 patients recruited, data presented for 127 patients. No information about the 5 patients not included; the number is small, so it is inlikely that this could bias the results.

Lambie 1980.

Methods	Randomized controlled trial
Participants	No. = 48; Gender = 44 males/4 females; Age = 22‐60 (mean 46) years Inclusion criteria: Alcohol dependents showing symptoms of withdrawal; had been drinking within the last 48 hours.
Interventions	Group A (22): sodium valproate 400 mg eight hourly; Group B (27) control.
Outcomes	Efficacy: Symptoms of withdrawal; degree of severity of symptoms
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Pateints were randomly allocated to one of the two groups"
Allocation concealment?	Unclear risk	"Pateints were randomly allocated to one of the two groups"
Blinding? subjective outcomes	High risk	"Patients were unaware of which group they were as other drugs described below were also given" COMMENT: physicians were not blind to treatment; The control group received tranquillizer and/or Chlormethiazole, but it is not specified if the drugs were identical in appearance
Blinding? Objective outcomes	Low risk	"Patients were unawere of which group they were as other drugs described below were also given" COMMENT: The outcomes are unlikely to be influenced by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no drop out from the study

Lapierre 1983.

Methods	Randomised controlled trial
Participants	No. = 40, Gender: 83% male; age range: 24‐60 years Inclusion criteria: patients admitted to hospital for moderate to acute symptoms of alcohol withdrawal; history of alcohol abuse of at least 5 years; current episode of heavy drinking of at least 10 days duration. Exclusion criteria: history or positive urine drug screen for other substances' addiction; acute infections; head or major bony injures; medical conditions involving the cardiovascular; endocrine; pulmonary; and nervous system; gross and severe physical deterioration secondary to excessive alcohol intake resulting in severe renal, hepatic, nutritional, hematological and electrolytic disturbances; and schizophrenic illness
Interventions	Group A (20) Chlormethiazole, Group B (20) chlordiazepoxide
Outcomes	Efficacy: Changes in AWS; TSA; SSA; Safety: adverse events, severe life‐treating adverse events, mortality; Acceptability: dropouts, dropouts due to adverse events; Other: psychophysiological assessment
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	High risk	"Treatment were randomly administered sequentially as the patients were included in the study"
Allocation concealment?	Unclear risk	"Treatment were randomly administered sequentially as the patients were included in the study"
Blinding? subjective outcomes	Low risk	"Double blind was assured by the double dummy technique in which the patient received equal number of either of the two active drugs and the placebo of the other". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	"Double blind was assured by the double dummy technique in which the patient received equal number of either of the two active drugs and the placebo of the other". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	No withdrawn from the study

Longo 2002.

Methods	Randomised controlled trial
Participants	No. = 16, Gender: 50% male; age: 18‐65 Inclusion criteria: DSM‐IV criteria for alcohol dependence; desire to quit drinking; acceptable health; reliable and compliant; CIWA‐Ar score > 8 and < 20 on entry. Exclusion criteria: history of seizures or delirium tremens; significant medical co‐morbidity; abuse or dependence on drugs other than alcohol, cannabis, nicotine or caffeine; DSM‐IV diagnosis of an Axis I psychiatric disorder for which pharmacotherapy was required; pregnancy or lack of birth control; treatment within the month prior to screening and during the six week period with medications which might influence drinking outcomes; no fixed domicile or collateral informant
Interventions	Group A (5) Depakote 5‐day detoxification, Group B (5) Depakote plus 6‐week maintenance, Group C (7) chlordiazepoxide
Outcomes	Efficacy: changes in CIWA‐Ar scores, ADS; TLFB, drinking diary; treatment utilization; CGI; ASI; VCS; OCDS; VSS; SIP; Laboratory values; Safety: adverse events, mortality;
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	High risk	"Patients were sequentially randomised into one of three group"
Allocation concealment?	Unclear risk	"Patients were randomised to receive a standard benzodiazepine detoxification or Depakote detox plus maintenance"
Blinding? subjective outcomes	Low risk	"CIWA‐Ar raters were unaware of subjects cohort group assignment" COMMENT: we judged that the study was not blinded because the treatment differed in duration and way and frequency of drug administration.
Blinding? Objective outcomes	Low risk	COMMENT: the outcomes are unlikely to be influenced by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no withdrawn from the study

Lucht 2003.

Methods	Randomised controlled trial
Participants	No. = 127, Gender: 93% male; mean age = 43.1 years Inclusion criteria: patients with alcohol dependence (ICD‐10) admitted for alcohol detoxification therapy; alcohol withdrawal syndrome; age > or = 18 years; direct admission. Exclusion criteria: delirium on admission; contraindications or severe side effects against the study medications; other drug/substance dependence; pregnancy and lactation; psychosis; severe physical diseases; more than 5 single doses of study medications 2 weeks prior to study
Interventions	Group A (31) Chlormethiazole (max. 3840 mg/day) for 9 days, Group B (28) carbamazepine (max. 1200 mg/day) for 9 days, Group C (34) diazepam (max. 80 mg/day) for 9 days,
Outcomes	Efficacy: seizure; delirium; CIWA‐Ar score; global improvement, VAS; SCL‐90‐R; Safety: adverse events, mortality; Acceptability: dropouts
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	High risk	"Assignment took place in blocks of 10 in order of admission ( first 10 patients into group A, next 10 patients into group B, etc)
Allocation concealment?	High risk	"Assignment took place in blocks of 10 in order of admission ( first 10 patients into group A, next 10 patients into group B, etc)
Blinding? subjective outcomes	High risk	"patients, physicians and nurses were not blind to the study medication"
Blinding? Objective outcomes	Low risk	"patients, physicians and nurses were not blind to the study medication" COMMENT: lack of blinding unlikely to influence outcomes
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no withdrawn from the study

Madden 1969.

Methods	Randomised controlled trial
Participants	No. = 100; Gender: 73 males/27 females Inclusion criteria: alcoholics in treatment of dependency on alcohol and drugs showing alcohol withdrawal features on the day of admission;
Interventions	Group A:  Chlormethiazole; Group B: trifluoperazine
Outcomes	Efficacy: Overall degree of improvement on each of the initial three full days of treatment compared with the patients' state on the day of admission; daily assessment of the presence or absence of abstinence features
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Patients were randomly allocated to one or other preparation"
Allocation concealment?	Unclear risk	"Patients were randomly allocated to one or other preparation"
Blinding? subjective outcomes	Unclear risk	"identical capsules were used" COMMENT: it is not clear if personnel and outcome assessor were blind to treatment
Blinding? Objective outcomes	Low risk	"identical capsules were used" COMMENT: outcomes are unlikely to be influenced by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no withdrawn from the study

Malcolm 1989.

Methods	Randomised controlled trial
Participants	No. = 66, Gender: 100% male; age range: 18‐65 years Inclusion criteria: met DSM‐III criteria for alcohol dependency, MMSE >25; CIWA = or > 20. Exclusion criteria: history of daily use of CNS active drugs, including prescription, nonprescription, and illicit agents; 5 or more days of illicit drug abuse (other than alcohol) in the 30 days before admission; allergic or adverse reactions to oxazepam or carbamazepine; manic‐depressive illness, schizophrenia, or dementia; history of hepatic encephalopathy, jaundice, ascites, diabetes, renal disease, neurologic disease (excluding peripheral neuropathy), or leukopenia; liver function transaminase levels (SGOT, LDH, SGPT) 2.5 times higher than normal; total WBC <4000/mm3; platelet count <100,000/mm3; participating in any drug
Interventions	Group A (32) carbamazepine 800 mg/day for 7 days, Group B (34) oxazepam 120 mg/day for 7 days
Outcomes	Efficacy: changes in CIWA‐Ar scores, physiological measures; neurological measures; self‐report measures; standard psychological testing ( SCL‐90‐R, [Beck depression inventory, State‐Trait anxiety inventory, Wechsler Memory scale)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Subjetcs were blindly assigned to a group who received carbamazepine or to a group who received oxazepam"
Allocation concealment?	Unclear risk	"Subjetcs were blindly assigned to a group who received carbamazepine or to a group who received oxazepam"
Blinding? subjective outcomes	Low risk	study described ad "double blind". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment.
Blinding? Objective outcomes	Low risk	study described ad "double blind". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment.
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	20 subjects dropped out from the study. Reason for drop out give, No difference in number of subjects dropped out between groups

Malcolm 2002.

Methods	Randomised controlled trial
Participants	No. = 136, Gender: 75% male Inclusion criteria: met DSM‐IV criteria for alcohol dependence and alcohol withdrawal; blood alcohol level = or <0.1 g/dl; residence within 50 miles of the study site; MMSE = or >26; admission score on CIWA‐Ar = or >10. Exclusion criteria: all substance abuse syndromes other than alcohol, nicotine or cannabis; major Axis I psychiatric disorder; use of medication in the preceding 30 days that could alter the withdrawal process; history of head injury or other neurologic illness including idiopathic epilepsy; medical instability; electroencephalogram abnormalities; grossly abnormal laboratory values
Interventions	Group A (61) carbamazepine 600‐800 mg on day 1 tapering to 200 mg on day 5, Group B (75) lorazepam 6‐8 mg on day 1 tapering to 2 mg on day 5.
Outcomes	Efficacy: changes in CIWA‐Ar scores; ADS; daily drinking log; Zung Anxiety scale; Beck depression inventory; ability to return to work; sleep quality measures; Safety: mortality
Notes	Malcolm2 2002 is the same RCT assessing different AW outcomes.
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"Subjects randomisation was based on on a computer generated schedule administered by a research pharmacist not involved in data collection"
Allocation concealment?	Low risk	"Subjects randomisation was based on on a computer generated schedule administered by a research pharmacist not involved in data collection"
Blinding? subjective outcomes	Low risk	Study describes as double blind". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment.
Blinding? Objective outcomes	Low risk	Study describes as double blind". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment.
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"Retention rate did not differ between the groups". Reported flow chart of treatment retention

Malcolm 2007.

Methods	Randomised controlled trial
Participants	No.= 101 Gender: 75 % male; mean age: 41. Inclusion criteria: meet DSM‐IV1 criteria for alcohol dependence and for current alcohol withdrawal syndrome;  medically stable and not likely to require hospitalisation for medical complications within 10 days of entry into the study;  score of 10 or higher (CIWA‐Ar); negative urine drug screens for benzodiazepines, other sedative‐hypnotics, opiates, and amphetamine. Exclusion criteria: history of taking medications known to ameliorate or intensify the AWS ; diagnosis of any other substance‐dependence syndrome other than alcohol , except cannabis and cocaine; history of idiopathic epilepsy, schizophrenia, bipolar disorder, or dementia; liver function tests 4 times higher than the upper range of normal; history of hepatic encephalopathy, jaundice, ascites, insulin‐dependent diabetes mellitus, or renal insufficiency .
Interventions	Group A (20) Gabapentin, mid‐range dose 300 mg 3 times a day for 3 days and 300 mg twice daily for the last day, Group B (21) lorazepam 2 mg 3 times a day for the first 3 days and 2 mg twice daily on the last day of treat­ment
Outcomes	Efficacy: Insomnia measured by CIWA‐AR scale; Quality of sleep measured by Beck Depression Inventory (BDI); Sleepiness measured by the Eporth Sleepiness Scale;
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	Subjects were randomised using a stratified permuted block method in which individuals with particular characteristics were assigned to medication groups with constraint that the assignment was balanced within successive blocks of subjects. We used a relatively small block size of q=4in order to promote a high degree of balance. Stratifications were based on the intersection of previous alcohol treatment history (0‐1 or more than 1 past treated withdrawals) and sex (male or female).
Allocation concealment?	Unclear risk	not specified if providers including subjects were aware of the criteria used to allocate patients to groups
Blinding? subjective outcomes	Low risk	"double blind study" "Medications prepared for the project were identical capsules prepared and distributed by the Alcohol Research Center Shared Scientific Core pharmacists under supervision by the University Research Pharmacy Office"
Blinding? Objective outcomes	Low risk	"double blind study" "Medications prepared for the project were identical capsules prepared and distributed by the Alcohol Research Center Shared Scientific Core pharmacists under supervision by the University Research Pharmacy Office"
Incomplete outcome data addressed? All outcomes except withdrawal	Unclear risk	reported the number of subjects excluded or who dropped out for treatment but not divided for group assignment

Manhem 1985.

Methods	Randomized controlled trial
Participants	N= 67; Age: 32‐60 Inclusion criteria: alcohol dependents (DSM III) with acute alcohol withdrawal syndrome requiring drug therapy; Exclusion criteria: history, medical examination or laboratory evidence suggesting severe neurologic, psychiatric, hepatic or cardio‐vascular illness. patients using prescription or illicit drugs
Interventions	Group A (10) Chlormethiazole 4 x 500 mg; Group B (10)  clonidine 4 x 150 ìg
Outcomes	Efficacy: Withdrawal Assessment Scale;
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	Unclear
Allocation concealment?	Unclear risk	Unclear
Blinding? subjective outcomes	Unclear risk	Unclear
Blinding? Objective outcomes	Unclear risk	Unclear
Incomplete outcome data addressed? All outcomes except withdrawal	Unclear risk	Unclear

Mariani 2006.

Methods	Randomised controlled trial
Participants	No.= 27 Gender: 70 % male; mean age: 44. Inclusion criteria: met DSM‐IV1 criteria for alcohol dependence; age between 18 and 60 years;  score of ten or higher as measured by the CIWA‐Ar); willingness to sign informed consent. Exclusion criteria: any major DSM‐IV Axis I non‐substance‐related psychiatric disorder; any known allergy to gabapentin or phenobarbital; pregnancy; documented AIDS (HIV+ status was not an exclusion criterion); severe unstable medical illness; alcohol withdrawal delirium; use of opioids (methadone maintenance was not an exclusion criterion) or sedative‐hypnotic agents (eg, benzodiazepines or barbiturates).
Interventions	Group A (14) gabapentin Day 1 total of 2400 mg in the first 24 hours Day 2‐4. 600 mg Group B (13) Phenobarbital Day 1‐3 60mg , day 4. 30mg
Outcomes	Efficacy: Withdrawal symptom (CIWA‐Ar); mood (POMS); craving (ACS); irritability (BDHI);anxiety (Hamilton?A);dysphoria (RDS); sleep (SPS, hours of sleep)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"patients were randomised to receive one of the two treatments"
Allocation concealment?	High risk	"The principal investigator conducted all screening interviews and administered all rating scales. The study was open‐label—all participants, research staff, and clinical staff were aware of the identity of the treatment medication.
Blinding? subjective outcomes	High risk	"The study was open‐label—all participants, research staff, and clinical staff were aware of the identity of the treatment medication."
Blinding? Objective outcomes	Low risk	"The study was open‐label—all participants, research staff, and clinical staff were aware of the identity of the treatment medication". COMMENT: objective outcomes unlikely to be biased by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	The proportion of completers did not differ between treatment groups: gabapentin, 71%; phenobarbital, 62%( p : 0.70). reason for drop out given

McGrath 1975.

Methods	Randomised controlled trial
Participants	No: 100, Gender: not reported, age range: not reported Inclusion criteria: patients with acute withdrawals phase of alcoholism. No further detailde given Exclusion criteria: not reported
Interventions	Group A: Chlormethiazole (50) group B: Chlordiazepoxide (50)
Outcomes	Efficacy: delirium Acceptaibility: drop out
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"One hundred consecutive admissions to a specialized unit for the treatment of alcoholism in a psychiatric hospital were allotted code numbers and in accordance with a list of randomised numbers"
Allocation concealment?	Low risk	"The true nature of the medication was known only to the hospital pharmacist who packaged them for each patient according to number. It was agreed that the code could be broken only if a patient's condition deteriorated to the extent that knowledge of the medication he was receiving was essential."
Blinding? subjective outcomes	Low risk	"As the two preparations being tested were very different in presentation, one being a tablet and the other a capsule with Arachis oil, it was decided that both tablet and capsule would be given to each patient. Patients received either active capsules and placebo tablets, or vice versa."
Blinding? Objective outcomes	Low risk	"As the two preparations being tested were very different in presentation, one being a tablet and the other a capsule with Arachis oil, it was decided that both tablet and capsule would be given to each patient. Patients received either active capsules and placebo tablets, or vice versa."
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	14 patients dropped out from Chlordiazepoxide group and 7 from Chlormethiazole because feelings of tension and restlessness or to breaking of the code due to the development of delirium tremens

Murphy 1983.

Methods	Randomized controlled trial
Participants	No. = 68; Gender: 51 males/17 females; Age: 21‐65 years Inclusion criteria: addicted to alcohol with withdrawal symptoms Exclusion criteria: alcoholic hallucinatory syndrome or delirium; psychosis of any kind or past psychosis not related to alcohol abuse; clinical or documented evidence of brain damage including the Wernicke‐Korsakoff syndrome; intercurrent physical illness necessitating intervention; any evidence of epilepsy not related to alcohol abuse; ingestion of psychotropic drugs less than 24 h before admission.
Interventions	Group A (29) Chlormethiazole; Group B (30) Tiapride; Group C (9) placebo
Outcomes	Efficacy: Depression (Wakefield Inventory of Zung's Self‐Rated Depressive Scale); Severity of Alcohol Dependence Questionnaire; withdrawal symptoms (Severity Assessment Scale of Gross); craving for alcohol;
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Patients were randomly assigned to one of the treatment group after stratification for age, sex and blood alcohol concentration"
Allocation concealment?	Unclear risk	"Patients were randomly assigned to one of the treatment group after stratification for age, sex and blood alcohol concentration"
Blinding? subjective outcomes	Low risk	"The study is a double blind, blindness being achieved by the use of double dummy technique". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	"The study is a double blind, blindness being achieved by the use of double dummy technique". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	11 patients dropped out of the trial. Rason for drop out given. Number of drop out balanced among group (5 from Tiapride, 4 from Chlormethiazole, 2 from placebo)

Myrick 2000.

Methods	Randomized controlled trial
Participants	No. = 11; Gender = Both males and females. Age = 18‐65 Inclusion criteria: Alcohol dependents (DSM II R) requiring alcohol detoxification; (CIWA‐Ar) scale of 6 or higher. Exclusion criteria: serious medical conditions; major psychiatric disorders; liver function tests 3 times greater than normal.
Interventions	Group A (6)  divalproex 500mg 3 times a day for 4 days)+Lorazepam 2mg; Group B (5) lorazepam 2mg
Outcomes	Efficacy: CIWA‐Ar score;
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Subjects were placed randomly on Depakote group. The CIWA‐ar only group is the comparison group".
Allocation concealment?	Unclear risk	"Subjects were placed randomly on Depakote group. The CIWA‐ar only group is the comparison group".
Blinding? subjective outcomes	Low risk	"Open label clinical trial". Blindness of outcome assessor: "CIWA‐Ar raters were unaware of treatment assignment"
Blinding? Objective outcomes	Low risk	"Open label clinical trial". COMMENT: objective outcomes unlikely to be biased by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no withdrawn from the trial

Nimmerichter 2002.

Methods	Randomized controlled trial
Participants	No. = 98; Gender = 89 males/9 females; Age = 31.7‐50.7 Inclusion criteria: Patients meet at least 3 DSM‐IV and ICD‐10 criteria for the diagnosis of alcohol dependence; severity of withdrawal syndrome = or > 20 points on the CIWA‐Ar,; elevated GGT and/or MCV and/or relative carbohydrate‐deficient transferrin levels. Exclusion criteria: patients already detoxified or dependent on substances other than alcohol; illnesses such as decompensated liver cirrhosis, acute pancreatitis, reduced respiratory function, acute pulmonary or bronchial disease, myocardial infarction during the preceding 6 months and cardiac arrhythmia; dementia, schizophrenia, polysubstance misuse or dependence or use of benzodiazepines and hypno‐sedatives; suspected sleep apnoea syndrome; allergy to either of the study drugs
Interventions	Group A (33) Chlormethiazole 1000 mg daily; Group B (33) GHB 50 mg GHB/kg body wt; Group C (32)  GHB 100 mg GHB/kg body wt
Outcomes	Efficacy: Withdrawal syndrome (CIWA‐Ar); clinical and biological parameters; Lubeck Craving‐Risk‐Relapse questionnaire (LCRR‐1); additional study medication; Acceptabilty: side effects;
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"Participants were randomised to one of the three treatments by means of a computerized randomisation schedule"
Allocation concealment?	Unclear risk	"Participants were randomised to one of the three treatments by means of a computerized randomisation schedule"
Blinding? subjective outcomes	Low risk	Double blind, double dummy. Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	Double blind, double dummy. Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"11 patients discontinued the study prematurely". Reason for drop out given. Analysis performed on the intention to treat basis

Radouco‐Thomas 1989.

Methods	Randomised controlled trial
Participants	No.= 67; Gender: 100% male; age range: 32‐60 years Inclusion criteria: patients admitted to the detoxification unit of Quebec Hospital; met the DSM‐III criteria for alcohol dependence; presented acute alcohol withdrawal syndrome requiring drug therapy. Exclusion criteria: history, medical examination or laboratory evidence suggesting severe neurologic, psychiatric, hepatic or cardiovascular illness and the use of prescription or illicit drugs (polydrug addiction
Interventions	Group A (30) oral phenobarbital, Group B (30) oral chlordiazepoxide. Additional medication doses were given according to the clinician judgment
Outcomes	Efficacy: seizures, CIWA‐Ar score, REG; vital signs and sleep evaluation; DSST, PPT; HSCL‐35, Zung Self Rating Anxiety Scale, Zerssen bipolar mood test‐Z; changes in blood alcohol levels; Safety: adverse events, severe life‐treating adverse events, mortality; Acceptability: dropouts
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"After enrolment, patients were assigned to receive the next in a sequentially numbered supply of medication. The sequence was previously determined by the hospital pharmacist from a table of random numbers with a blocked design to ensure a roughly equal numbers in the two treatment group and in the different season of the year"
Allocation concealment?	Low risk	"After enrolment, patients were assigned to receive the next in a sequentially numbered supply of medication. The sequence was previously determined by the hospital pharmacist from a table of random numbers with a blocked design to ensure a roughly equal numbers in the two treatment group and in the different season of the year"
Blinding? subjective outcomes	Low risk	Double blind. "Double dummy administration procedure". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	Double blind. "Double dummy administration procedure". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	7 patients drop out from the study. Reason for drop out given. No differences in the number of drop out between groups

Rathlev 1994.

Methods	Randomized controlled trial
Participants	No. = 100; Gender = 96 males/4 females Age (mean) = placebo, 41.5 years; phenytoin, 42.4 years. Inclusion criteria: Patients with a witnessed generalized seizure in the setting of decreasing alcohol intake, and had no other known or suspected cause of seizures. Exclusion criteria: known or potential causes of seizure by head CT scan, EEG, lumbar puncture, or laboratory studies; onset of seizures before age 25; focal seizures or status epilepticus; positive phenytoin level; focal neurologic examination; known ingestion of drugs that potentially cause seizures; treatment with benzodiazepines for alcohol withdrawal symptoms; significant head trauma; pregnancy; phenytoin allergy.
Interventions	Group A (49)  phenytoin 15 mg/kg at a rate equivalent to 50 mg/min over 20 minutes by iv pump; Group B (51)  placebo
Outcomes	Efficacy: seizure; Acceptability: completion of a six‐hour observation period after infusion.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"Patients were randomised to receive either phenytoin or saline placebo according to a table of random number"
Allocation concealment?	Unclear risk	"Patients were randomised to receive either phenytoin or saline placebo according to a table of random number"
Blinding? subjective outcomes	Low risk	Double blind study. "The drugs ware prepared in identical ampoules and were administered in a blinded fashion". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	Double blind study. "The drugs ware prepared in identical ampoules and were administered in a blinded fashion". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no withdrawn from the study

Reoux 2001.

Methods	Randomized controlled trial
Participants	No. = 36; Gender = 35 males/1 female; Age = 40.8‐58 Inclusion criteria: volunteers seeking treatment for alcohol dependence and withdrawal;  met DSM‐IV criteria for alcohol dependence; experiencing at least moderate acute alcohol withdrawal (score of 10 or greater on the CIWA‐Ar scale) Exclusion Criteria: 6 or more days of illicit drug use, other than cannabis, in the 30 days before admission; allergic or adverse reaction to prior use of oxazepam or valproate; current use of an anticonvulsant medication or benzodiazepine; pregnancy; medical condition potentially risky
Interventions	Group A (18)  divalproex sodium 500 mg 3 times a day; Group B (18)  inactive placebo
Outcomes	Efficacy: Total milligrams of oxazepam required to control AWS symptoms; withdrawal symptoms (CIWA‐Ar); changes in SCL‐90 scores; Acceptability: side‐effects
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Subjects were randomly assigned to either a placebo or divalproex sodium condition."
Allocation concealment?	Unclear risk	"Subjects were randomly assigned to either a placebo or divalproex sodium condition."
Blinding? subjective outcomes	Low risk	Study described as double blind. Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	Study described as double blind. Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"7 subjects (3 in the divalproex, 4 in the placebo) dropped out before the remission of the acute withdrawal symptoms and were not included in the analysis"

Ritola 1981.

Methods	Randomized controlled trial
Participants	No. = 68; Gender = 100% males; Age =37.4‐41.6. Inclusion criteria: patients considered sufficiently severe as to require hospitalisation. Exclusion Criteria: clear abnormalities in cardiac, hepatic or renal functions; using psychoactive agents other than the trial preparations
Interventions	Group A (34): carbamazepine 200 mg; Group B: (34) Chlormethiazole 300 mg
Outcomes	Efficacy: Withdrawal symptoms; patient's subjective feeling; overall evaluation of the treatment (patient, nurse, physician); Acceptability:side effects;
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"The patients were randomly allocated to two treatments"
Allocation concealment?	Unclear risk	"The patients were randomly allocated to two treatments"
Blinding? subjective outcomes	Low risk	"The double blind and double dummy technique was achieved by using placebo capsules and placebo tablets which were given toghether with the arrpopiate active drugs". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	"The double blind and double dummy technique was achieved by using placebo capsules and placebo tablets which were given toghether with the arrpopiate active drugs". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	5 patients (44 from Chlormethiazole and 1 from carbamazepine) dropped out from the study. Reason for drop out given

Robinson 1989.

Methods	Randomized controlled trial
Participants	No. = 32; Gender: 26 males/6 females Age (mean): Clonidine, 50.8±9.3 years; Chlormethiazole, 44.8 ± 8.4 years. Inclusion criteria: patients requiring drug treatment for the suppression of withdrawal symptoms. Exclusion Criteria: no significant symptoms of alcohol withdrawal (WSRscore < 3); standing systolic blood pressure < 100 mmHg; major withdrawal symptoms (WSR score > 20); history of previous withdrawal seizures; multidrug dependency problem; taking drugs affecting sympathetic nervous activity  in the week prior to admission
Interventions	Group A (16):  Chlormethiazole; Group B: (16) clonidine
Outcomes	Efficacy: alcohol withdrawal symptoms (WSR scale);
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Medication was administered in a randomised double blind manner".
Allocation concealment?	Unclear risk	"Medication was administered in a randomised double blind manner".
Blinding? subjective outcomes	Low risk	"Medication was administered in a randomised double blind manner".. Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	"Medication was administered in a randomised double blind manner".. Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	High risk	"The study was stopped when 25% of the patients had been withdrawn from the trial. All 16 patients who received chlormethiazole completed the trial uneventfully. However there was a significant number (8) of withdrawals for the clonidine group (P:0.002)

Rosenthal 1998.

Methods	Randomized controlled trial
Participants	No. = 42; Gender: 36 males/6 females;  Mean age: Valproate group, 43.5±6.7 years; Phenobarbital group, 41.5 ± 10.7 years. Inclusion criteria: alcohol dependents in withdrawal (DSM‐IV criteria); willingness to sign informed consent, age between 18 and 55 years; and criteria for alcohol dependence in withdrawal. Exclusion criteria: any major DSM‐IV Axis I disorder; documented liver disease; known allergy to valproate; pregnancy; documented AIDS (HIV+ not an exclusion criterion); coexisting use of cocaine or illicit opioids (methadone maintenance not an exclusion criterion); abuse of a half‐life benzodiazepine
Interventions	Group A phenobarbital; Group B valproate
Outcomes	Efficacy: withdrawal symptoms and signs: Modified Selective Severity Assessment (MSSA); modified Abstinence Symptoms Questionnaire (ASQ);Profile of Mood States (POMS‐SF); Buss‐Durkee Hostility Index (BDHI).
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Patients were randomised on day 1 to receive either phenobarbital or valproate"
Allocation concealment?	Unclear risk	"Patients were randomised on day 1 to receive either phenobarbital or valproate"
Blinding? subjective outcomes	High risk	"The study was open label"
Blinding? Objective outcomes	Low risk	"The study was open label" COMMENT:outcomes are unlikely to be influenced by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	5 patients dropped out from study, 4 from valproate and one from control. Reason fro drop out given

Rothstein 1973.

Methods	Randomized controlled trial
Participants	No. = 200 Inclusion criteria: have a history of chronic alcoholism; have been drinking heavily every day for at least five days immediately preceding his admission; give written, informed consent Exclusion criteria: previous long‐term diphenylhydantoin therapy for convulsive disorders (other than withdrawal seizures); seizures during the two weeks just before admission
Interventions	Group A (100) diphenylhydantoin 2 x 200 mg orally; Group B chlordiazepoxide + thiamine(100)
Outcomes	Efficacy: Seizures
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"The 200 patients were randomly assigned to either a diphenylhydantoin group or a control group using the following procedure:at the start of the study we obtained a random number list of 200 number balanced after each 20 to assure equal odd and even numbers.Upon admission each new patient eligible was assigned the nest available number in the random number list; cases with odd numbers received diphenylhydantoin and cases with even numbers served as control. "
Allocation concealment?	Unclear risk	"The 200 patients were randomly assigned to either a diphenylhydantoin group or a control group using the following procedure:at the start of the study we obtained a random number list of 200 number balanced after each 20 to assure equal odd and even numbers.Upon admission each new patient eligible was assigned the nest available number in the random number list; cases with odd numbers received diphenylhydantoin and cases with even numbers served as control. "
Blinding? subjective outcomes	High risk	blinding not mentioned COMMENT: we judged that study was not blind because the diphenylhydantoin group received this drug as an adjunct to the drugs received by the control group
Blinding? Objective outcomes	Low risk	blinding not mentioned COMMENT: outcome unlikely to be biased by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no withdrawn from the study

Sampliner 1974.

Methods	Randomized controlled trial
Participants	No. =157. Inclusion criteria: history of seizures in adulthood; no medications in the two weeks before admission;  steady, heavy alcoholic intake for the preceding four weeks.
Interventions	Group A (70)  diphenylhydantoin 300 mg/day; Group B (66) placebo
Outcomes	Efficacy: seizures
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"Patients were assigned randomly to diphenylhydantoin or placebo according to a computer generated set of numbers"
Allocation concealment?	Unclear risk	"Patients were assigned randomly to diphenylhydantoin or placebo according to a computer generated set of numbers"
Blinding? subjective outcomes	Unclear risk	"Diphenylhydantoin or placebo were blindly administered." " Diphenylhydantoin and placebo were placed in gelatine capsules" COMMENT: it is not clear if also personnel and outcome assessors were blind to treatment assignment
Blinding? Objective outcomes	Low risk	"Diphenylhydantoin or placebo were blindly administered." " Diphenylhydantoin and placebo were placed in gelatine capsules" COMMENT: it is not clear if also personnel and outcome assessors were blind to treatment assignment but outcomes are unlikely to be biased by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	8 patients from diphenylhydantoin group and 13 from the placebo group were excluded from the study. Reason for exclusion given

Santo 1985.

Methods	Randomized clinical trial
Participants	No. = 38; Gender: All males Age: Group 1: 44 years, Group 2: 43 years. Inclusion criteria: Patients with alcohol withdrawal syndrome hospitalised for reasons other than alcoholism Exclusion criteria: severe delirium tremens.
Interventions	Group A (15) Tetrabamate 300 mg; Group B (17)  Tiapride 100 mg.
Outcomes	Efficay: Number of patients with withdrawal symptoms at day 14; duration of symptoms per group
Notes	Study in Spanish
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	a tables of random numbers was used
Allocation concealment?	Unclear risk	a tables of random numbers was used
Blinding? subjective outcomes	Low risk	study described as double blind Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	study described as double blind Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	3 patients from each group were excluded from the study. Reason for exclusion given

Schik 2005.

Methods	Randomized clinical trial
Participants	No.= 29 Gender: 83 % male; mean age: 41. Inclusion criteria: voluntary admitted to hospital with a alcohol withdrawal syndrome according to ICD‐10 Exclusion criteria: polydrug abuse; delirium tremens upon admission; severe hepatic or haematological complications; hyponatraemia; abnormal ECG; known epileptic seizures;  history of head injury; neuroleptic or antiepileptic medication as well as MAO‐inhibitor intake
Interventions	Group A (14 ) carbamazepine, 600 mg on days 1 ‐ 3, 300 mg on day 4 and a last dose of 100 mg on day 5. Group B (15) oxcarbazepine , 900 mg on days 1 ‐ 3, 450 mg on day 4 and a final dose of 150 mg on day 5
Outcomes	Efficacy: Withdrawal symptom (CIWA‐Ar); Acceptability: subjective perception of side affect and symptoms (VAS)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"14 patients were assigned randomly to each of the two treatment groups".
Allocation concealment?	Unclear risk	"14 patients were assigned randomly to each of the two treatment groups."
Blinding? subjective outcomes	High risk	single blind "The type of medication was not disclosed to the patient, but the physician who measured withdrawal symptoms of the particular patient was aware of the treatment regimen."
Blinding? Objective outcomes	Low risk	"The type of medication was not disclosed to the patient, but the physician who measured withdrawal symptoms of the particular patient was aware of the treatment regimen." COMMENT: obljective outcomes are unlikely to be biased by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no withdrawn from the study

Seifert 2004.

Methods	Randomized controlled trial
Participants	No. =159; Gender = 150 males/9 females; Age = 18‐83 Inclusion criteria: met the DSM‐III‐R criteria for alcohol dependence. Exclusion criteria: less than 18 years of age; chronic obstructive disease; poor pulmonary function or pneumonia; bradycardia; systolic blood pressure of less than 95 mmHg; second‐ or third‐degree atrioventricular node block; history of current use or abuse of benzodiazepines, barbiturates, clonidine, or beta‐adrenergic receptor blockers.
Interventions	Group A (21) carbamazepine; Group B (16)  Chlormethiazole
Outcomes	Efficacy; Memory test (Benton); Zahlen‐Verbindungs test ;Beck Depression Inventory; Anxiety Sensitivity Index; Verbal Lern‐ und Merkfahigkeitstest; Mehrfach‐Wortschatz‐Intelligenz test (MWT‐B); European Addiction Severity Index (EuropASI); Clinical Institute Withdrawal Scale (CIWA‐A);
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	study described as randomised
Allocation concealment?	Unclear risk	study described as randomised
Blinding? subjective outcomes	High risk	study described as single blind COMMENT: personnel and outcome assessor not blinded
Blinding? Objective outcomes	Low risk	study described as single blind COMMENT: oucomes unlikely to be biased by lack of blindness
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no withdrawn from the study

Spies 1996.

Methods	Randomised controlled trial
Participants	No. = 159; Gender: 94% male; age range: 18‐83 years Inclusion criteria: multiple‐injured alcohol‐dependent patients who met the DSM‐III‐R criteria for alcohol dependence. Exclusion criteria: age < 18 years; chronic obstructive lung disease and poor pulmonary function or pneumonia; bradycardia (heart rate < 45/min); systolic blood pressure < 95 mmHg; second‐ or third‐degree atrioventricular node block; history of current use or abuse of benzodiazepines, barbiturates, clonidine, or beta‐adrenergic receptor blockers
Interventions	Group A (50) Clomethiazole/haloperidol, Group B (55) flunitrazepam/haloperidol, Group C (54) flunitrazepam/clonidine.
Outcomes	Efficacy: changes in CIWA‐Ar scores, Duration of controlled or assisted mechanical ventilation; Safety: mortality; Acceptability: dropouts, dropouts due to adverse events
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Patient was randomised to receive one of the three treatment"
Allocation concealment?	Unclear risk	"Patient was randomised to receive one of the three treatment"
Blinding? subjective outcomes	Unclear risk	Study reported as blinded. "The investigator who documented the alcoholism related history was unaware of the treatment and of the complication during the ICU stay" COMMENT: it is not clear if the patients and outcome assessor were blinded.
Blinding? Objective outcomes	Low risk	Study reported as blinded. "The investigator who documented the alcoholism related history was unaware of the treatment and of the complication during the ICU stay" COMMENT: outcomes unlikely to be biased by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	6 patients from the flunitrazepam/clonidine group, 10 patients from the Chlormethiazole/haloperidol group, 5 patients in the flunitrazepam/haloperidol group were excluded from the studies for medical reasons described in the study

Stanhope 1989.

Methods	Randomized controlled trial
Participants	No. = 304; Gender = 256 males; 48 females age = 40.5 years (mean age for carbamazepine); 39.9 years (mean age for placebo) Inclusion criteria Patients presenting for detoxification from alcohol Exclusion criteria: allergy to carbamazepine or Chlormethiazole reported before entry; known use of anticonvulsant before entry; previous repeated early self‐discharge; two admissions by Langton Centre terminated under 3 days duration; a previous admission already included in the study; alcohol not the principal drug of dependence.
Interventions	Group A (157)  carbamazepine; Group B (147) placebo
Outcomes	Efficacy: Daily withdrawal scores; seizures;Acceptability: Mode of discharge (completed detoxifications); Length of stay;Safety: adverse events;
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	High risk	"Alternate allocation was used rather than random allocation for logistic reason. The initial case was allocated randomly; and on five subsequent occasions, when the code was inadvertently broken, a fresh random start was made"
Allocation concealment?	High risk	"Alternate allocation was used rather than random allocation for logistic reason. The initial case was allocated randomly; and on five subsequent occasions, when the code was inadvertently broken, a fresh random start was made"
Blinding? subjective outcomes	Unclear risk	"Identical tablets were supplied by Ciba Geigy" COMMENT: it is not clear if personnel and outcome assessor were blind
Blinding? Objective outcomes	Low risk	"Identical tablets were supplied by Ciba Geigy" COMMENT: outcomes unlikely to be biased by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	7 patients withdrawn from the study. Reason for withdrawn given. Not specified form which group patients withdrawn, but numbers are small

Stuppaeck 1992.

Methods	Randomised controlled trial
Participants	No. = 60; Gender: 82% male Inclusion criteria: DSM‐III criteria for alcohol dependence; CIWA‐Ar score > 20 on admission; be able to sign informed consent. Exclusion criteria: age <18 or >65 years; severe somatic illness; polysubstance dependence; pre‐treatment with psychotropic drugs; full‐blown alcohol delirium
Interventions	Group A (29) carbamazepine, Group B (29) oxazepam Additional B‐polyvitamin compound was given orally.
Outcomes	Efficacy: seizures, delirium, changes in CIWA‐Ar scores; self‐rating adjective checklist; CGI scale; Safety: adverse events, mortality; Acceptability: dropouts, dropouts due to adverse events
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	"Patients were randomly allocated to a double blind study comparing CBZ abs oxazepam"
Allocation concealment?	Unclear risk	"Patients were randomly allocated to a double blind study comparing CBZ abs oxazepam"
Blinding? subjective outcomes	Low risk	Study defined as double blind. "Oxazepam and CBZ were administered in identical capsules". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Blinding? Objective outcomes	Low risk	Study defined as double blind. "Oxazepam and CBZ were administered in identical capsules". Blinding of outcomes assessor: it was not stated if the outcome assessor was blind, but because the outcomes were recorded during the treatment we judged that they were assessed by the blind personnel who gave the treatment
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"4 patients from each group dropped out from treatment." Reason for drop out given

Stuppaeck 1998.

Methods	Randomized controlled trial
Participants	No. = 38. Inclusion criteria: Inpatients met DSM‐IV criteria for alcohol dependence and a minimum CIWA ‐A score of 18.
Interventions	Group A (19) vigabatrin 2000 mg/day from day1‐3, then placebo from day4‐7; Group B (19) oxazepam 120 mg/day from day1‐3 and 90 mg/day from day 4‐
Outcomes	Efficacy: Modified German version of the Clinical Institute Withdrawal Assessment for Alcohol Scale (CIWA‐A); Clinical Global Impression Scale (CGI); Visual Analogue Scale (VAS); Fischer's Unwanted Symptoms Check List (FUSCL)
Notes	Meeting abstract
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	unclear
Allocation concealment?	Unclear risk	unclear
Blinding? subjective outcomes	Low risk	double blind
Blinding? Objective outcomes	Low risk	double blind
Incomplete outcome data addressed? All outcomes except withdrawal	Unclear risk	unclear

Teijeiro 1975.

Methods	Randomized Controlled Trial
Participants	No. = 55
Interventions	Group A (30)  atrium 300 mg; Group B (25)  Heminiurine
Outcomes	Efficacy: Tolerance; Tremor; Anxiety
Notes	Study in French
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	B Unclear
Allocation concealment?	Unclear risk	Unclear
Blinding? subjective outcomes	Unclear risk	Unclear
Blinding? Objective outcomes	Unclear risk	Unclear
Incomplete outcome data addressed? All outcomes except withdrawal	Unclear risk	Unclear

Thompson 1975.

Methods	Randomized controlled trial
Participants	No. = 34; Gender = 26 males/8 females Age = 22‐58 Inclusion criteria: Patients admitted consecutively to the Osler Medical Service with advanced delirium tremens
Interventions	Group A (17) paraldehyde 10 ml rectally every 30 minutes; Group B (17)  diazepam 10 mg intravenously initially and then mg intravenously every 5 minutes
Outcomes	Efficacy: Time for induction of a calm state; duration of delirium ; dose of diazepam and paraldehyde given after initial calming until patient was entirely recovered from delirium tremens; Safety: adverse events .
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"17 patients have delirium tremens not complicated by other serious illnesses; 17 patients had delirium tremens complicated by serious ilness. The two groups were allocated to therapy by means of separate series of sealed envelopes ordered from random number tables"
Allocation concealment?	Low risk	"17 patients have delirium tremens not complicated by other serious illnesses; 17 patients had delirium tremens complicated by serious ilness. The two groups were allocated to therapy by means of separate series of sealed envelopes ordered from random number tables"
Blinding? subjective outcomes	High risk	blindness not mentioned COMMENT: the study was judged not blind because the way of administration were different for two treatments
Blinding? Objective outcomes	Low risk	blindness not mentioned COMMENT: outcomes unlikely to be biased by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	no withdrawn from the study

Tubridy 1988.

Methods	Randomised controlled trial
Participants	No. = 102; Age range: 21‐65 years Inclusion criteria: alcohol dependent participants admitted to St John of God Hospital, Dublin. Exclusion criteria: patients sensitive to benzodiazepines or Chlormethiazole; dependent on or abused substances other than alcohol; psychotic or suffered from another mental disorder; serious physical illness; pregnant or lactating; taking other psychotropic drugs; history of withdrawal seizures
Interventions	Group A (49) Chlormethiazole, Group B (51) alprazolam All patients received daily intravenous injections of vitamin B complex and ascorbic acid on days 1‐5
Outcomes	Efficacy: seizures, global improvement, doctor's assessment of efficacy, patient's assessment of efficacy, Patient's general physical condition; presence of symptoms of alcohol withdrawal throughout the trial; HARS; doctor's and patient's global ratings of well‐being; changes in blood tests measures; Safety: adverse events, severe life‐treating adverse events, mortality; Acceptability: dropouts, dropouts due to adverse events
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"Patients were randomised to one or the two treatment group by consecutive assignment of patients identification numbers that had previously been distributed between the two treatments by a computer generated list of random numbers"
Allocation concealment?	Unclear risk	"Patients were randomised to one or the two treatment group by consecutive assignment of patients identification numbers that had previously been distributed between the two treatments by a computer generated list of random numbers"
Blinding? subjective outcomes	Unclear risk	blindness not mentioned
Blinding? Objective outcomes	Low risk	blindness not mentioned COMMENT: outcomes unlikely to be biased by lack of blinding
Incomplete outcome data addressed? All outcomes except withdrawal	Low risk	"Data from two patients in each group were lost"

ACL: Adjective Check List; ACS: Alcohol Craving Scale; ADS: Alcohol Dependence Scale; ALT: Alanine Aminotransferase Transaminase; ASI: Addiction Severity Index; ASQ: Abstinence Symptoms Questionnaire; AST: Aspartate Aminotransferase Transaminase; AWS: alcohol withdrawal symptoms; BAC: Blood Alcohol Concentration; BDHI: Buss‐Durkee Hostility Inventory BDI: Beck Depression Inventory; CGI: Clinical Global Impression; CIWA‐Ar: Rivised Clinical Institude Withdrawal Assessment for Alcohol; CLO: Chlormethiazole; CNS: Central Nervous System CPRS: Comprehensive Psychiatric Rating Scale; CSE: Clinical Side‐Effect Scale; DES: Daily Estimation Scale; DOTES: Dosage Record and Treatment Emergent Symptom Scale; DSM III: Diagnostic and Statistical Manual for Mental Disorders, third edition; DSM IV: Diagnostic and Statistical Manual for Mental Disorders, fourth edition; DSST: Digit Symbol Substitution Test; EuropASI: European Addiction Severity Index; FUSCL: Fischer's Unwanted Symptoms Check List; GGT: Gamma Glutamyl Transferase HARS: Hamilton Anxiety Rating Scale; HSCL: Hopkins Symptoms Check List; ICD‐10: Internationa Classification of Diseases; LCRR‐1: Lubeck Craving‐Risk‐Relapse questionnaire; MADRS: MontgomeryÅsberg Depression Rating Scale; MAO Inhibitors: Monoamine Oxidase inhibitors; MAWS: Mainz Alcohol Withdrawal Scale; MMSE: Mini‐Mental State Examination; MSSA: Modified Selective Severity Assessment; MWT‐B: Mehrfach‐Wortschatz‐Intelligenz test; NGI: Nurse's Global Impression; NOSIE: nurse's observation scale for inpatient evaluation; OCDS: Obsessive Compulsive Drinking Scale; PPT: Purdue Pegboard Test; PSQI: Pittsburgh Sleep Quality Index; RCT: randomised controlled trial; RDS: Reward Deficiency Syndrome REG: Rada‐Extensive Grid; SCL‐90‐R: Symptom Checklist 90 revised SGOT: Serum Glutamate Oxaloacetate Transaminase; SGPT: Serum Glutamic Pyruvic Transaminase; SIP: Short Index of Problems; SKT: Syndrom‐kurztest; SPS: Sleep Promoting Substance SSA: Selected Severity Assessment; STAI: State Anxiety Inventory; TLFB: Time Line Follow Back; TSA: Total Severity Assessment; VAS: visual analogue scale; VCS: Visual Craving Scale; VSS: Visual Success Scale; WHO: World Health Organization WSR: Withdrawal Symptom Rating;

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Addolorato 2002	Type of intervention not in the inclusion criteria
Adinoff 1994	Type of intervention not in the inclusion criteria
Aliyev 2008	Type of intervention not in the inclusion criteria
Ansoms 2000	Type of study design not in the inclusion criteria
Anton 2009	Outcome measures presented in a way not suitable for meta‐analysis
Athen 1976	Type of intervention not in the inclusion criteria
Athen 1977	Duplicate publication
Athen 1986	Study design not in the inclusion criteria
Balldin 1992	Study design not in the inclusion criteria
Baltieri 2003	Type of intervention not in the inclusion criteria
Banki 1983	Study design not in the inclusion criteria
Block 1974	Type of outcome measures not in the inclusion criteria
Bonnet 2009	Study design not in the inclusion criteria
Borg 1980	Study design not in the inclusion criteria
Brady 2002	Type of outcome measures not in the inclusion criteria
Bullock 1989	Type of intervention not in the inclusion criteria
Caspari 1992	Study design not in the inclusion criteria
Cazzato 1982	Study design not in the inclusion criteria
Chick 2000	Type of intervention not in the inclusion criteria
Cushman 1985	Type of intervention not in the inclusion criteria
Delamaire 1986	Type of outcome measures not in the inclusion criteria
Demmel 2000	Type of outcome measures not in the inclusion criteria
Dobrydnjov 2004	Type of intervention not in the inclusion criteria
Franz 2001	Study design not in the inclusion criteria
Frecker 1982	Study design not in the inclusion criteria
Gallant 1989	Study design not in the inclusion criteria
Gallant 1992	Study design not in the inclusion criteria
Glatt 1965	Duplicate publication
Gordon 2006	Type of intervention not in the inclusion criteria
Gottesleben 1995	Type of intervention not in the inclusion criteria
Griffith 1986	Type of outcome measures not in the inclusion criteria
Gual 2001	Type of intervention not in the inclusion criteria
Gual 2002	Type of intervention not in the inclusion criteria
Guarino 1989	Type of outcome measures not in the inclusion criteria
Hague 1976	Type of intervention not in the inclusion criteria
Harfst 1967	Study design not in the inclusion criteria
Heil 1990	Type of outcome measures not in the inclusion criteria
Hillbom 1989	Type of outcome measures not in the inclusion criteria
Hutchison 2001	Type of intervention not in the inclusion criteria
Ilyuchina 1995	Study design not in the inclusion criteria
Janiri 1996	Type of intervention not in the inclusion criteria
Johnson 2003	Type of outcome measures not in the inclusion criteria
Johnson 2004	Type of outcome measures not in the inclusion criteria
Kaim 1974	Study design not in the inclusion criteria
Kaplan 1972	Study design not in the inclusion criteria
Karam‐Hage 2003	Type of outcome measures not in the inclusion criteria
Karst 2002	Type of intervention not in the inclusion criteria
Keaney 2001	Type of intervention not in the inclusion criteria
Kiefer 2003	Type of intervention not in the inclusion criteria
Kramp 1979	Type of outcome measures not in the inclusion criteria
Kranzler 1996	Type of outcome measures not in the inclusion criteria
Kraus 1985	Type of intervention not in the inclusion criteria
Lange‐Ass 2003	Study design not in the inclusion criteria
Le Bon 2003	Type of intervention not in the inclusion criteria
Le Fauve 2004	Type of intervention not in the inclusion criteria
Leivonen 1966	Study design not in the inclusion criteria
Loo 1986	Type of outcome measures not in the inclusion criteria
Malcolm 1992	Type of intervention not in the inclusion criteria
Martinotti 2007	Type of intervention not in the inclusion criteria
Mendels 1985	Type of intervention not in the inclusion criteria
Miller 1988	Type of outcome measures not in the inclusion criteria
Montejo 2007	Study design not in the inclusion criteria
Mueller 1997	Type of intervention not in the inclusion criteria
Mukherjee 1983	Type of intervention not in the inclusion criteria
Muller 1969	Study design not in the inclusion criteria
Myrick 2009	Outcome measures presented in a way not suitable for meta‐analysis
Naranjo 2000	Type of intervention not in the inclusion criteria
Newman 1995	Study design not in the inclusion criteria
Pettinati 2000	Type of intervention not in the inclusion criteria
Ponce 2005	Study design not in the inclusion criteria
Rodgers 1999	Study design not in the inclusion criteria
Rubio 2002	Type of intervention not in the inclusion criteria
Rychlik 2001	Type of outcome measures not in the inclusion criteria
Sass 1996	Type of intervention not in the inclusion criteria
Schmidt 1994	Type of outcome measures not in the inclusion criteria
Schulte 1987	Study design not in the inclusion criteria
Schwarz 1982	Type of outcome measures not in the inclusion criteria
See 2002	Duplicate publication
Shaw 1994	Type of intervention not in the inclusion criteria
Sillanpaa 1979	Duplicate publication
Smith 1987	Type of intervention not in the inclusion criteria
Snel 1983	Type of intervention not in the inclusion criteria
Sobcyzk 1980	Study design not in the inclusion criteria
Solomon 1983	Type of intervention not in the inclusion criteria
Spies 1995	Type of outcome measures not in the inclusion criteria
Sternebring 1983	Study design not in the inclusion criteria
Stojek 1987	Study design not in the inclusion criteria
Streeton 2001	Study design not in the inclusion criteria
Swift 2003	Study design not in the inclusion criteria
Tacke 1975	Type of intervention not in the inclusion criteria
Teijeiro 1976	Type of outcome measures not in the inclusion criteria
Tempesta 2000	Type of intervention not in the inclusion criteria
Thomas 1964	Study design not in the inclusion criteria
Tress 1977	Study design not in the inclusion criteria
Wadstein 1986	Duplicate publication
Wegner 1965	Study design not in the inclusion criteria
Welbel 1982	Study design not in the inclusion criteria
Whitworth 1996	Type of intervention not in the inclusion criteria
Wiesbeck 1999	Type of intervention not in the inclusion criteria

AW: alcohol withdrawal 
 CCT: controlled clinical trial 
 RCT: randomized controlled trial

Characteristics of studies awaiting assessment [ordered by study ID]

Gottesleben 1994.

Methods	RCT
Participants	28 patients during drug therapy for alcoholic pre delirium
Interventions	slow‐release carbamazepine (Timonil registered trade mark 150 retard) versus Chlomethiazole
Outcomes	duration and severity of withdrawal symptoms.
Notes	we are trying to find the study

Hart 1964.

Methods	it is impossible from title to know the study design
Participants	175 of alcohol withdrawal
Interventions	promazine and paraldehyde
Outcomes	it is impossible from title to know which outcomes were considered
Notes	we are trying to find the study

Trevisan 2008.

Methods	RCT
Participants	Fifty‐seven male veterans presenting for alcohol detoxification
Interventions	valproic acid, gabapentin, or placebo for a total of 4 weeks.
Outcomes	symptoms of physiologic withdrawal during the first 5 days of treatment and doses of benzodiazepine required;
Notes	we are trying to find the study

Contributions of authors

Vecchi performed the literature searches; Mitrova organised papers collection; Minozzi and Amato reviewed the papers, abstracted data from the papers for meta‐analysis. Amato and Minozzi wrote abstract, introduction, discussion and conclusions sections. Minozzi wrote methods , assessed methodological quality of included studies and wrote results sections. Davoli supervised to all the process and all authors provided comments to the final version.

Sources of support

Internal sources

• No sources of support supplied

External sources

• EDAP Project (Evidence for Drugs and Alcohol Policy) sponsored by the European Community‐ Directorate Public Health (Grant Agreement SPC.2002454), Not specified.

Declarations of interest

None.

New search for studies and content updated (conclusions changed)