gabapentin versus chlordiazepoxide for outpatient alcohol ... · going treatment for alcohol...

Benzodiazepines have been the drugsof preference for managing alcohol

withdrawal (AW) in the US.1-4 Severalbenzodiazepines carry Food and DrugAdministration–approved indications forthe treatment of AW. While these agentsare considered the gold standard for thetreatment of AW, several adverse effectsshould be considered when benzodi-azepines are used for outpatient detoxifica-tion. These adverse effects of benzodi-azepines include potential abuse anddepen dence, memory impairment, confu-sion, sedation, and ataxia, as well as the in-teraction with alcohol.1,4

Although the effectiveness and safetyof treating AW with benzodiazepineshave been established,2,4 there is ongoinginterest in identifying treatments for AWthat are nonsedating, lack psychomotorimpairment, and are not likely to ad-versely affect patients’ transition to on-going treatment for alcohol dependence.

The Annals of Pharmacotherapy ■ 2013 July/August, Volume 47 ■ 961

Gabapentin Versus Chlordiazepoxide for Outpatient Alcohol

Detoxification Treatment

Christopher J Stock, Lindsay Carpenter, Jian Ying, Tom Greene

theannals.com

Drug, Alcohol, and Substance Abuse

Author information provided at end of text.© 1967-2013 Harvey Whitney Books Co. Allrights reserved. No part of this document maybe reproduced or transmitted in any form or byany means without prior written permission ofHarvey Whitney Books Co. For reprints of anyarticle appearing in The Annals, please [email protected]

BACKGROUND: Benzodiazepines are used to treat alcohol withdrawal (AW) butcause cognitive impairment, sedation, and ataxia, and interact with alcohol. Non ben -zo diazepine anticonvulsants are promising and possibly safer alternatives for thetreatment of AW. OBJECTIVE: To compare follow-up measures of Epworth Sleepiness Scale (ESS),Penn Alcohol Craving Scale (PACS), ataxia rating, and Clinical InstituteWithdrawal Assessment for Alcohol revised (CIWA-Ar) symptoms betweenalcohol-dependent individuals randomized to treatment with gabapentin orchlordiazepoxide.METHODS: A randomized, double-blind study was conducted in US veterans withalco hol withdrawal (DSM-IV criteria). Subjects requiring hospitalization or taking ben -zo diazepines or nonbenzodiazepine anticonvulsants were excluded. Twenty-sixparti cipants were randomized: 17 received gabapentin and 9 received chlor di aze -pox ide. Gabapentin doses were 1200 mg orally for 3 days, followed by 900 mg, 600mg, and 300 mg for 1 day each. Chlordiazepoxide doses were 100 mg orally for 3days, followed by 75 mg, 50 mg, and 25 mg for 1 day each. CIWA-Ar, ESS, PACSscales and evaluation for ataxia were administered daily. RESULTS: Follow-up mean ESS and PACS scores did not differ significantlybetween treatment groups in the early treatment period (days 1-4) but were lower(mean difference –3.70; 95% CI –7.21 to –0.19; p = 0.04) and (mean difference–6.05; 95% CI –12.82 to 0.72; p = 0.08), respectively, at the end of the treatmentperiod (days 5-7) in gabapentin-treated subjects. CIWA-Ar scores were reducedsimilarly in both groups. Ataxia was not observed. No significant adverse eventswere noted. Limitations include our small sample size and 35% loss to follow-upat the end of the treatment period. CONCLUSIONS: In ambulatory veterans with symptoms of alcohol withdrawal,gabapentin treatment resulted in significantly greater reduction in sedation (ESS)and a trend to reduced alcohol craving (PACS) by the end of treatment comparedto chlordiazepoxide treatment. Although limited by the small sample size, thesuggestion of reduction in sleepiness and less craving warrants replication of thestudy with a larger sample. Ann Pharmacother 2013;47:961-9.Published Online, 18 Jun 2013, theannals.com, doi: 10.1345/aph.1R751

The nonbenzodiazepine anticonvulsants carbamazepine,divalproex, and gabapentin have shown promise for treatingAW. These medications have been found to reduce the inten-sity of AW symptoms during a 5- to 10-day study period.2,5-9Gabapentin has been evaluated for AW in open as well asblinded studies of AW for inpatients and outpatients.10-13Gabapentin has been as effective as lorazepam or phenobar-bital in reducing withdrawal symptoms over time in patientsexhibiting mild to moderately severe symptoms of AW.12-14Myrick et al. demonstrated that gabapentin and lorazepam re-duced withdrawal symptoms during 4 days of active treat-ment.10 Although not their primary aim, they also noted thatsubjects treated with gabapentin had reduced probability ofdrinking and experienced less sedation during the activetreatment phase compared to lorazepam-treated subjects.

Anticonvulsants do not produce dependence and, com-pared with benzodiazepines, may not be as likely to cause se-dation to the same degree or amnesia.3,5 Conversely, dival-proex use in AW may be limited by other adverse effects suchas gastrointestinal disturbances, confusion, and tremor.8Dizzi-ness, pruritus, vomiting, and nausea are common adverse ef-fects of carbamazepine. Divalproex and carbamazepine use islimited in patients with severe hepatic and hematologic dis-eases.8 In addition, carbamazepine is problematic in patientsreceiving multiple medications because of its ability to inducehepatic enzymes.5 Gabapentin may offer benefit with im-proved tolerability and is not prone to drug interactions seenoften with carbamazepine and divalproex. The most commonadverse effects are sedation and dizziness, but these effectsgenerally do not lead to discontinuation. Gabapentin mayhave a potential role in AW treatment because of its lack ofsignificant drug interactions, lack of hepatic metabolism, andlack of the need for therapeutic monitoring.2,8

Previously published trials with gabapentin have pri-marily focused on reduction of withdrawal symptoms(Clinical Institute Withdrawal Assessment for Alcohol re-vised [CIWA-Ar] scores), and any comparison of cravingor safety aspects (eg, sleepiness and ataxia) was secondary.This study was undertaken to compare sedation, craving,and evidence of ataxia among individuals receivinggabapentin or chlordiazepoxide in fixed doses for ambula-tory treatment of mild to moderate AW.

The aims of the study were to compare follow-up mea-sures of Epworth Daytime Sleepiness Scale (ESS),15 PennAlcohol Craving Scale (PACS),16 and CIWA-Ar17 scores, andincidence of ataxia between alcohol-dependent outpatientsrandomized to either a fixed-dose tapering regimen ofgabapentin or chlordiazepoxide for treatment of mild to mod-erate AW.

MethodsA randomized, double-blind, double-dummy 7-day trial

comparing gabapentin to chlordiazepoxide was performed.

The study was reviewed and approved by the institutionalreview boards at the University of Utah and the VeteransAffairs (VA) Salt Lake City Health Care System. Thestudy was conducted under IND 66,492 issued by the Foodand Drug Administration and registered as NCT01573052on ClinialTrials.gov.

PARTICIPANTS

All patients referred to the George E Wahlen VA Medi-cal Center’s Mental Health Outpatient Detoxification Clin-ic were screened for participation in this study. Any veter-an meeting DSM-IV criteria for AW (2 or more of the fol-lowing criteria: sweating, pulse >100 beats/min, handtremor, insomnia, nausea or vomiting, sensory disturbance,agitation, anxiety, seizure) was invited to participate in thisstudy. Individuals were excluded if they had any acutelyunstable medical or major Axis I psychiatric conditions re-quiring emergency management or urgent hospitalization;comorbid current benzodiazepine, opioid, or barbiturateabuse/dependence; current/active seizure disorder requir-ing chronic anticonvulsant drug therapy; or were receivingcurrent β-blocker, clonidine, gabapentin, benzodiazepines,or other drug therapy likely to affect alcohol withdrawalsymptoms. Signed informed consent was obtained prior toany study procedures and after a complete description ofthe study was given to the subjects.

TREATMENT/INTERVENTIONS

Chlordiazepoxide 25 mg and matching placebo capsuleswere obtained from the University of California, San Fran-cisco Drug Product Services Laboratory, Department ofClinical Pharmacy. Gabapentin 300 mg and matchingplacebo capsules were provided by Pfizer Pharmaceuticals.The chlordiazepoxide/placebo and gabapentin/placebocapsules were not identical in appearance. Study medica-tions were packaged into a 7-day medication organizer.The dosing for each subject was either gabapentin 1200mg or chlordiazepoxide 100 mg orally days 1-3,gabapentin 900 mg or chlordiazepoxide 75 mg day 4,gabapentin 600 mg or chlordiazepoxide 50 mg day 5, andgabapentin 300 mg or chlordiazepoxide 25 mg day 6.

Once enrolled, subjects were randomly assigned tochlordiazepoxide or gabapentin treatment and received theassigned study medication box and dosing instructions.Each dose consisted of 1 capsule containing the activemedication according to randomized treatment (eithergabapentin 300 mg or chlordiazepoxide 25 mg) and aplacebo capsule that matched the nonrandomized medica-tion. Patients were instructed to take both nonidentical cap-sules for each prescribed dose.

Adherence was assessed by pill counts and serum sam-ples were obtained for study drug analysis. Serum samples

962 ■ The Annals of Pharmacotherapy ■ 2013 July/August, Volume 47 theannals.com

CJ Stock et al.

were batched and sent to a contract laboratory so that studyclinicians were blinded to the results until after the trialwas completed. Presence of the assigned study medicationin the blood served as a surrogate adherence marker.

Subjects also received prescriptions for daily oral thera-peutic multiple vitamin tablets, folic acid 1 mg, and thi-amine 100 mg.

ASSESSMENTS

All clinic appointments and study assessments for AWtreatment were conducted in the Mental Health OutpatientDetoxification Clinic.

A focused medical and psychiatric history and physicalexamination, breath alcohol concentration, vital signs,CIWA-Ar, and blood chemistry including transaminasesand complete blood cell count were obtained for baselinefor all participants. Subjects were seen on weekdays forclinic follow-up appointments.

Assessments performed at each study visit includedCIWA-Ar, ESS, PACS, and a brief examination to docu-ment mental status and assess coordination, which includ-ed assessment of stance, tandem gait, Romberg test, prona-tor drift, toe and heel walk, rapid-alternating-movements,and point-to-point movements. Breath alcohol concentra-tion (BAC) was also measured (AlcoSensor, Intoximeters,Inc.) at each visit.

Subjects returned on day 7 for final assessments and re-turn of the medication box. Following successful comple-tion of treatment for AW, participants were referred for on-going care in the VA’s substance abuse treatment program.

PROTOCOL MODIFICATIONS, POWER CALCULATIONS, ANDSTATISTICAL PLAN

When initially approved by the IRB, the study was de-signed to evaluate group differences in a composite endpoint that collapsed ESS, PACS, ataxia, and cognitionscores into discrete categories. The required sample sizewith this categorical outcome was 104 subjects. Followinginitial study approval, but during the period of active sub-ject enrollment, significant changes in the VA’s emergencydepartment (ED) and mental health department patienttriage processes and changes in the referral base limited re-cruitment to 26 subjects. As a result, before unlocking ourexisting dataset and examining the outcome data, wereevaluated the protocol end points and statistical analysisplan. The power analysis and statistical plan were revisedas described below to estimate the treatment effects on theESS and PACS as continuous variables and approval ofthese revisions was received from the IRB before unlock-ing and unblinding the database. Support for the revisedanalysis plan and the conduct of the data analyses was pro-vided by the Study Design and Biostatistics Center of the

Center for Clinical and Translational Science at the Uni-versity of Utah.

Previous research with the ESS15 reported a medium ef-fect size difference (d = 0.52 of 1 SD) between normal andmoderately impaired groups and a larger effect size differ-ence (d = 1.05) between severely impaired and moderatelyimpaired groups. A revised sample size of 26 subjects withcomplete data would provide an 80% power (with a 2-sidedα = 0.05) to detect a mean difference in each quantitativeoutcome equal to d = 0.96 SDs given a correlation of at least0.60 between the baseline and follow-up measurements; theminimal detectable effect size is increased to d = 1.08 afteraccounting for a 15% dropout rate during the first 4 days,and up to d = 1.28 after accounting for the 35% drop out rateat the end of the treatment period (Figure 1).

Unpooled 2-sample t-tests and Fisher exact tests wereused as appropriate to compare baseline subject character-istics between the 2 treatment groups. Separate mixed-ef-

Gabapentin vs Chlordiazepoxide for Outpatient Alcohol Detoxification

The Annals of Pharmacotherapy ■ 2013 July/August, Volume 47 ■ 963theannals.com

Figure 1. Consolidated Standards for Reporting of Trials study flow di-agram. CIWA-Ar = Clinical Institute Withdrawal Assessment for Alcoholrevised.

fects models were used to analyze group differences onmean values during the early stage (days 1-4) and latestage (days 5-7) of the treatment period for ESS, PACS,and CIWA-Ar while adjusting for initial baseline (day 0)values for the outcome being analyzed. The ESS andPACS scores were designated as the coprimary outcomevariables. Each follow-up measurement was given equalweight when constructing contrasts to compare the meanfollow-up scores during each of the 2 follow-up periodsbetween the randomized treatment groups.

The statistical analyses were performed using an intent-to-treat strategy in which each subject is analyzed accord-ing to his or her randomized treatment assignment, irre-spective of adherence to the assigned intervention. All 26participants were included in the mixed-effects analysis forthe study outcomes by retaining all available observationsprior to the time of loss to follow-up for subjects who with-drew early from the trial.

All hypothesis tests were performed with a comparison-wise 2-sided α level of 5%, without adjustment for multi-ple comparisons and are reported as mean differences be-tween groups with 95% CI and p values.

ResultsSUBJECTS

Four hundred twenty-nine patients were referred to theVA Salt Lake City Health Care System’s Mental HealthOutpatient Detoxification clinic from 2004 to 2009. Twen-ty-six veteran alcohol-dependent individuals (1 female and25 male) were enrolled and randomized to the 2 treatmentarms. There were 17 subjects randomized tothe gabapentin group and 9 subjects to thechlordiazepoxide group. Clinic patients whowere not being seen for alcohol dependence,alcohol-dependent patients with minimal with-drawal symptoms, as well as those requiringemergency treatment or hospitalization due tounstable medical or psychiatric conditions orwho were actively receiving exclusionarymedications (see Methods section) were ex-cluded from participation. Reasons for exclu-sion of these 403 clinic patients are outlined inFigure 1.

Baseline Characteristics

Gabapentin subjects were slightly youngeron average. A higher percentage of chlordi -azepoxide subjects were prescribed psycho -tropics (primarily selective serotonin reuptakeinhibitor antidepressants and/or trazo done) butthis difference was not significant. The groupsdid not differ significantly in age of reported

first drink, years of alcohol abuse, or aspartate aminotrans-ferase serum concentration. BAC readings at baseline av-eraged 0.071% alcohol (range 0-0.30%) and were not sig-nificantly different between the gabapentin and chlor-diazepoxide groups (p = 0.78). BACs dropped to zero inall subjects by the day 1 visit. A similar percentage of par-ticipants in both groups had comorbid drug dependence.Baseline characteristics are outlined in Table 1.

Dropouts and Noncompleters

Eleven (65%) subjects in the gabapentin group and 6(67%) in the chlordiazepoxide group completed the study.The median number of days of clinic follow-up was 5.Noncompleters included 3 gabapentin subjects who expe-rienced worsening withdrawal (including 1 who reportedan unwitnessed seizure) and were sent to the ED for evalu-ation, and 3 gabapentin subjects who were clinically im-proved on day 4 but did not return for follow-up appoint-ments on days 5-7. The 3 noncompleters in the chlor-diazepoxide group included 1 subject who experiencedworsening withdrawal (including a minor fall) and wassent to ED for evaluation and 2 chlordiazepoxide subjectswho were lost to follow-up after randomization withoutcompleting additional visits (Figure 1).

OUTCOMES

Baseline scores on the ESS were well balanced betweenthe 2 treatment groups (mean baseline scores were 6.2 inthe gabapentin group and 4.7 in the chlordiazepoxidegroup, p = 0.47) (Table 1), representing a low chance of


CJ Stock et al.

Table 1. Baseline CharacteristicsGabapentin Chlordiazepoxide

Characteristic (n = 17) (n = 9) p Value

Age, mean (SD) 51.1 (7.7) 55.8 (4.8) 0.11aAge at first alcohol use, years, 15.6 (2.8) 17.8 (2.6) 0.10amean (SD)

Years of alcohol use, mean (SD) 27.5 (7.4) 28.1 (5.9) 0.85aBaseline BAC, %, mean (SD) 0.08 (0.1) 0.06 (0.2) 0.78aComorbid drug dependence, n (%) 4 (24) 2 (22) 1.0bConcomitant psychotropic 12 (71) 9 (100) 0.13bmedication, n (%)

Baseline AST, IU/L, mean (SD)c 78.6 (59.2) 67.3 (53.0) 0.68aBaseline ESS, mean (SD) 6.2 (6.20) 4.7 (4.2) 0.47aBaseline PACS, mean (SD) 30.5 (5.0) 30.0 (5.8) 0.85aBaseline CIWA-Ar, mean (SD) 7.7 (3.1) 8.8 (6.3) 0.64a

AST = aspartate aminotransferase; BAC = breath alcohol concentration; CIWA-Ar= Clinical Institute Withdrawal Assessment for Alcohol revised; ESS = EpworthSleepiness Scale; PACS = Penn Alcohol Craving Scale.aStudent t-test.bFisher exact test.cValues were missing for 3 gabapentin and 3 chlordiazepoxide subjects.

dozing, but ranged from 0 (no chance of dozing) to 24(high chance of dozing). Baseline ESS scores did not cor-relate significantly with baseline BACs, that is, high BACswere not systematically associated with high ESS scores(Pearson R = 0.30, p = 0.14). The adjusted mean ESSscore did not differ significantly between the randomizedgroups during the early stage (adjusted mean ESS 0.66higher in the gabapentin group vs the chlordiazepoxidegroup; 95% CI –1.93 to 3.26; p = 0.61). During the latestage, however, the adjusted mean ESS score was signifi-cantly lower with gabapentin compared to chlordiazepox-ide (mean difference –3.70; 95% CI –7.21 to –0.19; p =0.04) as shown in Figure 2 and in Table 2.

Baseline scores on the PACS were also well balanced be-tween the 2 groups (p = 0.85). The average baseline scoreswere 30.0 (gabapentin) and 30.5 (chlordiazepoxide) (strongcraving) and ranged from 17 to 35 (moderate to unbearablecraving) (Table 1). The adjusted mean PACS score did notdiffer significantly between the treatment groups during the

early stage period (1.39 lower in the gabapentin group; 95%CI –6.48 to 3.70; p = 0.59). During the late stage, there wasa trend toward a lower adjusted mean PACS score in thegabapentin group compared to the chlordiazepoxide group(mean difference = 6.05; 95% CI –12.82 to 0.72; p = 0.08),as shown in Figure 3 and Table 2.

No participant in either group had evidence of ataxiaduring follow-up appointments; however, one chlor-diazepoxide subject reported an unwitnessed fall at homewith no residual injury. There were no clinically significantalterations in blood pressure or heart rate.

Scores on the CIWA-Ar averaged 7.7 (gabapentin) and8.8 (chlordiazepoxide) at baseline (range 0-20) and werewell balanced between the treatment groups (p = 0.64)(Table 1). Scores in both groups declined during the studyperiod, showing gradual reduction in withdrawal symptomseverity through the final day. Our mixed-model analysisshowed no significant difference in adjusted follow-upmean scores between groups during either early or latestages (Table 2).

All subjects reported taking study medications and only1 of the 14 serum samples (in an individual assigned togabapentin) was negative for the expected drug (96% med-ication adherence rate).

DiscussionDuring this 7-day trial, subjects who received a fixed

dose tapering regimen of gabapentin to treat mild to mod-erate alcohol withdrawal experienced significantly reduceddaytime sleepiness and a nonsignificant trend toward re-duction in alcohol craving by the end of the treatment peri-od compared to subjects who received chlordiazepoxide.Participants who completed treatment had similar reduc-tions in withdrawal symptoms over the treatment periodregardless of treatment group assignment and between-group differences were nonsignificant. Dropout rates andreasons for dropout were similar in both treatment groups.



Table 2. Results by Early- and Late-Stage Treatment Perioda

Early Stage (days 1-4) Late Stage (days 5-7)Adjusted Mean (95% CI) Difference Adjusted Mean (95% CI) Difference

Gabapentin Chlordiazepoxide (95% CI) Gabapentin Chlordiazepoxide (95% CI)Scale (n = 15) (n = 7) p Value (n = 12) (n = 5) p Value

ESS 5.97 (4.08 to 7.86) 5.31 (2.76 to 7.86) 0.66 (–1.93 to 3.26) 2.65 (0.44 to 4.85) 6.35 (3.13 to 9.57) –3.70 (–7.21 to –0.19) p = 0.61 p = 0.04

PACS 15.66 (12.74 to 18.58) 17.05 (12.66 to 21.45) –1.39 (–6.48 to 3.70) 10.74 (7.09 to 14.39) 16.78 (10.92 to 22.64) –6.05 (–12.82 to 0.72) p = 0.59 p = 0.08

CIWA-Ar 5.86 (4.19 to 7.53) 6.64 (4.04 to 9.24) –0.77 (–3.71 to 2.16) 4.33 (2.10 to 6.56) 3.29 (–0.28 to 6.88) 1.03 (–3.15 to 5.22) p = 0.60 p = 0.62

CIWA-Ar = Clinical Institute Withdrawal Assessment for Alcohol revised scale; ESS = Epworth Sleepiness Scale; PACS = Penn Alcohol Craving Scale.aThe adjusted means and mean differences were estimated using a mixed-effects model, controlling for the baseline level of the outcome being analyzed.

Figure 2. Adjusted mean (SE) ESS scores under a mixed-effects mod-el, controlling for baseline ESS scores. Treatment phase 0 = baseline, 1= early stage (days 1-4), and 2 = late stage (days 5-7). CDP = chlor-diazepoxide; ESS = Epworth Sleepiness Scale; GB = gabapentin.

These data add evidence to support use of gabapentin asan alternative to benzodiazepines for use in outpatienttreatment of mild to moderate AW. Gabapentin has beenrecommended as a good alternative to benzodiazepinesdue to speculation that it may dampen kindling (neuralsensitization leading to progressive worsening of subse-quent withdrawal episodes, particularly seizures),8,10,18-20 issafe, is less likely to interact with or increase sedation if al-cohol is consumed,21 and, as in the present study, may be aseffective as benzodiazepines in treating the symptoms ofalcohol withdrawal.12,14,22

Additionally, patients might resume drinking if with-drawal symptoms are not adequately managed or if themedication they take does not suppress the craving for al-cohol use or actually aggravates the craving. Preclinicaland clinical studies suggest that benzodiazepine use mayincrease craving and early relapse to alcohol use23 while, asin this study and others,10,24,25 gabapentin use does not trig-ger alcohol craving.

Because of the risk that patients may resume drinking al-cohol in an unmonitored outpatient or home setting, it is im-perative to choose a medication that, unlike benzodiazepines,does not increase sedation. Subjects treated with gabapentinin this study reported less likelihood of dozing than did chlor-diazepoxide-treated subjects. Myrick et al.21 also found thatgabapentin did not increase daytime sleepiness, and therewas no significant difference between placebo and gaba pen -tin, even in subjects who continued to drink alcohol.

The lack of differences in ESS and PACS scores duringthe early stage of treatment but the presence of a differencein ESS scores and a trend toward difference in PACSscores in the late stage of withdrawal treatment are consis-tent with the pharmacokinetic and pharmacodynamic pro-file of the study medications. Initial effects of chlor-

diazepoxide and continued accumulation of chlordiazepox-ide metabolites even after the dose is reduced would be ex-pected to lead to continued or even worsening sleepinessduring the study period. General lack of sedation and theshorter half-life of gabapentin, along with a lack of activemetabolites to accumulate, would be expected to result inless sleepiness. Myrick et al.10 also found that compared tolorazepam (a shorter half-life than chlordiazepoxide and noactive metabolites), gabapentin-treated subjects experi-enced less alcohol craving and sleepiness.

One of the major limitations of this study was the inabil-ity to achieve the sample size stipulated in the originalpower calculations. However, after reassignment of thestudy outcome measures and statistical analysis plan, theachieved samples size of 26 subjects was sufficient to de-tect a substantial treatment effect on ESS and PACS scores.Although there was a trend toward treatment effect on thePACS score, the difference seen in this small sample didnot reach significance. Furthermore, this smaller samplesize did not take into account the 33-35% noncompletionrate in the study groups. We had planned to recruit more el-igible individuals from the VA’s Mental Health Detoxifica-tion Clinic; however, during the recruitment period, unan-ticipated circumstances limited the pool of eligible veter-ans and we were unable to enroll more subjects to mitigatethe data lost from noncompleters. Different results mayhave been seen with a larger sample.

The difference in sample size between the treatmentgroups, which resulted because the randomization schemedid not invoke a permuted block design, limited the statis-tical power of the trial. However, randomization of morepatients to one or the other treatment group did not lead tobias in estimates of treatment effects with regard to differ-ences seen in the ESS and PACS scores. However, unequalrandomization does reduce the precision of treatment effectestimates compared to equal randomization.

Other investigators14,26 have suggested that selected sub-types of alcohol-dependent subjects may have different re-sponse to gabapentin or benzodiazepines and a larger samplesize would be desirable to establish whether these individualdifferences exist. Unfortunately, neither our study nor the oth-ers14,26 (with 26, 60, and 37 subjects, respectively) had ade-quate power to characterize response in subtypes of subjects.

There is a risk that the apparent differences betweentreatment groups during the late-stage treatment periodcould have resulted from loss to follow-up before thattime. The use of a mixed model for statistical analysis wasdesigned to minimize the risk of bias for loss to follow-upby accounting for baseline and the early-stage values evenif the subject missed the late-stage value, but that bias maystill have resulted if those data were not missing at random.

The average CIWA-Ar score at randomization in oursample was 8, lower than the score often described in guide-lines as the threshold for medication initiation.2,17,27,28 These


CJ Stock et al.

Figure 3. Adjusted mean (SE) PACS scores under a mixed-effects mod-el, controlling for baseline PACS scores. Treatment phase 0 = baseline,1 = early stage (days 1-4), and 2 = late stage (days 5-7). CDP = chlor-diazepoxide; GB = gabapentin: PACS = Penn Alcohol Craving Scale.

guidelines are targeted at treatment of hospitalized patientswho have more frequent contact, monitoring, and reassur-ance from clinical staff. All participants in the present outpa-tient study met DSM-IV criteria for alcohol withdrawal syn-drome, and reduction in CIWA-Ar was not the primary out-come measured in this study. The normal course of alcoholwithdrawal is gradual reduction of withdrawal symptomsover 72-96 hours, with and without medications to inter-vene.2,7,17 Schacht et al.27 used a CIWA-Ar threshold of 7,lower than the average in this study, for treatment assign-ments and stated that this score threshold is more typical foroutpatient detoxification treatment settings.

The drugs compared in this trial differ significantly intheir metabolic and pharmacokinetic profiles. Therefore,dose equivalence cannot be established. Chlordiazepoxidewas chosen as the control medication as it represents the lo-cal standard of care. The dose of chlordiazepoxide used inthis study may be lower than what is typically recommendedfor inpatient detoxification. The goals for outpatient detoxifi-cation, however, include avoidance of sedation and mini-mizing the risk of falls. Guidelines and recommendations foroutpatient treatment recommend initial chlordiazepoxidedoses of 30-100 mg/day.28-31 Doses of chlordiazepoxide usedin this study were well within this recommended range. Ini-tial doses of gabapentin used in other studies10,11,14 rangedfrom 900 to 3200 mg/day, with doses tapered over 4-9 days.Myrick et al.10 showed that a gabapentin starting dose of1200 mg/day is more effective than lower doses in reducingCIWA-Ar scores in an outpatient setting.

Finally, clinical consequences of ongoing or worseningsedation may include diminishment of patients’ functionalabilities, falls, and failure to follow through with referral totreatment. These real-life complications may result fromongoing withdrawal symptoms, from sedation due to med-ications used to treat withdrawal symptoms, or other fac-tors, and the incidence of these events is relatively low.The sample size needed to determine whether the rate ofsuch events differs between treatment arms was probablynot achieved in our study. Other studies10,11,14 also have notincluded these consequences as primary outcomes, andthose samples were likely too small to have done so.

In this randomized trial, gabapentin treatment resulted insignificantly less daytime sleepiness and a nonsignificanttrend toward less alcohol craving compared to chlor-diazepoxide treatment. As expected, CIWA-Ar scores weregradually reduced in both chlordiazepoxide and gabapentingroups, showing no difference favoring either drug. In ad-dition, no subjects had evidence of ataxia during medica-tion exposure. Alcohol withdrawal complications occurredat similar rates in both groups with one participant experi-encing a seizure.

Overall, our study supports gabapentin as a tolerablemedication that is less sedating, and may lead to less crav-ing in outpatients treated for alcohol withdrawal compared

to chlordiazepoxide. Given the small sample, there is notenough information to generalize these results or furthercharacterize likely responders versus nonresponders. Fur-ther trials with larger samples are warranted to establishgabapentin as a safer alternative to chlordiazepoxide andother benzodiazepines.

Christopher J Stock PharmD, Clinical Pharmacy Specialist, Men-tal Health Care Center (116), Veterans Affairs Salt Lake City HealthCare System, Salt Lake City, UT Lindsay Carpenter BS, Research Program Specialist, MentalHealth Care Center (116), Veterans Affairs Salt Lake City HealthCare SystemJian Ying PhD, Statistician, Department of Internal Medicine, Schoolof Medicine, University of Utah, Salt Lake CityTom Greene PhD, Professor, Department of Internal Medicine,School of Medicine, University of Utah Correspondence: Dr. Stock, [email protected]/Online Access:www.theannals.com/cgi/reprint/aph.1R751Conflict of Interest: Authors reported none Acknowledgments: Dr. Stock had full access to all of the data inthe study and takes responsibility for the integrity of the data and theaccuracy of the data analysis. This material is the result of work sup-ported with resources and the use of facilities at the Salt Lake CityVeterans Affairs Health Care System. The views expressed in thisarticle are those of the authors and do not necessarily reflect the po-sition or policy of the Department of Veterans Affairs or the US gov-ernment. Funding was provided by unrestricted grants from the West-ern Institute for Biomedical Research, Salt Lake City, Utah, and byPfizer Pharmaceuticals. Pfizer Pharmaceuticals supplied Neurontin(gabapentin) 300-mg capsules and matching placebo for use in thisstudy. Dr. Stock was the Investigational New Drug (IND) sponsor un-der Food and Drug Administration IND 66,492. This investigationwas supported by the University of Utah Study Design and Bio-statistics Center, with funding in part from the Public Health Servicesresearch grant numbers UL1-RR025764 and C06-RR11234 fromthe National Center for Research Resources. Except as noted above,funding agencies did not participate in study design, data collection,data analysis, or manuscript writing or review. Clinical Trials registration: NCT01573052: http://clinicaltrials.gov/ct2/show/NCT01573052?term=gabapentin+alcohol+withdrawal&rank=3© 1967-2013 Harvey Whitney Books Co. All rights reserved. No partof this document may be reproduced or transmitted in any form orby any means without prior written permission of Harvey WhitneyBooks Co. For reprints of any article appearing in The Annals, pleasecontact [email protected]

References1. Lejoyeux M, Solomon J, Ades J. Benzodiazepine treatment for alco-

holic-dependent patients. Alcohol Alcoholism 1998;33:563-75.2. Mayo-Smith MF; ASAM Working Group. Pharmacologic management

of alcohol withdrawal. A meta-analysis and evidence based practiceguideline. JAMA 1997;278:144-51.

3. Amato L, Minozzi S, Davoli M. Efficacy and safety of pharmacologicalinterventions for the treatment of the alcohol withdrawal syndrome (re-view). Cochrane Database Syst Rev 2011;6:1-28.

4. Amato L, Minozzi S, Vecchi S, Davoli M. Benzodiazepines for alcoholwithdrawal (review). Cochrane Database Syst Rev 2010;3:1-112.

5. Minozzi S, Amato L, Vecchi S, Davoli M. Anticonvulsants for alcoholwithdrawal (review). Cochrane Database Syst Rev 2010;3:1-108.

6. Reoux JP, Saxon AJ, Malte CA, et al. Divalproex sodium in alcoholwithdrawal: a randomized double-blind placebo-controlled trial. AlcoholClin Exp Res 2001;25:1324-9.

7. Malcolm R, Ballenger JC, Sturgis ET, Anton R. Double-blind controlledtrial comparing carbamazepine to oxazepam treatment of alcohol with-drawal. Am J Psychiatry 1998;146:107-9.



8. Leggio L, Kenna GA, Swift RM. New developments for the pharmaco-logical treatment of alcohol withdrawal syndrome. A focus on non-ben-zodiazepine GABAergic medications. Prog Neuropsychopharmacol BiolPsychiatry 2008;32:1106-17.

9. Stuppaeck CH, Barnas C, Hackenberg K, Miller CH, FleischhackerWW. Carbamazepine monotherapy in the treatment of alcohol withdraw-al. Int Clin Psychopharmacol 1990;5:273-8.

10. Myrick H, Malcolm R, Randall PK, et al. A double-blind trial ofgabapentin versus lorazepam in the treatment of alcohol withdrawal. Al-cohol Clin Exp Res 2009;33:1582-8.

11. Bonnett U, Banger M, Leweke FM, et al. Treatment of acute alcoholwithdrawal with gabapentin: results from a controlled two-center trial. JClin Psychopharmacol 2003;23:514-9.

12. Voris J, Smith NL, Rao SM, Thorne DL, Flowers QJ. Gabapentin for thetreatment of ethanol withdrawal. Subst Abuse 2003;24:129-32.

13. Mariani JJ, Rosenthal RN, Tross S, Singh P, Anand OP. A randomized,open-label, controlled trial of gabapentin and phenobarbital in the treat-ment of alcohol withdrawal. Am J Addict 2006;15:76-84.

14. Bonnet U, Hamzavi-Abedi R, Specka M, Wildfang J, Lieb B, ScherbaumN. An open trial of gabapentin in acute alcohol withdrawal using an oralloading protocol. Alcohol Alcoholism 2010;45:143-5.

15. Johns M. A new method for measuring daytime sleepiness: the EpworthSleepiness Scale. Sleep 1991;14:540-5.

16. Flannery B, Volpicelli J, Pettinati H. Psychometric properties of the PennAlcohol Craving Scale. Alcohol Clin Exp Res 1999;23:1289-95.

17. Sullivan JT, Sykora K, Schneiderman J, Naranjo CA, Sellers EM. As-sessment of alcohol withdrawal: the revised Clinical Institute WithdrawalAssessment for Alcohol scale (CIWA-Ar). Br J Addict 1989;84:1353-7.

18. Wojnar M, Bizon Z, Wasilewski D. Assessment of the role of kindling inthe pathogenesis of alcohol withdrawal seizures and delirium tremens.Alcohol Clin Exp Res 1999;23:204-8.

19. Gonzales LP, Veatch LM, Ticku MK, Becker HCl. Alcohol withdrawalkindling: mechanisms and implications for treatment. Alcohol Clin ExpRes 2001;25:197S-201S.

20. Modesto-Lowe V, Huard J, Conrad C. Alcohol withdrawal kindling: isthere a role for anticonvulsants? Psychiatry (Edgmont) 2005;2:25-31.

21. Myrick H, Anton R, Voronin K, Wang W, Henderson S. A double-blindevaluation of gabapentin on alcohol effects and drinking in a clinical lab-oratory paradigm . Alcohol Clin Exp Res 2007;31:221-7.

22. Malcolm R, Myrick LH, Veatch LM, Boyle E, Randall PK. Self-reportedsleep, sleepiness, and repeated alcohol withdrawals: a randomized, doubleblind, controlled comparison of lorazepam vs gabapentin. J Clin Sleep Med2007;3:24-32.

23. Poulos CX, Zack M. Low-dose diazepam primes motivation for alcoholand alcohol-related semantic networks in problem drinkers. Behav Phar-macol 2004;14:503-12.

24. Roberto M, Gilpin NW, O’Dell LE, et al. Cellular and behavioral interac-tions of gabapentin with alcohol dependence. J Neurosci 2008;28:5762-71.

25. Fureiri F, Nakamura-Palacios E. Gabapentin reduces alcohol consump-tion and craving: a randomized, double-blind, placebo-controlled trial. JClin Psychiatry 2007;68:1698-701.

26. Anton R, Myrick H, Baros A, et al. Efficacy of a combination of flumazeniland gabapentin in the treatment of alcohol dependence relationship to alco-hol withdrawal symptoms. J Clin Psychopharmacol 2009;29:334-42.

27. Schacht JP, Randall PK, Waid LR, et al. Neurocognitive performance, al-cohol withdrawal, and effects of a combination of flumazenil and gaba pen -tin in alcohol dependence. Alcohol Clin Exp Res 2011;35:2030-8.

28. Elholm B, Larsen K, Hornnes N, Zierau F, Becker U. Alcohol withdraw-al syndrome: symptom-triggered versus fixed-schedule treatment in anoutpatient setting. Alcohol Alcoholism 2011;46:318-23.

29. Allan C, Smith I, Mellin M. Detoxification from alcohol: a comparisonof home detoxification and hospital-based patient care. Alcohol Alco-holism 2000;35:66-9.

30. Saitz R, Mayo-Smith MF, Roberts MS, Redmond HA, Bernard DR,Calkins DR. Individualized treatment for alcohol withdrawal. A random-ized double-blind controlled trial. JAMA 1994;272:519-23.

31. Reid R. Alcohol misuse and withdrawal intervention and screeningguidelines: adult. http://www.ghc.org/all-sites/guidelines/alcohol.pdf (ac-cessed 2012 Jan 9).

EXTRACTOGabapentina vs Clordiazepoxida para el Tratamiento Ambulatoriode Detoxificación de AlcoholCJ Stock, L Carpenter, J Ying, T GreeneAnn Pharmacother 2013;47:961-9.TRASFONDO: Las benzodiacepinas se usan para tratar el retiro de alcoholpero causan deterioro cognitivo, sedación, ataxia, e interactúan conalcohol. Los anticonvulsivos no benzodiacepínicos son prometedores yposiblemente alternativas más seguras para el tratamiento del retiro dealcohol. OBJETIVOS: Comparar las medidas de seguimiento de la Escala deSomnolencia de Epworth (ESE), la Escala del Deseo de ConsumirAlcohol de PENN (EDCAP), la clasificación de ataxia y los síntomas deretiro de alcohol (CIWA-Ar) entre sujetos dependientes de alcoholasignados al azar a recibir tratamiento con gabapentina (GB) oclordiazepoxida (CDB).MÉTODOS: Se llevó a cabo un estudio aleatorio, doble ciego en veteranoscon retiro de alcohol (criterios de DSM-IV). Se excluyeron sujetos querequirieron hospitalización o tomar anticonvulsivos benzodiacepínicos ono benzodiacepínicos. Veintiséis sujetos fueron asignados al azar, 17recibieron GB, y 9 recibieron CDP. Las dosis de GB fueron 1200mgorales por 3 días, seguido por 900 mg, 600 mg, y 300 mg por 1 día cadauno. Las dosis de CBD fueron 100 mg orales por 3 días, seguido por 75mg, 50 mg, y 25 mg por 1 día cada uno. Diariamente se administraronlas escalas CIWA-Ar, ESE, EDCAP y la evaluación para ataxia.RESULTADOS: El seguimiento promedio de las puntuaciones de ESE yEDCAP no fueron diferentes entre los grupos de tratamiento durante elperíodo de tratamiento temprano (días 1-4), pero fueron más bajas(diferencia promedio –3.70, 95% IC) (–7.21, –0.19) p = 0.04) y (diferenciapromedio –6.05, 95% IC (–12.82, 0.72), p = 0.08), respectivamente, alfinal del período de tratamiento (días 5-7) en los sujetos tratados con GB.Las puntuaciones de CIWA-Ar se redujeron similarmente en ambosgrupos. No se observó ataxia. No se observaron eventos adversos signi -fi cativos. Las limitaciones incluyen un tamaño de muestra pequeña y un35% de pérdida del seguimiento al final del período de tratamiento. CONCLUSIONES: En veteranos ambulatorios con síntomas de retiro dealcohol, el tratamiento de GB resulta en una reducciónsignificativamente mayor de la sedación (ESE) y una tendencia a reducirel deseo de consumir alcohol (EDCAP) al final del tratamiento encomparación con CDP. A pesar de la limitación del tamaño pequeño dela muestra, la observación de una reducción en la somnolencia y menosdeseos de consumo justifican la replicación con una muestra mayor.

Traducido por Rafaela Mena

RÉSUMÉLa Gabapentine Contre la Chlordiazépoxide Dans la Cure deDésintoxication de l’Alcool en Clinique AmbulatoireCJ Stock, L Carpenter, J Ying, T GreeneAnn Pharmacother 2013;47:961-9.CONTEXTE: Les benzodiazépines sont utilisées pour traiter le sevrage del’alcool, mais causent des troubles cognitifs, la sédation, l’ataxie etinteragissent avec l’alcool. Les anticonvulsivants non-benzodiazépinessont des alternatives prometteuses et éventuellement plus sécuritairespour le traitement du sevrage.OBJECTIFS: Comparer les mesures de suivi de l’échelle de somnolenceEpworth (EES), l’envie d’alcool (PENN alcool Scale Craving (PACS))et les symptômes d’ataxie et de sevrage de l’alcool (CIWA-Ar) entre lessujets dépendants à l’alcool randomisés à un traitement avec lagabapentine (GB) ou avec la chlordiazépoxide (CDB ).


CJ Stock et al.

MÉTHODES: Une étude randomisée, en double aveugle a été menée chezles vétérans avec un sevrage alcoolique (critères DSM-IV). Les sujetsnécessitant une hospitalisation ou un traitement aux benzodiazépines oudes anticonvulsivants autres que les benzodiazépines ont été exclus. Vingt-six sujets ont été randomisés, 17 ont reçu la GB et 9 reçu la CDP. Lesdoses de GB doses étaient de 1200 mg par voie orale pendant 3 jours, puis900 mg, 600 mg et 300 mg pour 1 journée chacune. Les doses de CBDétaient de 100 mg par voie orale pendant 3 jours, puis 75 mg, 50 mg, et25 mg pour 1 journée chacune. Les évaluations à l’aide des outilsCIWA-Ar, ESS, PACS et pour l’ataxie ont été administréesquotidiennement.RÉSULTATS: La moyenne des scores ESS et PACS ne différaient pas entreles groupes de traitement pendant la période de traitement précoce (1-4jours) mais ont été plus faibles (différence moyenne -3.70, IC

95% –7.21, –0.19; p = 0.04) contre –6.05, IC95% –12.82, 0.72; p =0.08) à la fin de la période de traitement (jours 5-7). Les scores CIWA-Ar ont été réduits de façon similaire dans les 2 groupes. L’ataxie n’a pasété observée. Aucun effet indésirable significatif n’a été noté. Les limites del’étude comprennent la petite taille de l’échantillon et 35% de perte desuivi à la fin de la période de traitement.CONCLUSIONS: En mode ambulatoire, le traitement avec la gabapentine aentraîné une réduction significativement de la sédation (ESS) et unetendance à la diminution de désir d’alcool (PACS) à la fin du traitementpar rapport à la chlordiazépoxide. Bien que limitée par la petite taille del’échantillon, la réduction de la somnolence et de l’envie moindre pourl’alcool suggèrent de répéter l’expérience avec un échantillon plus large.

Traduit par Nicolas Paquette-Lamontagne



gabapentin versus chlordiazepoxide for outpatient alcohol ... · going treatment for alcohol...

Documents