« back · www.benzo.org.uk »

Protracted Withdrawal Syndromes
from Benzodiazepines

First published:
Journal of Substance Abuse Treatment
Vol. 8 pp. 19-28 · 1991

Professor C Heather Ashton DM, FRCP

School of Neurosciences
Division of Psychiatry
The Royal Victoria Infirmary
Queen Victoria Road
Newcastle upon Tyne NE1 4LP

The Ashton Manual · Professor Ashton's Main Page

Abstract: The benzodiazepine withdrawal syndrome is a complex phenomenon which presents serious difficulties in definition and measurement. It is particularly difficult to set out precise limits on its duration. Many withdrawal symptoms are a result of pharmacodynamic tolerance to benzodiazepines, some mechanisms for which are discussed. Such tolerance develops unevenly in different brain systems and may be slow to reverse. Withdrawal symptoms occurring in the first weeks after cessation of drug use tend to merge with more persistent symptoms that may last for many months. These prolonged symptoms do not necessarily constitute "true" pharmacological withdrawal symptoms, but are nevertheless related to long-term benzodiazepine use. Such symptoms can include anxiety, which may partly result from a learning deficit imposed by the drugs, and a variety of sensory and motor neurological symptoms. The protracted nature of some of these symptoms raises the possibility that benzodiazepines can give rise not only to slowly reversible functional changes in the central nervous system, but may also occasionally cause structural neuronal damage.

Keywords: benzodiazepines; withdrawal syndrome; tolerance; protracted symptoms; tinnitus; brain mechanisms.

Drug withdrawal syndromes, in general, tend to consist of mirror images of the drugs' initial effects. Thus, abrupt withdrawal from chronic usage of beta adrenoceptor antagonists such as propranolol may give rise to tachycardia and palpitations; abrupt withdrawal from antihypertensive doses of clonidine may be followed by hypertension, anxiety, and other signs of increased sympathetic activity. Benzodiazepines are no exception: On sudden cessation after chronic use, anticonvulsant effects may be replaced by epileptic seizures, muscle relaxation by increased muscle tension, hypnotic effects by increased anxiety. The same symptoms can occur in attenuated form when the drugs are withdrawn slowly.

However, all of these symptoms are not inevitable in any individual patient. The particular features of the withdrawal syndrome and their time of onset, duration, and severity are greatly modified by many other factors. Such factors include pharmacokinetic variables, dosage and duration of drug use, rate of withdrawal, the presence or absence of the original disorder (such as anxiety) for which the drug was prescribed, personality characteristics, physical makeup and susceptibility, and the use of concomitant treatments. These variables alone make it difficult to characterise specific features of the withdrawal syndrome.

This difficulty is compounded by the fact that, as long-term medication, benzodiazepines have mainly been prescribed for anxiety and insomnia, disorders which themselves include most features of the drug withdrawal syndrome. When such patients undergo reduction of benzodiazepine dosage, especially slow reduction, how can one specify which emergent symptoms are "true", drug-related withdrawal symptoms, which are "pseudowithdrawal" symptoms (Tyrer, Owen, & Dawling, 1983), which represent a return of the original anxiety state, and which are the natural reactions of an anxious personality undergoing the stress of withdrawal? In circumstances such as these, the benzodiazepine withdrawal syndrome becomes largely a matter of definition.

Nevertheless, the existence of a benzodiazepine withdrawal reaction, from both high and low (therapeutic) doses of benzodiazepines, is no longer in dispute, and many attempts have been made to define and measure it and to estimate its incidence and duration.

Definitions and Measurements

Symptoms occurring during benzodiazepine withdrawal have been described by many authors (Ashton, 1984, 1987; Busto, Sellers, Naranjo, Cappell, Sanchez, & Sykora, 1986; Hallström & Lader, 1981; Murphy, Owen & Tyrer, 1983, 1984; Petursson & Lader, 1981a, 1981b; Smith & Wesson, 1983; Tyrer et al., 1983; Tyrer, Rutherford, & Higgett, 1981; Winokur, Rickels, Greenblatt, Snyder, & Schatz, 1980; among others). Commonly described symptoms are shown in Figure 1. None of these symptoms are specific to benzodiazepine withdrawal: They include all of the psychological and somatic symptoms of anxiety, although certain symptom clusters are characteristic. Owen and Tyrer (1983) and Petursson and Lader (1981a & b) emphasised the appearance of new symptoms, not experienced before withdrawal and uncommon in anxiety states. These new symptoms include hypersensitivity to sensory stimuli (sound, light, touch, taste, and smell) and perceptual distortions (e.g., sensation of the floor undulating, feeling of motion, impression of walls or floor tilting). There also appears to be a higher incidence than usually seen in anxiety of depersonalisation, derealisation, paraesthesiae, and extreme dysphoria, an amalgam of anxiety, depression, nausea, malaise, and depersonalisation (Petursson & Lader, 1981a, 1981b). Visual hallucinations, distortion of body image, psychotic reactions, formication, muscle fasciculation and twitching (occasionally resembling myoclonus), and considerable loss of weight are also described during benzodiazepine withdrawal and are unusual in anxiety states.

Smith and Wesson (1983) and Ashton (1984) drew attention to the characteristic fluctuation of symptoms, which may wax and wane without obvious psychological provocation. Smith and Wesson (1983) suggest that this wavelike symptomatology is an important marker for distinguishing the low-dose benzodiazepine withdrawal syndrome from symptom re-emergence. However, since symptoms may fluctuate in the course of the day or over periods of days or weeks, accurate recording is difficult.

Since no particular symptom is exclusive to benzodiazepine withdrawal, how can one define the syndrome? Tyrer et al. (1981, 1983) have attempted various methods in placebo controlled studies:

  1. The appearance of two or more new symptoms during the withdrawal period. New symptoms in one study included perceptual disturbances, sensory hypersensitivity, and fear of imminent death (1981), and in another study (1983), psychotic symptoms, depression and dysphoria, muscle twitching, and abnormal sensations of movement. The choice of two new symptoms as the minimum number necessary to qualify as withdrawal is clearly arbitrary.

  2. An increase in self-rated symptoms (Comprehensive Psychiatric Rating Scale) to greater than 50% of baseline levels, followed by a return to lower values. Symptom resolution is clearly an important feature in differentiating between symptoms due to withdrawal and symptom re-emergence, emergence, or overinterpretation. As Smith and Wesson (1983, p. 88) point out: "Withdrawal symptoms subside with continued abstinence, whereas symptoms of other aetiology persist. "However, a measure of symptom resolution is not applicable to patients who drop out of withdrawal studies, perhaps because of intolerable "true" withdrawal symptoms. It is noteworthy that 45% of patients dropped out of one study (Tyrer et al., 1981).

  3. A combination of methods (1) and (2) so that a withdrawal reaction is defined as the appearance of new symptoms that resolved before the end of the study (20 and 14 weeks after the end of withdrawal [Tyrer et al., 1983]). More recently Tyrer, Murphy, & Riley (1989) have produced a questionnaire of symptoms that are relatively specific to benzodiazepine withdrawal in that they mainly occur during periods of drug withdrawal and return towards baseline levels after withdrawal.

  4. Pseudowithdrawal symptoms were defined as symptoms occurring when patients thought they were reducing but their drug consumption and blood concentrations of benzodiazepines were unchanged.

Such definitions, derived from double-blind, placebo-controlled studies, have been extremely helpful in the recognition of benzodiazepine dependence, especially low-dose dependence. However, they are of necessity arbitrary and can only be approximate, since the appearance or severity of any particular symptom or symptom cluster may actually represent a variable combination of true withdrawal, pseudowithdrawal, and re-emergence of anxiety, and the same patients liable to pseudowithdrawal reactions are also likely to be most vulnerable to true withdrawal effects.

Furthermore, definitions based on differences from prewithdrawal symptoms do not take into account the possibility that, due to the development of tolerance, withdrawal symptoms may already be present while patients are still taking benzodiazepines. Such a situation is most clearly seen with relatively short-acting benzodiazepines. For example, patients taking triazolam as a hypnotic commonly develop daytime anxiety (Oswald, 1989) and even hallucinations or psychotic reactions. These are almost certainly withdrawal effects since they are immediately relieved by taking the drug but eventually disappear after the drug is stopped (Ashton, 1987). Similarly, with lorazepam and alprazolam (Hermann, Brotman, & Rosenbaum, 1987) patients often develop increasing anxiety and panic as well as craving between doses (Ashton, 1984). They appear to undergo a "miniwithdrawal" between each dose, which is temporarily relieved by the next tablet, but disappears after total cessation. An analagous condition is seen with alcohol: alcoholics commonly complain of tremor and insomnia, symptoms which are temporarily relieved by alcohol but which only disappear after a period of abstinence. Even with long-acting benzodiazepines such as diazepam, there is usually a history in long-term users of steadily increasing anxiety, with the development over the years of new symptoms such as agoraphobia, often with perceptual distortions and depersonalisation, despite continued usage of these supposedly anxiolytic drugs. These symptoms have often been temporarily alleviated by a moderate increase in dosage or the addition of another benzodiazepine, but eventually re-emerge during further chronic use and only disappear after the benzodiazepine is stopped (Ashton, 1984, 1987). Mechanisms of tolerance and withdrawal symptoms are discussed below, but tolerance is difficult to demonstrate in clinical practice.

Because of these many immeasurable factors, it is doubtful whether the boundaries of a "true" benzodiazepine withdrawal syndrome can ever be clearly demarcated.


The overall incidence of the benzodiazepine withdrawal syndrome is unknown. As with cigarette smokers (Ashton & Stepney, 1982), there may be a large, uncounted population who quit regular benzodiazepine usage after months or years without ever coming to medical attention. Tyrer (1990) notes that it is surprising how many patients in ordinary practice have no difficulties whatsoever in reducing their benzodiazepines, and the incidence of a benzodiazepine withdrawal syndrome in general practice appears to be around 30% (Tyrer et al., 1981, 1983; Tyrer, 1989; Tyrer, Murphy, & Riley, in press; Tyrer, 1990). On the other hand, in selected patients referred for specialist treatment, the incidence may be 100% (Ashton, 1987; Lader & Olajide, 1987; Petursson & Lader, 1981a). It is also worth noting that withdrawal syndromes in the form of rebound insomnia (Kales, Scharf, & Kales, 1978) or more general symptoms can occur in experimental subjects and in nonanxious patients prescribed benzodiazepines for sports injuries (Lader, 1988). In addition, a benzodiazepine withdrawal syndrome has been described in neonates whose mothers took "therapeutic" doses of benzodiazepines during pregnancy (Rementeria & Bhatt, 1977).

Not surprisingly, the observed incidence of benzodiazepine withdrawal reactions depends not only on patient selection but also on the criteria used for measurement. In the study of Tyrer et al. (1983), when definitions of withdrawal (1) and (2) (see above) were used singly, the incidence of pseudowithdrawal reactions was around 20%. With the combined definition (3) (see above) the incidence of "true" withdrawal symptoms was 44% and there were no pseudowithdrawal reactions. This incidence of course only applies to those consenting to take part in the study and managing to finish it. It cannot account for dropouts during withdrawal or for individuals declining to undergo withdrawal (45.5% of eligible patients in the study of Tyrer et al., 1981). Thus the incidence of benzodiazepine withdrawal, like its diagnosis, becomes largely a matter of definition.


The identification of the benzodiazepine withdrawal syndrome is difficult enough; its duration is even more difficult to delineate. Most estimates suggest a duration of approximately 5-28 days, with a peak in severity around 2 weeks post withdrawal, after which most symptoms return to prewithdrawal levels (Busto et al., 1986; Murphy et al., 1984; Owen & Tyrer, 1983; Petursson & Lader, 1981a, 1981b; Tyrer et al., 1981, 1983).

To a large extent the apparent duration depends upon how long the patients are followed up, and several authors have drawn attention to the prolonged nature of postwithdrawal symptoms in some cases. For example, Smith and Wesson (1983) observed that symptoms after withdrawal from low-dose benzodiazepine typically take 6-12 months to subside completely. Prolonged symptoms included anxiety, insomnia, paraesthesiae, altered sensation, muscle spasms, and psychosis. Ashton (1984, 1987) reported a similar protracted time-course. Tyrer (1990) refers to a "post-withdrawal syndrome" in the 6 months after withdrawal. Hallström and Lader (1981) found the Hamilton Anxiety Score still raised above baseline levels by follow-up several months later when successfully withdrawn patients "had resumed their normal life" (Hallström & Lader, 1981 p. 237). Olajide and Lader (1984) suggested that depression may be an integral part of the benzodiazepine withdrawal syndrome and may last several months after withdrawal in susceptible individuals; this phenomenon was also observed by Ashton (1987). Busto, Fornazzari, and Naranjo (1988) described two cases in whom severe tinnitus first appeared during benzodiazepine withdrawal and persisted for 6 and 12 months after discontinuation before finally diminishing or disappearing. In one of these cases the tinnitus was immediately alleviated by diazepam in a double-blind placebo-controlled trial conducted over 1 week, 6 months after withdrawal; after a further 6 months of abstinence the tinnitus had became tolerable.

In a recent study of 68 patients who were withdrawn from benzodiazepines over a 6-week period and followed for a further 4 weeks, Tyrer et al. (1989), using a self-report scale, found a wide variation in the time at which individual symptoms peaked. Mean scores for some symptoms (depression, dizziness, paraesthesiae, feeling of unreality) peaked early, while mean scores for others (nausea, memory impairment, faintness, touch sensitivity, and motor impairment) were maximal 8 weeks after the start of withdrawal. Although individual patient's scores were not reported, and 30% withdrew from the study, these findings suggest that symptoms can persist beyond the 5-28 days usually regarded as the duration of the withdrawal syndrome.

Ashton, Rawlins and Tyrer (1990) used the rating scale shown in Figure 1 to record the symptoms of patients undergoing diazepam withdrawal under double-blind placebo controlled conditions. Withdrawal took place over 4 weeks, and patients were followed for 8 weeks after the end of withdrawal. Half the patients received placebo and the other half withdrawal. Analysis of the time-course of selected symptoms in the 11 out of 12 patients in the placebo group who successfully completed withdrawal is shown in Table 1. Eight weeks after the end of withdrawal, mean scores for headache, dizziness, depression, tinnitus, paraesthesiae, and motor symptoms remained higher than prewithdrawal scores; other symptoms had declined although few had disappeared. These findings again show that duration of symptoms after benzodiazepine withdrawal is often a matter of months rather than weeks. Different symptoms persisted in the Tyrer et al. (1989) and Ashton et al. (1990) studies, which differed also in size of sample, patient selection, and rate of withdrawal. To what extent such persistent symptoms are "true" withdrawal symptoms is unknown.

A further problem in assessing the duration of the withdrawal syndrome is the interpretation of the baseline (prewithdrawal) symptoms and anxiety scores. Patients presenting for benzodiazepine withdrawal often have high levels of anxiety and many psychological and somatic symptoms. Figures 2 and 3 show Hospital Anxiety Depression (anxiety) (Zigmond & Snaith, 1983) and symptom rating scores for 12 patients on benzodiazepines compared with the scores of 18 healthy university students approaching their exams. Both groups took placebo tablets and were followed for 20 weeks. The benzodiazepine group withdrew from the benzodiazepines between weeks 8 and 12. It is clear that the patients had considerably higher scores than the normal subjects on both scales, even at the beginning of the study while they were still taking benzodiazepines.

Certainly in these patients the benzodiazepines were not effectively controlling anxiety and, as argued above, it is possible that at least some of the presenting symptoms were due to "withdrawal" symptoms even in the presence of the drug, as a result of the development of drug tolerance. Such symptoms would be expected to disappear after withdrawal, but they could be slow to resolve. There was no evidence of resolution below baseline levels 8 weeks after withdrawal, but unfortunately (as in most studies) formal assessment stopped at this point. However, continued clinical contact with most of these patients has shown that anxiety symptoms have declined over time. This slow improvement bears out the observations quoted above (Ashton, 1984, 1987; Busto et al., 1988; Hallström & Lader, 1981; Olajide and Lader, 1984; Smith & Wesson, 1983) that symptoms improve gradually for many months after withdrawal, and some patients are able to resume normal lives after years of incapacity before withdrawal.

Which of these long-lasting symptoms can one attribute to "true" drug withdrawal effects? Is it possible to pinpoint a time at which the benzodiazepine withdrawal syndrome ends and to say with certainty that any residual symptoms must be due to other factors? The problem is similar to that of designating which effects of a bout of influenza or infectious mononucleosis can be attributed to the specific virus. Do such effects include only those of the viral toxaemia? Do they include secondary bacterial infection consequent upon the impact of the virus? Do they include the period of postviral lethargy and depression which, like benzodiazepine withdrawal symptoms, tends to recur in wave-like fashion for several months? Once again, the benzodiazepine withdrawal syndrome appears to slip through the fingers and one is led back to a question of definition. Probably a clear definition of duration is impossible because drug-induced perturbations of central neurotransmission merge imperceptibly into the background of individual, genetically determined, and learned patterns of brain activity.

Protracted Symptoms After Benzodiazepine Withdrawal

All the problems of definition discussed above are multiplied in any attempt to describe protracted benzodiazepine withdrawal syndromes. Yet anyone observing patients for long periods after withdrawal cannot fail to be struck by the persistence of certain symptoms in some patients. These may not be "true" benzodiazepine withdrawal symptoms; nevertheless they are possibly related to benzodiazepine use and often present a clinical problem. Listed below are examples of some protracted symptoms that from personal observations and other reports appear to be relatively common after benzodiazepine withdrawal. Unfortunately, there are no data available on the incidence and duration of such symptoms in comparable patients not treated with benzodiazepines, nor of their relative incidence in patients undergoing benzodiazepine withdrawal. Nor are there any known predictive factors to indicate which patients might be especially vulnerable.

Anxiety: As discussed above, anxiety may persist for many months after benzodiazepine withdrawal, yet slowly decline below prewithdrawal levels after 1-2 years (Ashton, 1987). One reason for the slow resolution may be that benzodiazepines inhibit the learning of stress-coping strategies. Such effects have been demonstrated in animals and man (Gray, 1987). Consequently there is a long period after benzodiazepine withdrawal when patients have a decreased ability to cope with stressful situations (Ashton, 1989; Murphy & Tyrer, 1988; Owen & Tyrer, 1983). Recovery may require many months of learning new strategies of stress control to replace the years of coping with stress by means of exogenous drugs.

Hence persisting anxiety after benzodiazepine withdrawal does not necessarily imply the re-emergence of an anxiety state existing before benzodiazepine treatment; it may represent the uncovering of a type of learning deficiency induced by long-term benzodiazepine use. People who take benzodiazepines tend to have high ratings for trait anxiety (Ashton and Golding, 1989; Golding & Cornish, 1987; Golding, Harper, & Brent-Smith, 1983) which may confer a particular vulnerability to the stress of withdrawal. Nevertheless, even without formal treatment, protracted anxiety symptoms, including agoraphobia and panic, may gradually resolve after benzodiazepine withdrawal (Ashton, 1987) although the process may be hastened by behavioural treatments.

Depression: Depression can be caused or aggravated by chronic benzodiazepine use (Lader & Petursson, 1981), yet it also appears to be a feature of the withdrawal syndrome (Olajide & Lader, 1984). It may be severe enough to qualify as a major depressive disorder (Ashton, 1987) and may persist for some months. Olajide and Lader (1984) suggest that the mechanisms for postwithdrawal depression may be central serotonin depletion, but there is no direct evidence for this. Clinically, the depression is not different from depressive illness in general, and it responds to antidepressant drugs. It is not clear whether protracted depressive symptoms are more common in patients with a previous history of depression or whether it recurs in subsequent years after withdrawal.

Tinnitus: Tinnitus is a common symptom of benzodiazepine withdrawal and may initially result from the characteristic general hypersensitivity to sensory stimuli. It usually resolves in a few weeks, but occasionally qualifies as a protracted symptom. Busto et al. (1988) describe two cases in which tinnitus persisted for 6 and 12 months after withdrawal and mention a third patient who was unable to withdraw from benzodiazepines because of severe tinnitus at each attempt. Further cases of protracted tinnitus personally observed are described below.

Case 1: Female, aged 54. Duration of benzodiazepine usage: 8 years. Withdrawal slowly from oxazepam 45mg daily in 1986. Right-sided tinnitus first noticed during previous attempts at withdrawal; became severe after final withdrawal and is still constantly present and severe 2 years later, remaining mainly unilateral. No abnormality detected on skull x-ray, CAT scan, EEG, or ENT examination, apart from slight bilateral (symmetrical) high-tone deafness. Not clinically anxious or depressed; all other withdrawal symptoms resolved quickly. Taking no medication.

Case 2: Female, aged 62. Duration of benzodiazepine usage: 18 years. Withdrawn slowly from diazepam 15mg daily in 1988. Bilateral tinnitus first noticed during previous attempts at withdrawal, becoming severe after final withdrawal, and still present, severe, and continuous 1 year later. ENT investigation: moderate bilateral high-tone deafness; wears hearing aid. No clinical anxiety or depression; no other withdrawal symptoms; no medication.

Case 3: Female, aged 58. Duration of benzodiazepine usage: 20 years. Withdrew from diazepam 20mg daily over 4 weeks in 1988. Developed acute psychotic reaction, which resolved in one week. First noticed left-sided tinnitus 1 month after withdrawal. Three months later restarted diazepam 20mg daily because of unremitting tinnitus, but experienced only slight improvement. Second withdrawal over one year 1988-9. Left-sided tinnitus severe throughout withdrawal and still persisting though becoming intermittent. ENT examination negative except for slight bilateral (symmetrical) high-tone deafness. Medication: mianserin 30mg daily, started after onset of tinnitus; carbamazepine recently added with no effect on tinnitus.

Case 4: Female, aged 70. Duration of benzodiazepine usage: 20 years. First noticed tinnitus, mainly right-sided, on withdrawal from flurazepam 30mg daily in 1985. Tinnitus continuous ever since, not relieved by a course of diazepam. ENT examination normal. Drugs: no psychotropics; inhalers for asthma.

Tinnitus is fairly common in the general population and the apparent relation to benzodiazepine use may be incidental, but these cases raise the suspicion that benzodiazepines may occasionally cause permanent or only slowly reversible brain damage. Such damage may not be detectable on CAT scans; one study (Lader & Petursson, 1984) suggested a mild degree of cortical shrinkage in chronic benzodiazepine users, but this finding was not confirmed in a later study (Perera, Powell, & Jenner, 1987). Two of the above patients also complained of unilateral headaches.

Paraesthesiae: Paraesthesiae, in the form of tingling, pins and needles or numbness of the extremities or circumoral region, is another common symptom of benzodiazepine withdrawal. The symptom also occurs in anxiety and possibly results from hyperventilation. Benzodiazepines depress the sensitivity of the respiratory centre to carbon dioxide (Gilmartin, Corris, Stone, Veale, & Gibson, 1988) and it is possible that the respiratory centre becomes hypersensitive during withdrawal, triggering hyperventilation. Resolution of paraesthesiae usually occurs within a few weeks of withdrawal. Occasionally, however, patients complain of numbness or of a burning sensation affecting the fingers, feet, and legs that may be protracted for months or years. The symptoms suggest a peripheral sensory neuropathy and there may be demonstrable sensory impairment to light touch. Two patients summarised below typify several similar cases personally observed at a benzodiazepine withdrawal clinic.

Case 1: Female, aged 56. Duration of benzodiazepine usage: 15 years. Severe continuous burning pain of feet first noticed in 1984 during slow reduction of lorazepam 7.5mg to 1.5mg daily. Changed to diazepam and slowly withdrawn over 1 year 1987-8. Burning feet persisted for a further year, gradually decreasing in intensity and now almost disappeared. No abnormal signs; good peripheral pulses. Medication; promethazine 50mg nocte since 1988.

Case 2: Female, aged 67. Duration of benzodiazepine usage: 12 years. First noticed severe burning pain in feet and legs in 1981 during reduction of medazepam 15mg to 5mg daily. Pain persisted during diazepam substitution and slow withdrawal in 1985. 1986: still severe burning pain in legs not relieved by chlordiazepoxide. 1989: pain still present. Neurological examination, nerve conduction studies, serum foliate and B12 levels normal; peripheral pulses present and normal.

Formication is also common during benzodiazepine withdrawal, and many patients temporarily complain of a feeling of insects crawling on the skin or of lice or nits in the hair. Occasionally, more bizarre sensations are reported, such as a feeling of slime or water running over the body, a sense of inner vibration, or a feeling of "trembling inside", and these symptoms may be protracted. Such symptoms may be variants of the burning sensations described above or may possibly be psychotic, but have a temporal relationship to benzodiazepine withdrawal.

Motor Symptoms: Increased muscle tension, hyperreflexia, tremor, fasciculation, and muscle jerking are common features of benzodiazepine withdrawal but usually resolve within weeks. Occasionally muscle jerking persists for a year or more after withdrawal, and the clinical picture may suggest myoclonus, tics, or exaggerated startle reactions. Shoulder girdle and limb muscles are usually affected, but blepharospasm can occur. Some patients complain of violent jerking of the whole body at the onset of sleep, which may occur 20 or more times a night. Restless legs syndrome may also be protracted. Some of these symptoms may result from increased peripheral sympathetic tone since they are often improved by propranolol. Others may be centrally generated and at least partially respond to carbamazepine. Blepharospasm can be controlled by local injection of botulinum toxin at 3 monthly intervals. Protracted motor symptoms such as these raise the possibility that benzodiazepines are capable of causing long-term hyperexcitability of motoneurons or central motor pathways.

Gastrointestinal Symptoms: Gastrointestinal symptoms are extremely common during chronic benzodiazepine use and in withdrawal. Many chronic benzodiazepine users have been investigated by gastroenterologists and found to have "irritable bowel syndrome" (Ashton, 1987). Gastrointestinal symptoms may be aggravated by hyperventilation (Lum, 1987) and may disappear completely after benzodiazepine withdrawal, even in patients who have had "irritable bowel syndrome" for years. Nevertheless, there remains a sizeable core of patients who complain of food intolerance and gaseous abdominal distension which first appears during withdrawal and is protracted for many months. Tests for specific food allergies almost always prove negative, and the condition is unresponsive to conventional treatment. Patients often turn to alternative or "fringe" medicine, undergo various forms of diet, and become convinced that they have intestinal candidiasis or damage to the immune system. None of these claims have scientific support, although Lum (1987) reports that hyperventilation provokes histamine release and that the incidence of food intolerance and pseudo-allergic reactions is high in chronic hyperventilators. The effect of benzodiazepine withdrawal on gastrointestinal function and on coriocosteroid and immune responses (known to be affected by stress) perhaps merits further attention.

Brain Mechanisms of Benzodiazepine Withdrawal Symptoms

The primary effect of benzodiazepines is enhancement of gamma-aminobutyric acid (GABA) activity on postsynaptic GABA-A receptors in the brain. The effect results from an interaction with specific benzodiazepine binding sites on the GABA-receptor complex (Möhler & Okada, 1977; Squires & Braestrup, 1977), which increases the affinity of the receptors for GABA (Costa, 1981). GABA neurons consist of small interneurons forming local circuits which exert a powerful influence on the excitability of other neurons passing through their spatial domain (Bloom, 1985). Such local GABA circuits are widely distributed throughout the brain, including the reticular formation, limbic system structures, cerebral and cerebellar cortex (Young & Kuhar, 1980). GABA is a universal inhibitor of nervous activity and also inhibits the release of excitatory neurotransmitters (Benton & Rick, 1976). Thus the actions of benzodiazepines include not only enhancement of GABA activity at many brain sites but also decreased release of acetylcholine, noradrenaline, dopamine and serotonin (Faefely, Pieri, Pole, & Schaffer, 1981). The clinical effects of benzodiazepines probably result from a combination of these primary and secondary effects at critical sites. For example, the anxiolytic effects may be due to decreased serotonergic and noradrenergic activity in septo-hippocampal pathways (Gray, 1981). Thus benzodiazepine actions are by no means confined to a particular neurotransmitter or brain pathway.

Any chronically used drug gradually engenders a series of homeostatic responses which tend to restore normal function despite the presence of the drug. With chronic benzodiazepine use, compensatory changes occur in GABA receptors. Such changes consist of decreased sensitivity of these receptors to GABA, probably as a result of alterations in affinity state and decreased density (Cowan & Nutt, 1982; Nutt, 1986). In addition, there are changes in the secondary systems controlled by GABA, so that the output of excitatory neurotransmitters tends to be restored, and/or the sensitivity of their receptors increases. The whole complex of primary and secondary changes eventually results in benzodiazepine tolerance.

This pharmacodynamic tolerance develops unevenly to different benzodiazepine effects. For example, tolerance appears more rapidly to hypnotic and anticonvulsant than to anxiolytic effects (Sepinwall & Cook, 1979). Tolerance to different effects may also vary between individuals, possibly due to variations in intrinsic GABA activity in different parts of the brain, which are in turn reflected in personality characteristics and susceptibility to stress. Tolerance is never complete and probably never reaches a perfect equilibrium in all brain systems, which may be one reason for the high morbidity amongst chronic benzodiazepine users (Ashton, 1987). Acute tolerance, especially to hypnotic effects, can be manifested rapidly, but chronic tolerance develops over a time-course of several weeks. Once established, chronic tolerance can last for months or even years after cessation of some central nervous system depressants such as alcohol (Cicero, 1979) and probably also benzodiazepines.

The development of pharmacodynamic tolerance sets the scene for the withdrawal syndrome. Cessation of the drug exposes all the adaptations which have accrued to counteract its presence, releasing a rebound of unopposed activity involving many neurotransmitters and their receptors and many brain systems. Clinically this state is manifested as the withdrawal syndrome, consisting of effects that are largely the opposite of those originally induced by the drug. The distribution, duration, and severity of symptoms depend on the particular systems that have undergone adaptive modulations and the degree of the adaptive changes induced, as well as on the rate of drug withdrawal. Some authors distinguish between rebound and withdrawal effects, but the mechanism is the same for both (Lader, 1988) Acute withdrawal effects are reversed by an appropriate dose of the drug, which restores the status quo.

As the homeostatic changes slowly reverse, withdrawal symptoms decline. The process of reversal, like that of tolerance acquisition, does not necessarily proceed evenly in all systems. The variable time of emergence and duration of individual symptoms during benzodiazepine withdrawal noted by Tyrer (1989) and Ashton et al. (1990) (Table 1) may reflect this uneven course. The perturbations of brain function induced by benzodiazepines are exceedingly complex, and it is not surprising that withdrawal symptoms are many and variable. Different symptoms may reflect disturbance of the balance between different neurotransmitter systems as suggested by Ashton (1984) and are likely to show large interindividual differences depending on personal characteristics and susceptibilities. As discussed above, it is difficult to set a definite time limit on the reversal of tolerance and, therefore, the end of the withdrawal syndrome. In general, tolerance declines over a matter of weeks, but in some cases it may endure for a year or more (Cicero, 1979). Delayed or slow reversal of tolerance may account for some protracted withdrawal symptoms.

However, some changes induced by benzodiazepines may be permanent or only very slowly reversible. Since benzodiazepines apparently inhibit learning, especially for coping with stress strategies (Gray, 1987), cessation after many years of use may expose a learning deficit, especially in the ability to cope with stress. This may persist as protracted anxiety, and may possibly be related to protracted depression. Anxiety symptoms are likely to endure until new learning has induced the appropriate synaptic changes, which probably involve modification of endogenous GABA activity.

Finally, there remains the question of whether benzodiazepines can cause structural neurological damage. Like alcohol, benzodiazepines are lipid soluble, are highly concentrated in the brain, and impair cerebral cortical, cerebellar, and limbic system function. It is possible that use over many years could cause physical changes such as cortical shrinkage, which may be only partially reversible. Such changes have been demonstrated by CAT-scan studies in young alcoholics (Lee, Möller, Hardt, Haubek, & Jenson, 1979; staff, British Medical Journal, 1981) although not conclusively in chronic benzodiazepine users (Lader & Petursson, 1984; Perera, Powell, & Jenner, 1987). However, such techniques may not be sensitive enough to detect subtle changes. Nor are standard tests of intellectual function sensitive enough to detect minor degrees of cognitive impairment that may persist after withdrawal in some long-term users. It remains possible that some protracted benzodiazepine withdrawal symptoms (including tinnitus and other neurological and psychological symptoms) could result from physicochemical neuronal damage.

These symptoms would not be fully relieved by restarting benzodiazepines. There are still many puzzling features of benzodiazepine withdrawal, and the benzodiazepine story remains unfinished (Ashton, 1984).


Ashton, H. (1984). Benzodiazepine withdrawal: An unfinished story. British Medical Journal, 288, 1135-1140.

Ashton, H. (1987). Benzodiazepine withdrawal: Outcome in 50 patients. British Journal of Addiction, 82, 665-671.

Ashton, H. (1989). Risks of dependence on benzodiazepine drugs: A major problem of long-term treatment. British Medical Journal, 298, 103-104.

Ashton, H., & Golding. J.F. (1989). Tranquillisers: Prevalence, predictors and possible consequences. Data from a large United Kingdom Survey. British Journal of Addiction, 84,541-546.

Ashton, H., Rawlins, M.D. & Tyrer, S.P. (1990). A double blind placebo controlled study of buspirone in diazepam withdrawal in chronic benzodiazepine users. British Journal of Addiction, 157, 232-238.

Ashton, H., & Stepney, R. (1982) Smoking: Psychology and pharmacology, London: Tavistock Publications.

Benton, D., & Rick, J.T. (1976). The effect of increased brain GABA produced by amino-oxyacetic acid on arousal in rats. Psychopharmacology, 49, 85-89.

Bloom, F.E. (1985). Neurohumoral transmission in the central nervous system. In A.G. Gilman, L.S. Goodman, T.W. Rall, & F. Muland (Eds.), The pharmacological basis of therapeutics (pp. 236-259). New York: Macmillan.

Busto, U., Fornazzari, L., & Naranjo, C.A. (1988). Protracted tinnitus after discontinuation of long-term therapeutic use of benzodiazepines. Journal of Clinical Psychopharmacology, 8, 359-362.

Busto, U., Sellers, E.M., Naranjo, C.A. Cappell, H.P., Sanchez, C.M., & Sykora, K. (1986). Withdrawal reaction after long-term therapeutic use of benzodiazepines. New England Journal of Medicine, 315, 654-659.

Cicero, T.J. (1979). Tolerance to and physical dependence on alcohol: Behavioural and neurobiological mechanisms. in Brain and pituitary peptides (pp.1603-1617). Munich: Ferring Symposium 1979.

Costa, E. (1981). The role of gamma-aminobutyric acid in the action of 1,4 benzodiazepines. In J.W. Lamble (Ed), Towards understanding receptors (pp. 176-183). Amsterdam: Elsevier/North Holland.

Cowen, P.J., & Nutt, D.J. (1982). Abstinence symptoms after withdrawal from tranquillising drugs: Is there a common neurochemical mechanism? Lancet,2,360-362.

Gilmartin, J.J., Corris, P.A., Stone, T.N., Veale, D., & Gibson, G.J. (1988). Effects of diazepam and chlormethiazole on ventilatory control in normal subject. British Journal of Clinical Pharmacology, 25, 766-770.

Golding, J.F., & Cornish, A.M. (1987). Personality and life-style in medical students: Psychopharmacological aspects. Psychology and Health, 1, 287-301.

Golding, J.F., Harper, T., & Brent-Smith, H. (1983). Personality, drinking and drug-taking correlates of cigarette smoking. Personality and Individual Differences, 4, 703-706.

Gray, J.A. (1981). Anxiety as a paradigm case of emotion. British Medical Bulletin, 37, 193-197.

Gray, J. A. (1987). The neuropsychology of emotion and personality. In S.M. Stahl, S.D. Iverson, & E.C. Goodman (Eds.), Cognitive neurochemistry (pp.171-190). Oxford: Oxford University Press.

Haefley, W., Pieri, L., Pole, P., & Schaffer, R. (1981). General pharmacology and neuropharmacology of benzodiazepine derivatives. In H. Hoffmeister, and G. Stille (Eds.), Handbook of experimental pharmacology (Vol. 55, 11, pp. 13-262). Berlin: Springer-Verlag.

Hallström, C., & Lader, M. (1981). Benzodiazepine withdrawal phenomena. International Pharmacopsychiatry, 16, 235-244.

Hermann, J.B., Brotman, A.W., & Rosenbaum, J.F. (1987). Rebound anxiety in panic disorder patients treated with shorter-acting benzodiazepines. Journal of Clinical Psychiatry, 48 (suppl. 10), 22-28.

Kales, A., Scharf, M.B., & Kales, J.D. (1978). Rebound insomnia: A new clinical syndrome. Science, 201, 1039-1041.

Lader, M. (1988). The psychopharmacology of addiction - Benzodiazepine tolerance and dependence. In M. Lader (Ed.), The psychopharmacology of addiction (pp. 1-14). Oxford: Oxford University Press.

Lader, M.H., & Olajide, D. (1987). A comparison of buspirone and placebo in relieving benzodiazepine withdrawal symptoms. Journal of Clinical Psychopharmacology, 7, 11-15.

Lader, M.H., & Petursson, H. (1981). Benzodiazepine derivatives - Side effects and dangers. Biological Psychiatry, 16, 1195-1212.

Lader, M.H., Ron, M., & Petursson, R.H. (1984). Computed axial brain tomography in long-term benzodiazepine users. Psychological Medicine, 14, 203-206.

Lee, K., Moller, L., Hardt, F., Haubek, A., & Jenson, E. (1979). Alcohol-injured brain damage and liver damage in young males. Lancet 2, 759-761.

Lum, L.C., (1987). Hyperventilation syndromes in medicine and psychiatry: A review. Journal of the Royal Society of Medicine, 80, 229-231.

Möhler, H., & Okada, T. (1977). Benzodiazepine receptors: Demonstration in the central nervous system. Science 198, 849-851.

Murphy, S.M., Owen, R.T., & Tyrer, P.J., (1984). Withdrawal symptoms after six week’s treatment with diazepam. Lancet, 1389.

Murphy, S.M., & Tyrer, P. (1988). The essence of benzodiazepine dependence. (pp. 157-167). Oxford: Oxford University Press.

Nutt, D. (1986). Benzodiazepine dependence in the clinic: Reason for anxiety? Trends in Pharmacological Sciences 7, 457-460.

Olijade, D., & Lader, M. (1984). Depression following withdrawal from long-term benzodiazepine use: A report of four cases. Psychological Medicine, 14, 937-940.

Oswald, I. (1989) Triazolam syndrome 10 years on. Lancet 1, 451-452.

Owen, R.T., & Tyrer, P. (1983) Benzodiazepine dependence: A review of the evidence. Drugs, 25, 385-398.

Perera, K.M.H., Powell, T., & Jenner, F.A. (1987). Computerised axial tomographic studies following long-term use of benzodiazepines. Psychological Medicine, 17, 775-777.

Petursson, H., & Lader, M.H. (1981a) Withdrawal from long-term benzodiazepine treatment. British Medical Journal, 283, 634-635.

Petursson, H., & Lader, M.H. (1981b). Benzodiazepine dependence. British Journal of Addiction, 76, 133-145.

Rementeria, J.L., & Bhatt, K. (1977). Withdrawal symptoms in neonates from intrauterine exposure to diazepam. Journal of Paediatrics, 90, 123-125.

Sepinwall, J., & Cook, L. (1979). Mechanisms of action of the benzodiazepines: Behavioural aspects. Federation Proceedings 39, 3024-3031.

Smith, D.E., & Wesson, D.R. (1983). Benzodiazepine dependency syndromes. Journal of Psychoactive Drugs, 15, 85-95.

Squires, R.F., & Braestrup, C. (1977). Benzodiazepine receptors in the rat brain. Nature, 266,732-734.

Staff. (1981). Minor brain damage and alcoholism. British Medical Journal, 2, 455-456.

Tyrer, P. (1990) Diagnosis of benzodiazepine dependence. Unpublished manuscript.

Tyrer, P., Murphy, S., & Riley, P. (1989). The benzodiazepine withdrawal symptom questionnaire. Journal of Affective Disorders, 19, 53-61.

Tyrer, P., Owen, R., & Dawling, S. (1983). Gradual withdrawal of diazepam after long-term therapy. Lancet 1, 1402-1406.

Tyrer, P., Rutherford, D., & Higgett, T. (1981). Benzodiazepine withdrawal symptoms and propranolol. Lancet, 1, 520-522.

Winokur, A., Rickels, K., Greenblatt, D.J., & Schatz, N.J. (1980). Withdrawal reaction from long-term low dosage administration of diazepam. Archives of General Psychiatry, 37, 101-105.

Young, W.S., & Kuhar, M.J. (1980). Radiohistochemical localisation of benzodiazepine receptors in rat brain. Journal of Pharmacology and Experimental Therapeutics, 212, 337-346.

Zigmond, A.S., & Snaith, R.P. (1983). The hospital anxiety and depression (HAD) scale. Acta Psychiatrica Scandinavica, 67, 361-370.


Ratings:    0=none    1=mild    2=moderate    3=severe
Drowsiness / Fatigue
Excitability (jumpiness, restlessness)
Depersonalisation / derealisation
Poor memory / concentration
Perceptual distortion
Agoraphobia / phobias
Panic attacks
Paranoid thoughts
Rage / aggression / irritability
Pain (limbs, back, neck)
Pain (teeth, jaw)
Tingling / numbness / altered sensation (limbs, face, trunk)
Stiffness (limbs, back, jaw)
Weakness ("jelly legs")
Muscle twitches
Dizziness / light-headedness
Blurred / double vision
Speech difficulty
Hypersensitivity (light, sound, taste, smell)
Insomnia / nightmares
Nausea / vomiting
Abdominal pain
Diarrhoea / constipation
Appetite / weight change
Dry mouth
Metallic taste
Difficulty in swallowing
Flushing / sweating
Frequency / polyuria / pain on micturition
Menorrhagia / PMT
Mammary pain / swelling
Skin rash / itching
Stuffy nose / sinusitis
Influenza-like symptoms
Sore eyes
Other symptoms (specify)

FIGURE 1. Withdrawal Symptoms Rating Scale



FIGURE 2. Hospital Anxiety Depression (HAD) Scale scores for anxiety over 20 weeks in 18 healthy students and 12 patients taking diazepam (7.5 SD 4.6 daily). Patients and students took placebo tablets until week 20. Patients withdrew from diazepam during weeks 8-12. One patient dropped out after 12 weeks for domestic reasons. Circles represent placebo group; triangles represent students.

FIGURE 3. Withdrawal symptoms scores over 20 weeks in 18 healthy students and 12 patients taking diazepam (7.5mg SD 4.6 daily). Patients and procedures as in Figure 2. Symptom rating shown in Figure 1. Circles represent placebo group; triangles represent students.


TABLE 1. Scores for selected symptoms before, during, and after diazepam withdrawal
(n = 11; mean total scores on withdrawal symptoms scale shown in Figure 1.

Score at start
of 4-week
withdrawal period
Maximum Score
(Weeks after start
of withdrawal)
Score 8 weeks
after end
of withdrawal
Declining Symptoms
(postwithdrawal score equal
to or less than prewithdrawal
  insomnia / nightmares
  nausea / vomiting
  perceptual distortion
  excitability / restlessness
  sensory hypersensitivity (a)
  poor memory / concentration
Persisting Symptoms
postwithdrawal score
greater than prewithdrawal
  anxiety (b)
  paraesthesiae (c)
  motor symptoms (d)
  1. Sensory hypersensitivity: to light, sound, taste, smell.

  2. Anxiety: Hospital Anxiety Depression (HAD) Scale (anxiety) scores.

  3. Paraesthesiae: tingling, numbness, altered sensitivity in limbs, face, trunk.

  4. Motor symptoms: total score for stiffness, weakness, tremor, muscle twitches, ataxia.

The Ashton Manual · Professor Ashton's Main Page

« back · top · www.benzo.org.uk »