« back · www.benzo.org.uk »
Discontinuation of Alprazolam
Treatment in Panic Patients
Abby J. Fyer, M.D., Michael R. Liebowitz, M.D., Jack M. Gorman,
Raphael Campeas, M.D., Andrew Levin, M.D., Sharon O. Davies, R.N.,
Deborah Goetz, B.S., and Donald F. Klein, M.D.
Alprazolam treatment was tapered in 17 panic patients at a rate of 10% of the starting dose every 3 days. Only four subjects completed withdrawal on schedule (4-5 weeks); four additional subjects discontinued treatment in 7-13 weeks. During withdrawal 15 patients had recurrent or increased panic attacks and nine had significant new withdrawal symptoms. Most common among the latter were malaise, weakness, insomnia, tachycardia, lightheadedness, and dizziness. None had seizures, psychosis, or significant neurological or EEG abnormalities. Results indicate that relapse and withdrawal are important considerations in the choice of alprazolam treatment for panic attacks. (Am J Psychiatry (1987; 144:303-308)
The triazolobenzodiazepine alprazolam is a new, effective antipanic agent (1-3). Alprazolam's rapid action and low side effect profile make it an important therapeutic addition. However, what happens when alprazolam is discontinued is unclear.
When a drug dose is lowered or discontinued, two types of symptoms may appear: those specifically due to drug withdrawal and those which are a recrudescence of the drug-suppressed illness (relapse).
Withdrawal symptoms are highly stereotyped, appear in proportion to daily dose and length of treatment, and disappear with time. In contrast, relapse shows the distinctive illness pattern and does not follow a course of temporary exacerbation followed by normalization. However, distinguishing between withdrawal and relapse in anxiety patients may be difficult, since some anxiety symptoms are often part of the withdrawal pattern.
Received Oct. 7, 1985; revised May 1 and Aug. 6, 1986; accepted Sept. 4, 1986. From the Anxiety Disorders Clinic, New York State Psychiatric Institute; and the Department of Psychiatry, Columbia University College of Physicians and Surgeons, New York, N.Y. Address reprint requests to Dr. Fyer, New York State Psychiatric Institute, 722 West 168th St., New York, NY 10032.
Supported in part by NIMH grants MH-33422 and MH-30906 and Research Scientist Development Award MH-00416 (Dr. Gorman) and by the Upjohn Company.
The authors thank Dr. Timothy Pedley for assistance in the neurological evaluation of patients and Mr. Tom Cooper for assay of alprazolam plasma levels.
Copyright © 1987 American Psychiatric Association.
Benzodiazepine withdrawal studies in anxious patients without panic disorder indicate that many have symptom recurrence or a withdrawal syndrome. Over 50% of patients whose drugs are discontinued experience symptoms such as insomnia, agitation, anxiety, tremor, irritability, headache, nausea, and dizziness (4-10). Recurrence of symptoms of illness at greater than pretreatment levels (rebound) has also been described (4, 8).
Abrupt benzodiazepine discontinuation can cause seizures (11, 12). Three cases of grand mal seizures on abrupt cessation of alprazolam have been reported (13, 14).
Here we will describe tapered alprazolam discontinuation in panic disorder patients. Specific questions we addressed were as follows: 1) Does tapered discontinuation of alprazolam induce a withdrawal syndrome? If so, what are its characteristics? 2) Does alprazolam discontinuation affect frequency and severity of panic attacks? 3) Does the illness become reestablished? 4) Can withdrawal be successfully completed without adjunctive treatment?
The 18 withdrawal study subjects were drawn from 30 treatment study patients (3). Twelve patients did not enter; six had terminated treatment before the withdrawal study, three developed major depressive disorder, and three refused to participate.
All 18 subjects had DSM-111 panic disorder or agoraphobia with panic attacks. Subjects were between 18 and 60 years old and without serious medical illnesses, current DSM-111 major depression, or alcohol or drug abuse in the past 6 months. Subjects were largely self-referred outpatients. Results of physical examination, CBC, blood chemistries, thyroid profile, EKG, EEG, and urinalysis were normal at baseline. Written informed consent was obtained from all subjects.
All 18 subjects participated in a 13-week treatment study (3). Responders only were scheduled to continue for an additional 12-week maintenance period before withdrawal. Therefore responders underwent withdrawal after a minimum of 24 weeks and nonresponders after 12-16 weeks of drug treatment.
Alprazolam was decreased every 3 days at as close to a rate of 10% of the original dose as was permitted by 0.5 mg tablets. The goal was to discontinue medication in 30 days.
Each week patients received separate packets for each dose of the upcoming week. As the dose decreased, active medication was replaced by indistinguishable placebo. Therefore, the number of daily tablets remained constant. Patients were told about the taper, but the rate was kept from them.
Patients also received an emergency bottle containing fifteen 0.5 mg alprazolam tablets. If a patient called between visits in significant distress, the treating psychiatrist could give an additional emergency dose of active alprazolam. If more than one extra dose was needed during a week, the patient was removed from the protocol at the next visit and a slower open withdrawal schedule was begun. Patients were also removed from the protocol at scheduled visits if they were so symptomatic that fixed tapering was hazardous. If even slow tapering led to a relapse, an alternative antipanic drug (e.g., imipramine) was started or alprazolam was increased to an effective level.
Research psychiatrist clinical assessment, neurological examination, and EEG were done before treatment, one day before withdrawal, and weekly during withdrawal.
Patients kept a record of anxiety symptoms. The number of panic attacks was recorded at weekly visits; in this report, spontaneous and situational panic attacks were combined. Clinical condition was rated on two panic and phobic scales that measured severity and change. The five factors measured were panic attacks, anticipatory anxiety, phobic avoidance, functional impairment, and overall clinical condition. These factors were rated on a 7-point scale (1=normal or markedly improved, 4=moderate or no change, 7=among the most seriously ill or markedly worse). Ratings for panic attacks combined frequency and intensity.
Withdrawal symptoms and side effects were rated on a scale containing 32 items that frequently accompany benzodiazepine withdrawal. Items were rated on a 4-point scale (0=none, 3=severe). For each subject the average number of new withdrawal symptoms was determined as follows: all scores for symptoms on the symptom and side effect scale that the subject had not had at more than a mild level score=0 or 1) before withdrawal but had at least moderately (score=2 or 3) during withdrawal were added together, and this sum was then divided by the number of withdrawal protocol weeks. Two subjects (patients 7 and 10) had incomplete scale ratings. Symptoms described in their charts as new during alprazolam discontinuation were considered withdrawal symptoms; however, withdrawal scores were not calculated.
Neurological examinations were done by a Boardcertified neurologist or a senior neurology fellow working under the latter's supervision. Blood samples for determination of alprazolam plasma levels were drawn weekly during withdrawal in a subgroup of patients. Levels were assayed by gas liquid chromatography according to the method of Greenblatt et al. (15).
Sixteen of the 18 subjects responded to alprazolam: 14 in the acute treatment study and two (subjects 3 and 16) during maintenance (see table 1). Mean length of drug treatment before withdrawal was 29.4 weeks (range, 23-42 weeks) for patients who entered the maintenance phase (acute treatment responders) and 14.75 weeks (range, 12-20 weeks) for those who did not (acute treatment nonresponders). Mean drug dose at the start of withdrawal was 5.25 mg/day (range, 2.5-8.0 mg/day) for acute responders and 6.0 mg/day (range, 3-8.5 mg/day) for acute nonresponders. Sixteen of the 18 subjects were panic free before withdrawal.
Only four of the 18 subjects completed withdrawal from alprazolam by following the fixed protocol (see table 1). Thirteen subjects were unable to complete withdrawal at that rate; seven terminated prematurely because of panic recurrence, five because of a combination of panic recurrence and new withdrawal symptoms, and one because of new withdrawal symptoms only. The four successfully tapered patients did not differ on demographic or treatment characteristics from the 13 who failed to follow the protocol.
TABLE 1. Characteristics of Panic Patients Who Did and Did Not Successfully Discontinue Alprazolam Treatment
|Number of panic attacks|
|Discontinuation||New Withdrawal Symptoms|
faintness, decreased appetite
fatigue, ataxia, irritability,
depression, tremor, weakness,
headaches, constipation, nasal
congestion, blurred vision
|4||23||F||No||5.5||0||14||5||2||4||0.80||3.2||Nausea, malaise, rigidity,|
|5||53||F||Yes||2.5||1.0||32||4||0||4||1.00||3.7||Malaise, increased appetite|
confusion, dizziness, impaired
insomnia, loss of appetite
|11||32||F||Yes||8.0||5.0||24||2||0||0||0||3.0||Blurred vision, tachycardia,|
|12||32||M||Yes||5.0||0.5||26||4||0||1||0.25||3.0||Insomnia, malaise, ataxia,|
|17||30||M||No||8.5||2.5||12||4||0e||3||7.80||2.2||Rash, dry mouth, sweating|
a Before start of discontinuation.
b Total spontaneous and situational panic attacks reported during discontinuation divided by number of weeks in discontinuation.
c See Method section for definition.
d No rating; data from clinical chart.
e Subject 17 had five partial but no full panic attacks during prewithdrawal week.
The 18th subject dropped out in her second tapering week after suicidal ideation following a marital argument. Since panic attacks recurred, her overall discontinuation outcome was unsuccessful; however, she was excluded from other analyses. (Her exclusion left 15 subjects who were panic free before withdrawal.)
Fifteen of the 17 patients had a recurrence or increase in panic attacks during discontinuation compared to their prediscontinuation status. One (subject 4) had a similar frequency of panic attacks before and during tapering. Only one patient (subject 11) had no panic attacks.
Eight of these 15 patients who relapsed (subjects 1, 2, 3, 6, 8, 12, 14, and 16) had mild to moderate panic symptoms. In this group, the average weekly number of panic attacks during tapering was 1.5 or less (see table 1). Typically these patients had one or two panic attacks a week for 1 or 2 weeks. Three subjects (1, 2, and 3) of the four patients who completed the protocol were in this group; the fourth (subject 4) had no change in panic attacks during tapering.
In contrast, the other seven patients (subjects 5, 7, 9, 10, 13, 15, and 17) constituted a "severe panic relapse" group. All but one (subject 5) averaged more than two panic attacks a week (range, 2.3-7.8) during withdrawal (see table 1). Subject 5 had a lower score (1.0) because her first 3 panic-free withdrawal weeks reduced the impact of her fourth week, in which four severe attacks led to alteration of her withdrawal schedule. Subject 15 had a dramatic increase in the intensity (but not the frequency) of panic attacks. In this group, calls to request a medication increase between visits were frequent. All subjects had several weeks when they experienced several panic attacks each week. Four (subjects 7, 9, and 17 during the protocol and subject 15 after the protocol) had a frequency and severity of panic attacks during withdrawal that exceeded pretreatment levels. The possible persistence of this "rebound" effect could not be evaluated, since treatment was reinstituted soon afterward in view of the patient's disturbance.
Before the tapering of alprazolam, 10 of 17 patients were rated well or only minimally ill on the "overall" item of the Panic and Phobia Severity Scale; however, at termination from the protocol none of the patients was rated completely recovered, and only two were well enough to be rated minimally ill.
TABLE 2. Commonly Reported
incomplete ratings; patient 7 had no pretreatment ratings,
and patient 10 no termination ratings.
Table 2 lists newly emergent withdrawal symptoms reported by at least three patients. The most commonly reported symptoms were quite similar to those typically reported during benzodiazepine withdrawal. However, seizures, psychotic symptoms, organicity, delirium, and syncope did not occur.
Fourteen of the 17 subjects reported at least two newly emergent withdrawal symptoms. One (subject 10) reported only one such symptom, and two (subjects 1 and 15) reported none. Table 1 lists the specific symptoms reported.
Six of the 15 symptomatic patients (subjects 8-11, 14, and 17) had mild to moderate withdrawal symptoms: either 1-2 weeks of moderate discomfort with several symptoms or 3-4 weeks with one symptom. The remaining nine patients (subjects 2-7, 12, 13, and 16) had more severe symptoms: either 1-2 weeks of considerable discomfort and numerous symptoms or persistent moderate difficulty with several symptoms over 3-4 weeks. Most patients had considerably more complaints than is reflected by the first column of table 2, since any symptom that also occurred during treatment was excluded from this list. The second column of table 2 lists the frequency with which these latter symptoms occurred.
Weekly neurological examinations were completed on 10 subjects, including six of the nine with severe withdrawal symptoms. Two subjects had mild hand tremors at one examination-one had hyperflexia and one mild ataxia. There were no other neurological abnormalities.
Weekly EEG recordings were completed on 10 subjects, including five of the nine with severe withdrawal. Two had excessive slowing on one EEG examination. Both abnormalities occurred in the postprotocol phase. All other EEGs were normal. None had spikes, sharp waves, or any signs suggestive of seizure activation.
Weekly alprazolam plasma levels of eight subjects during withdrawal showed sequential decreases paralleling dosage. Two subjects refused to participate in this aspect of the study because of needle phobia. Levels of the remaining eight patients were not measured regularly because of scheduling difficulties.
Ten of the 13 patients who did not complete alprazolam withdrawal during the protocol continued withdrawal at slower rates. Formal clinical ratings continued during this phase. Four of these patients (subjects 7, 9, 10, and 16) discontinued alprazolam. Their total tapering times were 7, 12, 13, and 6 weeks, respectively. Therefore, a total of eight of 17 patients completed alprazolam withdrawal by simple tapering of dose and support. Six additional patients attempted to taper alprazolam at rates of 0.5 mg-0.25 mg/week. All stopped because they were unwilling to tolerate persistent panic symptoms. All six started tricyclic antidepressants while taking 0.5-2.0 mg/day of alprazolam.
Of the eight patients who discontinued alprazolam treatment and did not take other antipanic medication (four during and four after the protocol), six were followed for 3-5 months after medication was discontinued. Three subjects (2, 9, and 16) remained well (no or very occasional panic attacks). One (subject 7) relapsed immediately, one (subject 1) at 4 months, and one (subject 4) at 5 months.
Does Alprazolam Induce Withdrawal, and If So,
What Are Its Characteristics?
Fourteen of our 17 subjects experienced at least two new withdrawal symptoms during alprazolam discontinuation. In half (nine of 17), symptoms were considered severe. However, there were no medically dangerous effects, and only one subject left the protocol because of withdrawal symptoms alone.
The new withdrawal symptoms observed were similar in type to those described in previous studies of withdrawal from therapeutic doses of benzodiazepines (4-10).
The rate of new withdrawal symptoms in our patients appears higher than rates reported in previous studies of benzodiazepine discontinuation. However, accurate comparisons are difficult. Earlier studies differed from ours with respect to subject diagnosis, length of drug treatment, history of previous benzodiazepine use, rate of drug decrease, and drug half-life. Since all of these factors affect severity of withdrawal symptoms, study variation may explain the observed differences.
What Effect Does Alprazolam Discontinuation Have on Frequency and Severity of Panic Attacks?
Our most striking finding was the rapid recurrence or increase of panic attacks in most of our patients (15 of 17 subjects, 14 of 15 subjects who were panic free). DuPont and Pecknold (16) have presented similar results from a withdrawal study of 60 panic patients treated with alprazolam for 8 weeks. In that study 89% (32 of 36) of panic-free patients had panic recurrence during withdrawal.
The rates of panic recurrence during alprazolam discontinuation observed by ourselves and DuPont and Pecknold appear considerably higher and more rapid in onset than those seen in panic patients tapered from tricyclic antidepressants (17) or monoamine oxidase (MAO) inhibitors (18). If confirmed, this finding may indicate that alprazolam discontinuation can precipitate panic attacks in panic patients. Another possibility is that this finding reflects the difference between the rapid elimination of alprazolam compared to the longer-lived effects of MAO inhibitors and tricyclic antidepressants, although the latter are not well documented. Additional alternative explanations might be differences between drug antipanic mechanisms or an as yet undetected distinction between high and low-potency benzodiazepines.
Further studies including comparison of panic and nonpanic patients during withdrawal of alprazolam and other benzodiazepine as well as nonbenzodiazepine compounds are necessary to address this question.
Transient Flare-Up Versus Relapse: Does the Illness Become Reestablished?
We are not able to determine whether the panic relapse during alprazolam discontinuation reflected a transient flare-up or a persistent relapse. For ethical reasons, 10 of our subjects either did not discontinue alprazolam or started treatment with another antipanic medication within a few weeks of or before discontinuing alprazolam. However, of the six subjects who were medication free after the protocol, three remained panic free for 1-5 months and two had only infrequent panic attacks that did not require treatment.
DuPont and Pecknold (16) stated that while 73 % of their patients had panic attacks during medication tapering, only 54% still had panic attacks 2 weeks after medication discontinuation. This, together with our findings, suggests that in some patients, recurrence of panic attacks may be temporary. However, the numbers are too small and follow-up too inconsistent for firm conclusions.
Can Discontinuation Be Successfully Completed Without the Use of Adjunctive Medication?
Our data suggest that with a 10% decrease rate, adjunctive medication may be required to successfully complete alprazolam discontinuation in at least one third of panic patients. Six of our 17 patients required adjunctive use of tricyclic antidepressants in order to complete tapering of alprazolam. DuPont and Pecknold indicated that a firm attitude on the part of the investigator generally increased the number of patients who successfully discontinued medication. Our goal was to take patients off the medication. However, since the possible consequences of alprazolam discontinuation are relatively unknown, we did tend to be cautious when confronted with a patient in significant distress.
One approach is to slow the tapering rate as discussed earlier. An alternative strategy is to discover a rapidly acting drug that blocks relapse if taken during drug discontinuation.
Klein has proposed a possible explanation of the rapid panic recurrence and withdrawal symptoms observed during alprazolam discontinuation (personal communication). He speculated that alprazolam exerts its antipanic effect by blocking afferent pathways to the locus ceruleus or other noradrenergic centers. Massive deafferentiation usually produces receptor hypersensitivity distal to the point of blockade (19). In this case, discontinuing alprazolam treatment would be expected to leave these hypothetical noradrenergic center receptors hypersensitive and prone to overresponse.
If this model is accurate, giving patients whose alprazolam treatment is being tapered off a quickacting α2 agonist such as clonidine might prevent symptomatic exacerbation during withdrawal. Clonidine decreases locus ceruleus firing by stimulating presynaptic inhibitory receptors (20, 21).
Limitations and Conclusions
This study has several major limitations. Most significant is the lack of a double-blind placebo control design. It is impossible to assess the degree to which a patient's anxiety about alprazolam discontinuation contributed to withdrawal symptoms or requests to stop discontinuation. Less severe benzodiazepine withdrawal symptoms have been reported in placebotreated patients withdrawn in a double-blind manner compared to those openly withdrawn (22).
Incomplete compliance data are a second problem. For 10 subjects there were no compliance data. We cannot rule out the possibility that the mildness of their withdrawal symptoms was due to augmentation of the protocol regimen by nonprotocol drugs. Future studies should include systematic urine toxicology data to rule out this possibility.
Ratings were done at the patients' scheduled weekly visits. If a patient became symptomatic and requested a dose increase between visits, by the time the formal rating occurred the most severe withdrawal effects had been blunted. The week's rating was an average of these events. The ratings may therefore have underestimated the severity of symptoms.
Our findings indicate that relapse, withdrawal symptoms, symptomatic exacerbation, and possible adjunctive medication during tapering must be major considerations in deciding whether to use alprazolam for the treatment of panic attacks. Further studies, including direct comparisons to tapering with other antipanic agents, variations in rate of alprazolam decrease, and the use of adjunctive medications, are necessary before alprazolam's therapeutic role can be fully assessed.
Chouinard G, Annable L, Fontaine R, et al: Alprazolam in the treatment of generalized anxiety and panic disorders: a doubleblind placebo-controlled study. Psychopharmacology 1982; 77: 229-233.
Alexander PE, Alexander DD: Alprazolam treatment for panic disorders. J Clin Psychiatry 1986; 47:301-304.
Liebowitz MR, Fyer AJ, Gorman JM, et al: Alprazolam in the treatment of panic disorders. J Clin Psychopharmacol 1986; 6: 13-20.
Fontaine R, Chouinard G, Annable L: Rebound anxiety in anxious patients after abrupt withdrawal of benzodiazepine treatment. Am J Psychiatry 1984; 141:848-852.
Rickels K, Case WG, Downing RW, et al: Long-term diazepam therapy and clinical outcome. JAMA 1983; 250:767-771.
Petursson H, Lader MH: Withdrawal from long-term benzodiazepine treatment. Br Med J 1981; 283:643-645.
Tyrer P, Rutherford D, Huggett T: Benzodiazepine withdrawal symptoms and propranolol. Lancet 1981; 1:520-522.
Pecknold JC, McClure DJ, Fleuri D, et al: Benzodiazepine withdrawal effects. Prog Neuropsychopharmacol Biol Psychiatry 1982; 6:517-522.
Tyrer P, Owen R, bawling S: Gradual withdrawal of diazepam after long term therapy. Lancet 1983; 1:1402-1406.
Covi L, Lipman RS, Pattison JH, et al: Length of treatment with anxiolytic sedatives and response to their sudden withdrawal. Acta Psychiatr Scand 1973; 49:51-64.
De Bard ML: Diazepam withdrawal syndrome: a case with psychosis, seizure, and coma. Am J Psychiatry 1979; 136:104105.
Rifkin A, Quitkin F, Klein DF: Withdrawal reaction to diazepam. JAMA 1976; 236:2171-2173.
Levy AB: Delirium and seizures due to abrupt alprazolam withdrawal: case report. J Clin Psychiatry 1984; 45:38-39.
Breier A, Charney DS, Nelson JC: Seizures induced by abrupt discontinuation of alprazolam. Am J Psychiatry 1984; 141: 1606-1607.
Greenblatt DJ, Dizoll M, Moschitto LJ, et al: Electron capture in gas chromotographic analysis of the triazolobenzodiazepines alprazolam and triazolam. Chromotography 1981; 225:205-207.
DuPont RL Jr, Pecknold JC: Alprazolam withdrawal in panic disorder patients, New Research Abstracts of the 138th Annual Meeting of the American Psychiatric Association. Washington, DC, APA, 1985.
Zitrin CM: Combined pharmacological and psychological treatment of phobias, in Phobia. Edited by Mavissakalian M, Barlow DH. New York, Guilford Press, 1981.
Kelly D, Guirguis W, Frommer E, et al: Treatment of phobic states with antidepressants: a retrospective study of 246 patients. Br J Psychiatry 1970; 116:387-398.
Brown IR: Molecular Approaches to Neurobiology. New York, Academic Press, 1982.
Hoehn-Saric R, Merchant AF, Keyser ML, et al: Effects of clonidine on anxiety disorders. Arch Gen Psychiatry 1981; 38: 1278-1282.
Liebowitz MR, Fyer AJ, McGrath P, et al: Clonidine treatment of panic disorder. Psychopharmacol Bull 1981; 17:122-123.
Winokur A, Rickels K: Withdrawal and pseudowithdrawal from diazepam therapy. J Clin Psychiatry 1981; 42:442-444.
Am J Psychiatry 144:3, March 1987