SCIENTIFIC INVESTIGATIONS
Self-Reported Sleep, Sleepiness, and Repeated Alcohol Withdrawals: A
Randomized, Double Blind, Controlled Comparison of Lorazepam vs Gabapentin
Robert Malcolm, M.D.1; L. Hugh Myrick, M.D.1,2; Lynn M. Veatch, Ph.D.1,2; Elizabeth Boyle, M.S.W.1; Patrick K. Randall, Ph.D.1
Charleston Alcohol Research Center, Medical University of South Carolina; 2Research Service, Department of Veterans Affairs Medical Center,
Charleston, SC
1
Study Objectives: Insomnia is a central symptom of alcohol withdrawal
and increases relapse potential. The primary objective of this study was
to compare the efficacy of gabapentin to lorazepam in alleviating sleep
disturbances and daytime sleepiness during an episode of alcohol withdrawal. The secondary objective of this study was to determine if drug
treatment efficacy differed by the patient history of previous treatments
for alcohol withdrawal.
Methods: Outpatients in treatment for alcohol withdrawal received a
4-day fixed-dose taper of gabapentin or lorazepam in a double-blind,
randomized, controlled trial with an 8-day follow-up. Daily across a 5
day outpatient treatment and Days 7 and 12 post-treatment, patients
self-reported daytime sleepiness using the Epworth Sleepiness Scale.
Self-reports of depression (Beck Depression Inventory) were completed
at Days 1,5,7 and 12. Staff assessed daily alcohol withdrawal using the
Clinical Institute Withdrawal Assessment for Alcohol. From these instruments, self-reported sleep and sleepiness were extracted and assessed
in the context of limited (0-1) or multiple (2 or more) previously treated
alcohol withdrawal episodes.
Results: Patients with limited previous withdrawals reported similar
treatment effects on self-reports of sleep and sleepiness for gabapentin
and lorazepam. In contrast, patients with multiple previous alcohol withdrawals receiving gabapentin reported reduced sleep disturbances and
sleepiness in comparison to those receiving lorazepam.
Conclusions: During treatment for alcohol withdrawal, gabapentin as
compared to standard therapy with lorazepam, was superior on multiple
sleep measures, in patients who had previous withdrawals. Lorazepam
subjects experienced rebound symptoms. Early drinking was related to
persisting insomnia with both drugs.
Keywords: Alcohol withdrawal, sleep, insomnia, benzodiazepine, lorazepam, gabapentin, sensitization, randomized clinical trial, detoxification,
anticonvulsant
Citation: Malcolm R; Myrick LH; Veatch LM et al. Self-reported sleep,
sleepiness, and repeated alcohol withdrawals: a randomized, double
blind, controlled comparison of lorazepam vs gabapentin. J Clin Sleep
Med 2007;3(1):24-32.
I
nsomnia is a central symptom of alcohol withdrawal; it is one of
8 core criteria in the DSM-IV diagnosis of alcohol withdrawal.1
Historically, clinicians have observed that the inability to sleep
during alcohol withdrawal often is a harbinger of delirium tremens and, conversely, a good night’s sleep early in the course of
alcohol withdrawal often precedes an amelioration of withdrawal
symptoms.2
The restorative forces of sleep extend into the post-withdrawal
time frame as well.3 Brower and colleagues4 found a highly significant sleep difference among alcoholics who were followed up
5 months after initial detoxification and treatment. They found
that while 30% of individuals with no complaints of insomnia
during treatment for alcohol withdrawal later relapsed to alcohol
use, 60% of individuals with insomnia during treatment for alco-
hol withdrawal later relapsed. Alcohol dependence is associated
with other syndromes that are also likely to disrupt sleep, such
as anxiety, depression, sleep disordered breathing, and increased
frequency of periodic leg movements.5 Furthermore, many alcoholics with documented periods of abstinence continue to show
fragmented sleep, reduced delta sleep, and difficulty falling
asleep.6-9
As relapses to drinking are recurring events in alcoholism, so
too are repeated withdrawals. The severity of a current alcohol
withdrawal episode in any given patient depends on a number
of factors. Four very significant factors include: the presence of
comorbid medical disorders, the quantity of alcohol consumed
prior to cessation, the duration of that episode of drinking, and
the number of previous withdrawal events over the lifetime of
the patient.10 This last factor, the number of previous withdrawal
episodes, is the focus of this study. That is, what are the effects of
past single vs multiple withdrawals on current sleep and sleepiness in the present withdrawal treated by lorazepam and gabapentin?
Ballenger and Post11 examined the clinical presentation of several hundred patients who had repeated episodes of treated alcohol withdrawal syndrome. They found that repeated episodes of
alcohol withdrawal syndrome were likely to worsen with each
successive episode, and they postulated that this might be an
“in vivo” model of neuronal sensitization, or kindling, as pro-
Disclosure Statement
This was not an industry supported study. Drs. Veatch, Malcolm, Myrick,
and Randall and Ms. Boyle have indicated no financial conflicts of interest.
Submitted for publication March 16, 2006
Accepted for publication August 24, 2006
Address correspondence to: Robert Malcolm, MD, Professor of Psychiatry, Pediatrics and Family Medicine, 67 President St., P.O. Box 250861,
Charleston, SC 29425, Tel: (843)792-5214; Fax: (843)792-7353; E-Mail:
[email protected]
Journal of Clinical Sleep Medicine, Vol. 3, No. 1, 2007
24
Comparison of Lorazepam vs Gabapentin
posed by Goddard and colleagues.12 Subsequently, several animal
studies have found an intensification of some withdrawal symptoms in animals that have gone through prior repeated withdrawals.13-15 Over a dozen studies have found that alcoholic patients
who have undergone multiple withdrawals are likely to have more
severe withdrawal symptoms than patients having their first or
second alcohol withdrawal. These symptoms include craving for
alcohol, heavy drinking after treatment, cognitive impairment,
and seizures. These symptoms are less responsive to benzodiazepine treatment than those in patients without multiple withdrawal
episodes, even when years of drinking and quantities of drinking
are equated.16 To the best of our knowledge, studies of subjective
complaints of sleep disturbance during an episode of treated alcohol withdrawal have not been examined in the context of a limited
number (0-1) of previous withdrawals vs history of multiple withdrawal episodes (2 or more).
In a previous study evaluating the anticonvulsant carbamazepine vs lorazepam to treat alcohol withdrawal, we found that the
carbamazepine treated group had improved sleep, and this finding persisted into the post-treatment period.17 Individuals taking
lorazepam had improved sleep during treatment as measured by
a self-report scale, but rebound insomnia occurred immediately
after treatment, and sleep remained poor in the immediate postdetoxification period. We did not assess improvement in sleep by
carbamazepine or lorazepam in the context of multiple previous
withdrawals. Carbamazepine has compared well to some benzodiazepines for the treatment of alcohol withdrawal in a few small
double-blind, placebo-controlled trials.17-19 Carbamazepine is not
ideal for treatment of alcohol withdrawal; it is limited by its multiple interactions with other medications and infrequent but toxic
effects related to low serum sodium, bone marrow dyscrasias, and
hepatotoxicity.20
In our most recent experimental clinical work comparing anticonvulsants to benzodiazepines, (unpublished observations) we
chose gabapentin because of its favorable side effect profile, lack
of interactions with other medications, lack of hepatic metabolism, and a favorable report of sleep improvement in alcoholics.21
Becker and colleagues,22 using an animal model, have shown that
gabapentin is effective in reducing sensitization after multiple alcohol withdrawals. We hypothesized that gabapentin would have
superior effects to lorazepam in reducing self-reported measure
of sleep difficulties during alcohol withdrawal in the group who
had undergone multiple previous treated (sensitized) withdrawals in the past. Excessive daytime sleepiness, to the best of our
knowledge, has not been extensively evaluated in a large sample
of patients during alcohol withdrawal. Accordingly, we measured
daytime sleepiness using the Epworth Sleepiness Scale.23 We did
not formulate any prediction regarding daytime sleepiness, since
both gabapentin and lorazepam can produce sedation.20 We also
evaluated the probability of a return to drinking behavior and its
relationship to insomnia during and following treatment.
ments, and word-of-mouth. All subjects had to demonstrate cognitive capacity to give informed consent, and subjects with breath
alcohol levels (BAL) above 80 mg/dL were re-consented on their
second day of treatment. All informed consent procedures were
approved by the Medical University of South Carolina (MUSC)
Institutional Review Board (IRB). Subjects had to have a Mini
Mental State Exam24 score of 26 (30 maximum) or higher and
have the clinically judged capacity to give study consent. Subjects
had to be judged medically stable and not likely to require hospitalization for medical complications within 10 days of entry into
the study. Subjects had to have a score of 10 or higher (indicating
moderate or greater withdrawal) on the Clinical Institute Withdrawal Assessment for Alcohol-Revised (CIWA-Ar).25 Subjects
were required to have a negative urine drug screens for benzodiazepines, other sedative-hypnotics, opiates, and amphetamine.
Subjects were excluded from the study if they had a history of
taking medications known to ameliorate or intensify the AWS.
These included, but were not limited to, tricyclic antidepressants,
anticonvulsants, antipsychotics, alpha-adrenergic agonists, betablockers, calcium channel antagonists, calcium agonists, SSRIs,
SNRIs, and buspirone. Patients were allowed to use nicotine and
caffeine. Subjects were excluded with a diagnosis of any other
substance-dependence syndrome other than alcohol dependence,
with the exception of cannabis, and cocaine. Subjects were excluded if they had a history of idiopathic epilepsy, schizophrenia,
bipolar disorder, or dementia. Subjects with liver function tests 4
times higher than the upper range of normal were excluded from
the study per IRB request. Subjects with a history of hepatic encephalopathy, jaundice, ascites, insulin-dependent diabetes mellitus, or renal insufficiency were also excluded.
PROCEDURES
Initial eligibility screening was generally performed by phone,
with participants providing an estimated amount of daily alcohol
use and, whenever possible, the number of previous medicated
withdrawals along with brief medical screening history. Subjects
were questioned regarding medications which might confound
or complicate alcohol withdrawal treatment. Subjects’ initial appointments for baseline measures and in-person screening were
generally scheduled between 9 a.m. and noon. This was to promote standardization of baseline data collection, first dose of
study medication, and decreased variability of measuring medication effects, diurnal sleep and wake patterns. For similar reasons, subjects were generally started on Mondays, Tuesdays, or
Wednesdays for initial baseline assessment.
Based on our previous open-label pilot work,26 subjects were
randomized to receive 1 of 3 doses of gabapentin or 1 comparison dose of lorazepam. 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. A medications
compliance log was maintained with subjects’ return of medication blister-packs and receipt of new medications. Medications
used in the study were identical capsules, containing gabapentin
400 mg, gabapentin 300 mg, gabapentin 200 mg, and lorazepam
2 mg. On the first day of alcohol withdrawal treatment, the highdose gabapentin group received 400 mg three times a day for 3
days and 400 mg twice daily for 1 day. The mid-range gabapentin
dose was 300 mg 3 times a day for 3 days and 300 mg twice daily
METHODS
Subjects
Subjects were treatment-seeking patients, 21-70 years old, of
both sexes and multiple ethnic groups, who met DSM-IV1 criteria
for alcohol dependence and for current alcohol withdrawal syndrome (AWS). All patients were recruited through referrals from
university clinics, emergency departments, community advertiseJournal of Clinical Sleep Medicine, Vol. 3, No. 1, 2007
25
R Malcolm, LH Myrick, LM Veatch et al
for the last day. The 200 mg dosing group was dosed identically.
Lorazepam 2 mg was administered 3 times a day for the first 3
days of the study and 2 mg twice daily on the last day of treatment. Subjects were randomized using a stratified, permuted block
method.27 Stratification was based on previous alcohol treatment
history (0-1 or >1 past treated alcohol withdrawals) and sex (male
or female). In the present study, because of the relatively small
number of subjects in each treatment arm that had undergone 2
or more withdrawals, we elected to aggregate all subjects in the
gabapentin arms into one group and then compare that group to
the group receiving lorazepam treatment. No placebo dosing arm
was used, due to the potential serious consequences of untreated
alcohol withdrawal and to the availability of effective comparison
treatments.28 Day 5 was the first follow-up day without medication, Day 7 was the second follow-up day without medication,
and Day 12 was the third follow-up day without medications.
Study Measures
CLINICAL INSTITUTE WITHDRAWAL ASSESSMENT
WA-AR)
FOR
Table 1—Clinical Institute Withdrawal Assessment for Alcohol-Revised (CIWA-Ar)
Insomnia Item 11
0
1
2
3
4
5
6
7
Awake one-half of time
Awake throughout the night
Epworth Sleepiness Scale23
This is a subject self-rating scale that uses 8 items on which
an individual might have the propensity to fall asleep during the
daytime. Each item is rated as to the probability of falling asleep
on a scale of 0 to 3. The Epworth scale was administered each
morning and subjects were asked to rate their degree of sleepiness
for the previous 24 hours. The Epworth scale has been studied in
a variety of disorders leading to excessive daytime sleepiness and
used as a measure of medication effects.31,32
ALCOHOL-REVISED (CI-
This is a clinician-rated checklist.25 It is a multidimensional instrument assessing nausea and vomiting, tremor, sweating, tactile
and visual disturbances, auditory disturbances, orientation, sensorium, anxiety, headache, and insomnia. Items 1-10 are included
in the overall assessment for total score. Item 11 (insomnia) is, by
convention, not included in the overall score and was scored and
analyzed separately for the present study. Table 1 depicts Item 11
of the CIWA-Ar. Response 0 is “able to sleep uninterrupted.” Response 1 is “complains of difficulty falling asleep.” Response 4 is
“awake one-half the time.” Response 7 is “awake throughout the
night.” High scores indicate poor sleep. In the present study, all
investigators, co-investigators, project coordinators, and research
assistants were trained with mock subjects and clinical patients in
withdrawal assessment prior to starting the study and periodically
retrained to achieve high congruence in scoring of this instrument.
For the majority of subjects seen in this study, one individual (the
project coordinator, EB) scored the CIWA-Ar. This was rated in
all subjects on a daily basis for each visit and generally between
the times of 9 a.m. and noon. The CIWA-Ar Item 11 was our outcome variable in also examining the probability of drinking the
next day.
NUMBER OF PREVIOUS WITHDRAWALS
As with our previous study,17 we used a semi-structured standard interview to question subjects about the number of previous
withdrawal episodes they had been treated for in the past. We used
a modified version of the Cognitive Lifetime Drinking History.33
Both inpatient and outpatient treated withdrawal episodes were
counted. Subjects were asked about a list of commonly used medications (benzodiazepines, barbiturates, etc.), using both generic
and trade names. Patients who were treated with vitamins, supportive care, and no specific medication to ameliorate withdrawal
symptoms were not counted as having a treated withdrawal.
OTHER PROCEDURES
All subjects were given a breath alcohol test at the start of each
daily visit to obtain a breath alcohol level (BAL). The time since
last drink was also recorded. Additional measures made at appropriate intervals were Timeline Followback of Drinking Behaviors, electrocardiographic examination, and appropriate physical
and laboratory assessments. These and other assessments are reported in our treatment outcome paper.(unpublished observations
of L.H. Myrick)
In summary, subjective reports of sleep were studied in the
context of drug treatment, current withdrawal severity, and depressive symptoms. Daytime sleepiness was assessed by selfreport using the Epworth Sleepiness Scale. Secondary analysis
further evaluated the impact of previous treatments for alcohol
withdrawal on these measures.
BECK DEPRESSION INVENTORY (BDI)
This is a 21-item self-rating questionnaire evaluating multiple
dimensions of mood and depression.29 Item 16, which was the
item analyzed separately for this study, asks “which of these statements best describes the way you have been feeling during the
past few days?” Responses in order include “I can sleep as well as
usual,” “I don’t sleep as well as I used to,” “I wake up 1-2 hours
earlier than usual and find it hard to get back to sleep,” and “I
wake up several hours earlier than I used to and can’t get back to
sleep.” A high score indicates poor sleep. A previous study using
this instrument and single item analysis of Item 16 reported that
a 1 point change in this item represented a 25% clinical improvement.30 The BDI was administered at baseline, Day 5 (one day after the end of medication treatment), and for both follow-up visits
on Days 7 and 12. The scale was modified to tell subjects that for
Days 5, 7, and 12 to use the interval of the previous time since the
last administration of the scale as their evaluation interval.
Journal of Clinical Sleep Medicine, Vol. 3, No. 1, 2007
Able to sleep uninterrupted
Complains of difficulty falling asleep
Statistical Analyses
Subjects were randomized using a stratified permuted block
method27 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=4 in order to promote a
26
Comparison of Lorazepam vs Gabapentin
Figure 1—Flowchart of disposition of subjects in the study.
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). Therefore this
method balanced treatment assignment in each of the combinations formed by crossing these 2 stratification variables.
All sleep-related dependent variables (CIWA-Ar-11, Beck16, and Epworth) were analyzed as a mixed model (SAS Proc
Mixed) with an unstructured variance/covariance matrix. Preliminary model estimates on the primary CIWA-Ar scores revealed
that in addition to the reduction in variance as scores declined
across time, simple covariance structures (heterogeneous compound symmetry, heterogeneous autoregressive, etc.) fit less well
by likelihood ratio criteria than did the unstructured matrix. The
same was true for most of the secondary variables in the main
study, so all dependent variables were analyzed with the unstructured matrix. Further, dependent variables across the treatment
and follow-up periods were conditioned on the baseline measurement for each and on the drinks per day variable derived from the
immediate pre-detoxification period (14 days). These adjustments
were particularly important in analyses distinguishing between
different previous alcohol withdrawal histories, as these covariates often differ between the groups and may otherwise obscure
or confound the interpretation of treatment results.
The basic design utilized a two way factorial, drug treatment
(gabapentin vs lorazepam), and the within subject variable time
for each dependent measure. Separate analyses were completed
for the 2 study phases (treatment and follow-up) with additional
analysis considering differences by day completed to further delineate results when justified by significant interactions.
A secondary analysis was completed to address the history of
previous alcohol treatment. This design utilized a three way factorial (previous alcohol withdrawal history (0-1 vs >1), drug treatment (gabapentin vs lorazepam), and the within subject variable
time which differed for different dependent measures. A variety
Journal of Clinical Sleep Medicine, Vol. 3, No. 1, 2007
of simple effects (e.g., drug by time within previous alcohol treatment levels) were used to further delineate results when justified
by significant or near significant interactions.
The relationship between insomnia and next-day drinking was
analyzed using a multilevel logistic regression. Scores on the
CIWA-Ar-11 variable were used as predictors for dichotomized
drinking (drinking vs not drinking) on the subsequent day. A random regression model was employed in which slopes are permitted to vary between individual subjects and hypotheses tested
regarding the population of slopes in the two groups.
RESULTS
The sample consisted of 449 patients screened by phone for the
study, 281 patients screened in person for the study, and 101 subjects signing informed consent at baseline and receiving at least
one dose of study medication. Figure 1 is a flowchart of subject
disposition. Seventy-five subjects received 4 days of medicated
treatment, and 68 subjects completed follow-ups at Days 5, 7,
and 12. The sample consisted of 25% females, 15% African
Americans, 3% Native Americans, and 1% Hispanic Americans.
The number of prior medicated withdrawals of the patients ranged
from 0 to 5. The subject population had a mean age of 41 years,
a mean educational level of 13.4 years, and a mean income of
$2,023/month. The mean duration of alcohol use was 15 years.
Sixty-five percent of patients reported 0-1 medication-treated
alcohol withdrawals, and 35% reported 2 or more withdrawals
from alcohol that were treated with medication. Age, sex, number of years drinking, number of standard drinks in the 14 days
prior to study entry, income, educational level, and number of
previously treated withdrawals did not differ statistically between
medication treatment groups. There were no differences in medication compliance among the groups.
27
R Malcolm, LH Myrick, LM Veatch et al
were no differences in self-reported side effects between participants treated with gabapentin vs lorazepam (p = .74). Throughout
the trial there was a trend for gabapentin subjects to report more
“high” (intoxication)(p = .17). The gabapentin subjects had a trend
of more pruritus (p = .17). One gabapentin participant was withdrawn due to rash and urticaria (treated with diphenhydramine,
followed by a complete recovery). No patients experienced delirium tremens in either the gabapentin or lorazepam group. There
was a trend for lorazepam participants to report more sedation (p
= .12).
Clinical Institute Withdrawal Assessment for Alcohol-Revised Sleep
Assessments
Self-reports of insomnia, as assessed from Item 11 of the
CIWA-AR, were differentially affected by drug administration,
both during treatment and follow-up. Patients receiving 600 mg,
900 mg and 1200 mg gabapentin did not differ and are combined
in the analysis. Depicted in Figure 2A are the trajectories prior to
initiation of drug treatment, during treatment and through the follow-up period. Across the days of treatment, a drug by treatment
day effect was found (F3,62 = 2.77, p = .049). Post hoc analysis of
this interaction indicated that on the first day of treatment, selfreports of insomnia were significantly lower in patients receiving
lorazepam in comparison to gabapentin (F1,62 = 7.48, p = .008).
In contrast, in the follow-up period, subjects previously taking
lorazepam experienced increased insomnia, while gabapentin
subjects continued to report a reduction in insomnia. These impressions were supported by statistical analysis indicating both
a drug by follow-up day effect (F1,59 = 4.82, p = .032), as well as
a significant difference on the 1st follow-up day in self-reports of
insomnia with subjects previously receiving lorazepam reporting
higher levels of insomnia in comparison to prior gabapentin treatment (F1,59 = 4.07, p = .048).
Analysis also identified a main effect of prior treatment history
of withdrawal on this measure of insomnia, both during treatment
(F1,60 =6.08, p =.016) as well as into the follow-up period (F1,57 =
4.41, p = .04). Depicted in Figure 2 are trajectories of the CIWAAr Item 11 insomnia rating for patients with a history of limited
(0-1) (Figure 2B) and multiple (>1) (Figure 2C) previous treatments for alcohol withdrawal. In patients with a limited treatment
history, there was a trend for lorazepam treatment to reduce selfreports of insomnia during treatment (F1,60= 3.45, p = .068). In the
follow-up period, prior lorazepam treatment initially resulted in
increased insomnia (F1,52 = 7.55, p = .008), while insomnia levels
reported by subjects previously receiving gabapentin remained
stable. A similar, but more pronounced effect was seen among
patients with a history of multiple previous alcohol treatments.
During treatment, lorazepam significantly reduced self-reports of
insomnia on the first day of treatment (F1,169 = 3.68, p = .013).
At the final follow-up day, insomnia reported by subjects with a
history of multiple previous alcohol treatments who had received
lorazepam treatment were significantly higher than all other subject groups (F3,60 = 2.89, p = .043).
Figure 2—Clinical rating of insomnia as assessed from Item 11 of
the Clinical Institute Withdrawal Assessment25 (CIWA). Figure 2A
depicts all subjects, regardless of prior treatment history. Figure 2B
depicts ratings for subjects with history of limited (0-1) prior alcohol
withdrawal treatments. Figure 2C depicts ratings for subjects with a
history of multiple prior alcohol withdrawal treatments. Lorazepam
initially reduced insomnia but a rebound increase in insomnia occurred at follow-up, most prominently in patients with a history of
multiple alcohol withdrawal treatments. In contrast, insomnia in patients receiving gabapentin treatment tended to improve throughout
the course of treatment and follow-up.
*difference between lorazepam and gabapentin treatment on day of
study (p<.05).
Adverse Events
The study was initially designed to compare 3 doses of gabapentin (totaling 600 mg, 900 mg, and 1200 mg/day) to three 2 mg
doses of lorazepam. Eleven individuals were randomized to the
low-dose gabapentin group (200 mg three times daily); 3 of them
experienced significant adverse events (2 with seizure-like activity during treatment, 1 with a syncopal episode requiring emergency room assessment), indicating that this dose was ineffective.
At the request of the Data Safety Monitoring Board, this dose was
discontinued and no further recruitment of subjects was made for
this dose. The eleven subjects in this group completed the study
and their data was included in the gabapentin dataset. No serious
adverse events were experienced by patients randomized to highdose or mid-range gabapentin or 2 mg lorazepam doses. There
Journal of Clinical Sleep Medicine, Vol. 3, No. 1, 2007
Beck Depression Inventory Sleep Assessments
Overall, lorazepam and gabapentin were equally efficacious in
reducing this measure of sleep difficulty during treatment, evidenced as a nonsignificant drug effect during treatment (p>.14).
However, during the follow-up period, significant differences
28
Comparison of Lorazepam vs Gabapentin
Figure 3—Subjective rating of sleep as assessed from Item 16 of the Beck Depression Inventory (BDI). Figure 3A depicts ratings for subjects
with history of limited (0-1) prior alcohol withdrawal treatments. Figure 3B depicts ratings for subjects with a history of multiple prior alcohol
withdrawal treatments. In patients with a history of multiple treated withdrawals, gabapentin treatment was more effective in reducing subjective
ratings of sleep disturbance.
*difference between lorazepam and gabapentin treatment on day of study (p<.05).
were noted by prior drug treatment across the period (F1,59 = 5.02,
p = .029). While discontinuation of gabapentin treatment did not
alter this measure, patients previously treated with lorazepam reported increased sleep difficulty, an effect which did dissipate by
the end of the study (data not shown). Further analysis indicated
that prior treatment history interacted with drug treatment, both
during treatment and follow-up ((F1,57 = 9.49, p = .003) and (F1,57 =
5.28, p = .025), respectively). Depicted in Figure 3 are the scores
on the sleep-related item of the Beck Depression Inventory for
subjects with limited (Figure 3A) and multiple (Figure 3B) previous treatments for alcohol withdrawal. During treatment, while
there were no significant differences among patients with a limited treatment history (p>.65), a highly significant drug main effect was noted among patients with a history of multiple prior
treatments (F1,57 = 18.15, p = .0001); subjects receiving lorazepam
reported significantly higher levels of sleep disturbance during
treatment and follow-up than those receiving gabapentin. (F1,57=
6.98, p = .011).
Figure 4—Subjective ratings of daytime sleepiness as assessed using
the Epworth Sleepiness Scale completed each morning. Subject reports of sleepiness were lower during treatment in patients receiving
gabapentin treatment, most prominently on Day 3.
*difference between lorazepam and gabapentin treatment on day of
study (p<.05).
Epworth Sleepiness Scale
Reports of daytime sleepiness, assessed using the Epworth
Sleepiness Scale, during treatment and follow-up for patients
treated with lorazepam or gabapentin are depicted in Figure 4.
An overall main effect of drug on sleepiness during treatment was
noted in the analysis (F1,65 = 4.42, p = .0394) with patients receiving lorazepam reporting significantly more sleepiness than those
receiving gabapentin. Post hoc analysis by treatment day indicated the effect to be most pronounced on day 3 of withdrawal (F1,65 =
5.54, p = .022). Additional analysis indicated that prior treatment
history had no impact on this effect (p>.78). No significant differences were noted during the follow-up period.
log(likelihood) = f(CIWA-Ar-11) have been converted to probabilities for presentation purposes. For both drug treatment groups,
increasing difficulty with sleep on any night increased the likelihood of drinking on the subsequent day. For examples, 70% of
subjects rating a 7 on CIWA-Ar Item 11 (“Awake throughout
the night”) drank the next day. The slope of the log-likelihood
function for the steeper gabapentin function was B=.27, SE=.09,
t(215)=3.1, p =.002. The lorazepam slope was shallower, but not
significantly different (t(215)= -1.28, p = .21) from the gabapentin slope. A random slopes model was fit, though we were unable
to detect significant individual differences in the slope of the prediction (x2(60) =53.8, p>.5). Intercepts (probability of sleep with
theoretical scores of zero on CIWA-Ar-11) did differ between the
drug groups (t(60)=2.079, p= .037). In the absence of insomnia
(i.e. CIWA Item 11 score of 0), subjects treated with lorazepam
had a higher probability of drinking the next day.
Relationship of Insomnia to Next-Day Drinking
The probability of drinking as a function of Item 11 from the
CIWA-Ar (reflecting sleep difficulty on the preceding night) for
subjects receiving gabapentin or lorazepam treatment is depicted in Figure 5. In this figure, the best fitting linear functions of
Journal of Clinical Sleep Medicine, Vol. 3, No. 1, 2007
29
R Malcolm, LH Myrick, LM Veatch et al
of withdrawal. These symptoms include seizures, craving for alcohol, heavy drinking after treatment, and cognitive impairment;
the symptoms are less responsive to benzodiazepine treatment
than similar symptoms in patients having their first or second alcohol withdrawal, even when years of drinking and quantities of
drinking are controlled.16 The present study suggests that, among
alcoholics with a history of multiple withdrawals, lorazepam is
less effective than gabapentin in reducing insomnia.
The strong relationship between sleep disruption and drinking
on the following day was independent of days into treatment and
monitored sleep difficulty. In this case, the only time at which
gabapentin-treated subjects were more likely to drink than lorazepam subjects occurred during the first 2 days of treatment, during
which the gabapentin-treated subjects were most likely to report
sleep disruption. A rebound in drinking (and other symptoms) occurred when subjects were taken off lorazepam, at which time
they showed more severe sleep disruption than the gabapentintreated individuals. While these data cannot establish a causative
link between insomnia and drinking, these results are supportive
of other studies that suggest sleep disruption is a determinant of
early relapse.
The mechanisms of action of gabapentin and lorazepam are
dissimilar. As a benzodiazepine, lorazepam’s effects are due to
actions in the inhibitory GABA system where benzodiazepines
act as agonists, allosterically binding to the benzodiazepine recognition site on the GABAA receptor.34 In contrast to the identified mechanisms of action for lorazepam, many biological mechanisms for gabapentin’s actions have been proposed, but thus far
the only specific receptor binding is at the alpha(2) delta subunit
of voltage-gated calcium channels.35 The effect of this subunit on
sleep is unknown, however, studies in rat have identified high
levels of this subunit in brain areas involved in sleep and arousal homeostasis.36 Clinically, in adults with localized epilepsy,
300 and 600 mg of gabapentin increased slow wave sleep, and
600 mg slightly decreased REM sleep.37 In nonalcoholics, singledose gabapentin improved several sleep outcome measures previously disrupted by a loading of alcohol.38 Gabapentin reduced
Stage 1 sleep, reduced the number of awakenings, and increased
slow wave sleep. A 600 mg dose of gabapentin slightly reduced
REM sleep. In an open-label trial of gabapentin and trazodone
in abstinent alcoholics with persistent insomnia, gabapentin was
the superior agent in improving sleep as measured by the Sleep
Problems Questionnaire.21
Why should gabapentin exert superior effects in the multiple
withdrawal group? The neural processes regulating sleep are
composed of an integrated network of cortical and subcortical
structures utilizing a number of neurotransmitters and neurohormones. Acetylcholine, amines and orexin are the prominent
wake-promoting neurotransmitters while the interaction of GABAergic, cholinergic and aminergic neurotransmission controls
NREM and REM sleep. It is this balanced interplay among these
various brain sites and neurotransmitter systems which is disrupted during alcohol withdrawal and further exacerbated with
repeated withdrawal. In preclinical models, repeated exposure to
alcohol with abrupt cessation leads to an imbalance of excitatory
and inhibitory amino acid neurotransmitters.10,39 Indeed, GABAA
functions are reduced during withdrawal, and glutamatergic functions are increased, with extracellular glutamate increasing with
each cycle of ethanol withdrawal.39 It is perhaps this imbalance of
the ratio of reduced inhibitory neurotransmitter in relation to in-
Figure 5—Probability of next day drinking as a function of prior
day score from Item 11 on the CIWA-Ar. Best fit linear functions
of the log(likelihood)=f(CIWA-Ar-11), converted to probabilities, for
patients receiving gabapentin or lorazepam treatment indicated that
increasing CIWA-Ar scores were predictive of drinking the next day.
The scale is anchored by 0: Able to Sleep Uninterrupted; 7: Awake
Throughout the Night
DISCUSSION
Outpatients being treated for mild to moderate alcohol withdrawal received a fixed-dose taper for 4 days of gabapentin or
lorazepam in a double-blind, randomized, controlled trial. Sleep
and daytime sleepiness were assessed using subjective rating instruments, and additional analyses evaluated the impact of limited
(0-1) or multiple (2 or more) previously treated alcohol withdrawal episodes. Overall, patients receiving lorazepam reported less
insomnia and more sleepiness early in treatment than patients receiving gabapentin. However, upon completion of treatment and
discontinuation of drug administration, patients previously treated
with lorazepam reported increased insomnia and daytime sleepiness, while patients previously treated with gabapentin continued
to report improvements in these self-reported measures of sleep.
By examining the sample based on the number of previous
treated alcohol withdrawals, we saw some interesting differences
emerge between lorazepam and gabapentin ratings of insomnia
and sleepiness. For the group that had gone through minimal
withdrawals (0-1), lorazepam was better in quickly improving
sleep. However, rebound insomnia was seen when lorazepam was
stopped (Figure 2B). In the multiple previous withdrawal group,
sleep improved only briefly with lorazepam, while insomnia worsened, even during treatment (Figure 2C). Sleep deteriorated further
in the post-treatment week. The gabapentin group showed slowly
improving sleep over the treatment week, and this improvement
was sustained in the post-treatment period. The BDI sleep score
provided further evidence of differences between lorazepam and
gabapentin in the multiple previously treated withdrawal groups
at Day 5, Day 7, and Day 12 (Figure 3B). Gabapentin was superior to lorazepam in reducing insomnia as assessed by the BDI,
an effect sustained throughout the post-treatment week. Epworth
scores indicated less daytime sleepiness in the gabapentin group
than the lorazepam group (Figure 4).
Several studies have found that alcoholic patients who have undergone multiple withdrawals are likely to have more severe withdrawal symptoms than those undergoing a first or second episode
Journal of Clinical Sleep Medicine, Vol. 3, No. 1, 2007
30
Comparison of Lorazepam vs Gabapentin
creased excitatory neurotransmitter that gives the picture of poor
sleep at night and lack of sleepiness in the daytime for the multiple
previous detoxification group. Since there is no placebo control
group, it is difficult to tell whether gabapentin is superior specifically, or simply that lorazepam is a less adequate treatment.
There are several limitations to the present study. While less
than 25% of the initially screened subjects enrolled in the study,
the dropout rate between initial screening and subject enrollment
is in the normal range, comparable to other alcohol withdrawal
studies. Objective electrophysiologic measures of sleep (PSG)
and daytime sleepiness (MSLT) would have been preferable, but
certainly would be a daunting task in outpatients in a sample size
this large. Numerous factors which could have been more rigorously examined for their influence on sleep were not studied
here. There are some limitations to use of the Epworth Sleepiness
Scale. First, this scale is novel for this situation, and it has not
been validated for alcohol withdrawal. Secondly, this scale is not
typically used to measure “next day” sleepiness. Finally, this scale
includes a hazardous situation (Item #8) that subjects were asked
to imagine rather than report the actual situation since driving was
discouraged. We did not examine the relationship of sleep to caffeine intake, or routine sleep hygiene measures. We did not use a
multidimensional standardized rating scale of sleep disturbance.
Subjective history of alcohol withdrawal may be inaccurate. Despite these limitations, our findings across instruments are generally consistent and are supported by findings from other studies
with patients and preclinical models.
11. Ballenger JC, Post RM. Kindling as a model for alcohol withdrawal
syndromes. Brit J Psychiat 1978;133:1-14.
12. Goddard GV, McIntyre DC, Leech CK. A permanent change in
brain function resulting from daily electrical stimulation. Exp Neurol 1969;25:295-330.
13. McCown TJ, Breese GR. Multiple withdrawals from chronic ethanol “kindles” inferior collicular seizure activity: Evidence for kindling of seizures associated with alcoholism. Alcohol Clin Exp Res
1990;14:394-9.
14. Becker HC, Hale RL. Repeated episodes of ethanol withdrawal
potentiate the severity of subsequent withdrawal seizures: An animal model of alcohol withdrawal “kindling”. Alcohol Clin Exp Res
1993;17:94-8.
15. Veatch LM, Becker HC. Electrographic and behavioral indices of
ethanol withdrawal sensitization. Brain Res 2002;946:272-82.
16. Duka T, Gentry J, Malcolm R, et al. Consequences of multiple withdrawals from alcohol. Alcohol Clin Exp Res 2004;28:233-46.
17. Malcolm R, Myrick H, Roberts J, Wang W, Anton RF, Ballenger
JC. The effects of carbamazepine and lorazepam on single versus
multiple previous alcohol withdrawals in an outpatient randomized
trial. J Gen Intern Med 2002;17:349-55.
18. Malcolm R, Ballenger JC, Sturgis ET, Anton R. Double-blind controlled trial comparing carbamazepine to oxazepam treatment of
alcohol withdrawal. Am J Psychiatry 1989;146:617-21.
19. Stuppaeck CH, Pycha R, Miller C, Whitworth AB, Oberbauer H,
Fleischhacker WW. Carbamazepine versus oxazepam in the treatment of alcohol withdrawal: a double-blind study. Alcohol Alcohol
1992;27:153-8.
20. Charney DS, Mihic SJ, Harris RA. Hypnotics and sedatives. In:
Gilman AG, ed. Goodman & Gilman’s The pharmacologic basis of
therapeutics, 10th ed. New York: McGraw Hill, 2001: 399-429.
21. Karam-Hage M, Brower KJ. Open pilot study of gabapentin versus
trazodone to treat insomnia in alcoholic outpatients. Psychiatry Clin
Neurosci 2003;57:542-4.
22. Veatch LM, Myrick DL, Becker HC. Effects of gabapentin treatment on behavioral and electrographic measures of seizure activity
following multiple ethanol withdrawals in mice. Alcohol Clin Exp
Res 2001;25:22A.
23. Johns MW. A new method for measuring daytime sleepiness: the
Epworth Sleepiness Scale. Sleep 1991;14:540-5.
24. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189-98.
25. Sullivan JT, Sykora K, Schneiderman J, Naranjo CA, Sellers EM.
Assessment of alcohol withdrawal: the revised clinical institute
withdrawal assessment for alcohol scale (CIWA-Ar). Br J Addict
1989;84:1353-7.
26. Myrick H, Malcolm R, Brady KT. Gabapentin treatment of alcohol
withdrawal. Am J Psychiatry 1998;155:1632.
27. Pocock SJ. Clinical trials: a practical approach. New York: Wiley,
1983.
28. Moskowitz G, Chalmers TC, Sacks HS, Fagerstrom RM, Smith
H, Jr. Deficiencies of clinical trials of alcohol withdrawal. Alcohol
Clin Exp Res 1983;7:42-6.
29. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring depression. Arch Gen Psychiatry 1961;4:56171.
30. McCall WV, Reboussin BA, Cohen W. Subjective measurement of
insomnia and quality of life in depressed inpatients. J Sleep Res
2000;9:43-8.
31. Dowling GA, Mastick J, Colling E, Carter JH, Singer CM, Aminoff
MJ. Melatonin for sleep disturbances in Parkinson’s disease. Sleep
Med 2005;6:459-66.
32. Thase ME, Fava M, DeBattista C, Arora S, Hughes RJ. Modafinil
augmentation of SSRI therapy in patients with major depressive
disorder and excessive sleepiness and fatigue: a 12-week, open-label, extension study. CNS Spectr 2006;11:93-102.
ACKNOWLEDGMENTS
Supported by the National Institute on Alcohol Abuse and Alcoholism (Grant P50 AA010761)
REFERENCES
1.
Diagnostic and statistical manual of mental disorders, 4th ed. Washington, DC: American Psychiatric Association, 2000.
2. Dickson HS. Essay on mania a potu. Available by the Waring Historical Library, Medical University of South Carolina. Charleston,
SC: Miller, AE, 1836: 9.
3. Crum RM, Ford DE, Storr CL, Chan YF. Association of sleep disturbance with chronicity and remission of alcohol dependence: data
from a population-based prospective study. Alcohol Clin Exp Res
2004;28:1533-40.
4. Brower KJ, Aldrich MS, Robinson EA, Zucker RA, Greden JF. Insomnia, self-medication, and relapse to alcoholism. Am J Psychiatry
2001;158:399-404.
5. Brower KJ. Insomnia, alcoholism and relapse. Sleep Med Rev
2003;7:523-39.
6. Mello NK, Mendelson JH. Behavioral studies of sleep patterns in alcoholics during intoxication and withdrawal. J Pharmacol Exp Ther
1970;175:94-112.
7. Zarcone V. Alcoholism and sleep. In: Passonant P, Oswald I, eds.
Pharmacology of the state of alertness. Oxford: Pergamon Press,
1979: 3-38.
8. Drummond SP, Gillin JC, Smith TL, DeModena A. The sleep of
abstinent pure primary alcoholic patients: natural course and relationship to relapse. Alcohol Clin Exp Res 1998;22:1796-802.
9. Currie SR, Clark S, Rimac S, Malhotra S. Comprehensive assessment of insomnia in recovering alcoholics using daily sleep diaries
and ambulatory monitoring. Alcohol Clin Exp Res 2003;27:12629.
10. Becker HC. Alcohol withdrawal: Neuroadaptation and sensitization.
CNS Spectrums 2000;4:38-65.
Journal of Clinical Sleep Medicine, Vol. 3, No. 1, 2007
31
R Malcolm, LH Myrick, LM Veatch et al
33. Russell M, Marshall JR, Trevisan M, et al. Test-retest reliability of the
cognitive lifetime drinking history. Am J Epidemiol 1997;146:97581.
34. Bowery NG, Smart TG. GABA and glycine as neurotransmitters: a
brief history. Br J Pharmacol 2006;147 (Suppl 1):S109-19.
35. Marais E, Klugbauer N, Hofmann F. Calcium channel α(2)δ subunitsstructure and gabapentin binding. Mol Pharmacol 2001;59:1243-8.
36. Differential distribution of voltage-gated calcium channel alpha-2
delta (α2δ) subunit mRNA-containing cells in the rat central nervous system and the dorsal root ganglia. J Comp Neurol. 2005
24;491(3):246-69.
37. Legros B, Bazil CW. Effects of antiepileptic drugs on sleep architecture: a pilot study. Sleep Med 2003;4:51-5.
38. Bazil CW, Battista J, Basner RC. Gabapentin improves sleep in the
presence of alcohol. J Clin Sleep Med 2005;1:284-7.
39. De Witte P. Imbalance between neuroexcitatory and neuroinhibitory
amino acids causes craving for ethanol. Addict Behav 2004;29:132539.
Journal of Clinical Sleep Medicine, Vol. 3, No. 1, 2007
32