Original Author: Paul Perry, Ph.D, BCPP
Latest Revisers: Paul Perry, Ph.D, BCPP, Kristine Bever-Stille,
Pharm.D.
Creation Date: 1996
Peer Review Status: Internally Peer Reviewed
INTRODUCTION
In considering the efficacy of benzodiazepines in the treatment of generalized anxiety disorder (GAD) there are two issues that require discussion prior to assessing the findings of the controlled trials conducted since criteria for the disorder were introduced in 1980. The first has to do with the legitimacy of the diagnosis. The separation of generalized anxiety disorder and panic disorder, that first appeared in DSM-III, was based in part on an observed difference in the treatment response of generalized- and panic-anxiety. The former appeared to respond specifically to benzodiazepines while the latter responded to antidepressants. However, as evidence has accumulated this difference in treatment response has been questioned and along with it, doubts have been raised about the validity of GAD. The second issue has to do with the design of anxiolytic drug studies. This is a matter of concern because biased findings often result from inadequately designed trials. Once these factors are placed in proper perspective, the efficacy studies using DSM-III-R diagnosed GAD can be critically examined according to the subject selection, study design, and data analysis.
GENERALIZED ANXIETY DISORDER
Two points need to be made with respect to the diagnosis of generalized anxiety disorder. The first is that their continues to be debate about whether this represents a reliable and valid diagnostic category. The second point is that patients who meet current criteria for this disturbance are usually not seen by psychiatrists but by primary care physicians. We are faced, then, with two major sources of uncertainty in examining the drug treatment of patients with generalized anxiety. We are unsure how these patients are to be distinguished from patients with other anxiety and depressive disorders and we, as psychiatrists are unfamiliar with the setting in which most patients of this kind are treated.
In DSM-III, anxiety neurosis was divided into panic disorder and generalized anxiety disorder. Panic disorder was distinguished by the presence of panic attacks but, since most patients with these attacks also had generalized anxiety symptoms there appeared to be a great deal of overlap between the two. The separation of these disorders depends upon reliable assessment of panic attacks which is difficult because no explicit criteria for them have been proposed and because, according to Tyrer, anxiety symptoms can be quite variable (Tyrer 1986). It should be noted that most of the data available since the introduction of DSM-III have tended to support the distinction in question. First of all, there are persons in the general population that meet criteria for generalized anxiety disorder. Also, most of the available data recently reviewed by Noyes, including that from family and twin studies, support the current classification (Noyes 1988). However, the data are limited and we may still find that generalized anxiety disorder is but a mild variant of panic disorder.
Recent studies show that patients with generalized anxiety disorder are commonly seen by family practitioners but are rarely encountered by psychiatrists. In fact, only about 10% of patients with anxiety disorders seen by psychiatrists qualify for the GAD diagnosis (Barlow et al 1986). Nevertheless, as we shall see most treatment studies have been conducted by psychiatrists. The efficacy of antianxiety drugs has most often been tested on psychiatric outpatients and, for this reason, most of the recommendations for their use have come from psychiatric physicians. Yet family practitioners often report that their anxious patients respond to simple reassurance or non-drug therapy. Clearly we have more to learn about the treatment of anxiety in the primary care setting.
RESEARCH METHODOLOGY
Solomon and Hart critiqued the research methodology of 78 double-blind anxiolytic drug studies that had been conducted between 1961 to 1977 (Solomon and Hart 1978). The authors assessed the adequacy of the studies according to the parameters of subject selection, study design, assessment of clinical response, and data analysis. Most of their criticisms apply to more recent anxiolytic trials and are therefore worth reviewing in some detail.
1. Subject selection. Major flaws in early studies included undefined and vague inclusion/exclusion and diagnostic criteria, if they included such criteria at all. This design deficit routinely led to a lack of uniformity in the patient populations being studied. Few of the studies reviewed by Solomon and Hart elaborated on the nature of the diagnoses making it difficult to determine whether patients' predominant symptoms were anxious, depressive or phobic in nature (Solomon and Hart 1978). This continued to be a problem when DSM-II criteria for anxiety neurosis were utilized. Because this diagnosis was broadly defined it was difficult to generate homogenous study populations. Additionally, anxiety neurosis was so poorly defined that it is difficult now to compare the results of more recent drug efficacy studies that utilized Research Diagnostic Criteria (RDC), DSM-III, or DSM-III-R criteria.
2. Study Design. Although the use of the double-blind, cross-over design represents an improvement over less controlled designs, the approach has serious problems. Twenty-eight (35%) of the 78 studies reviewed by Solomon and Hart (1978) utilized this design in comparing chlordiazepoxide, diazepam, oxazepam, and clorazepate to placebo for the treatment of anxiety neurosis. The difficulty stems from the possibility that patients who receive benefit from active medication may recognize lack of efficacy from placebo once they are crossed-over thus, effectively, breaking the blind. An additional fault of the cross-over design involves the confounding of treatment effects by carry-over effects occurring as patients are switched from one treatment to the next. A patient receiving diazepam, who has a 50 hour, elimination half-life for the drug, will require approximately five half-lives or 250 hours to completely washout the drug. Additionally, the "biological" therapeutic effects of anxiolytics probably require an even greater washout period. The amount of time required to observe relapses following discontinuation of lithium in manics, antidepressants in depressives, and antipsychotics in schizophrenics certainly is greater than the five half-lives required for a chemical "washout." Thus diazepam should probably have at least a ten-day "chemical washout" period and probably an even longer period of time for the "biological washout" so as not to bias the effects of the subsequent treatment. Most studies allowed a week if they included any drug-free interval at all (Solomon and Hart 1978). Inadequate sample size was still another problem. Solomon and Hart noted that 26 (33%) of the studies they reviewed utilized sample sizes of less than 20 patients per group, thus increasing the risk of type II errors (Solomon and Hart 1978). Tansella (1980) has recommended a between- patient or parallel design in which patients receive only one drug in the course of the study even though samples sizes of 30 to 50 per treatment group may be required.
3. Data Analysis. The use of untested instruments to assess clinical outcome leaves the reliability and validity of these measurements in doubt. Only 27 (34%) of the studies reviewed by Solomon and Hart utilized validated scales to assess anxiety (Solomon and Hart 1978). Also, most studies exhibited inadequate control of variables such as concurrent use of psychotropic agents; subject matching on age, race, education etc.; previous response to drugs; and duration of treatment. Most presented no information on washout periods, dropout rates or changes in study design. Solomon and Hart suggested a minimum washout of two weeks to prevent contamination of results by previous drug use (Solomon and Hart 1978). They noted that high dropout rates prior to the start of active medication might have been due to drug withdrawal or rebound anxiety problems.
Several studies specified minimum therapeutic dosage. However, this dosage was often either not achieved or not reported, making evaluation of the anxiolytic effect difficult. Also, the use of inappropriate statistical tests may have caused misinterpretation of resulting data. Fifty-six percent of the reviewed studies found benzodiazepines much better and 33% slightly better than placebo for the treatment of anxiety. However, 18% found no difference between the two and 1% found placebo better than diazepam. As the authors pointed out, most studies had major flaws in design or execution.
NON-BENZODIAZEPINE TREATMENTS
Before we assume that benzodiazepines are the treatment of choice for generalized anxiety disorder, the efficacy of these drugs needs to be compared with that of other drugs and psychological treatments for this disorder. A number of studies have examined various non-benzodiazepine treatments and the findings lead us to suspect that the tricyclic antidepressants and buspirone may also be effective drug treatments for generalized anxiety disorder. Non-drug therapies that have been known to work include relaxation and cognitive therapy. Of course, placebo or non-specific therapy may be by itself effective and any active treatment must be shown to be superior (Shapiro et al 1983).
The question of whether benzodiazepines work better than placebo in anxious patients was examined by Shapiro et al (1983)who conducted a six-week, double-blind, clinical study of 224 anxious outpatients recruited from a psychiatric clinic. Patients were randomly assigned to treatment with diazepam or placebo plus weekly brief psychotherapy. The mean daily dose of diazepam was 19.5 mg/d for those who completed the study. After one week, global and anxiety factor ratings significantly favored diazepam but not thereafter. Diazepam was superior to placebo for medium or high anxiety patients but not for those with low anxiety. Thus, after one week, diazepam did not appear to enhance the effects of placebo plus psychotherapy. This study, because of its adequate design and large sample size, raises questions about the efficacy of benzodiazepines. At least, the small difference in efficacy observed in this study indicates that trials of active medication in generalized anxiety disorder ought to include a placebo treatment group.
Prompted by questions regarding the efficacy of benzodiazepines, Kahn et al (1986) implemented an eight week, double-blind, placebo-controlled, multi-center trial of imipramine and chlordiazepoxide that included 242 patients recruited primarily by newspaper advertisements and secondarily from psychiatric outpatient clinics. The 121 completers were retrospectively diagnosed as having generalized anxiety disorder; those with panic and/or agoraphobia were eliminated from the final analysis. Patients with depression were also carefully excluded. Following a two-week placebo washout period, patients were randomly assigned to chlordiazepoxide, imipramine, or placebo for eight weeks. Mean dosages at week eight were chlordiazepoxide 55 mg/d and imipramine 135 mg/d. The results indicated that both imipramine and chlordiazepoxide were significantly superior to placebo. During the first week patients on chlordiazepoxide showed greater improvement than those on imipramine. However, subsequently, imipramine appeared to be significantly superior to chlordiazepoxide in its anxiolytic action.
Tyrer et al (1988) contrasted the efficacy of benzodiazepines to both antidepressants and psychological therapy in patients with anxiety disorders . They conducted a randomized, parallel-group study in 210 psychiatric outpatients. Patients had generalized anxiety disorder (71), dysthymic disorder (65) or panic disorder (74) according to the DSM-III classification and had not received psychotropic drugs for at least four weeks prior to inclusion in the study. Patients were allocated to one of five treatment groups: diazepam 5 mg (28), dothiepin 25 mg (28), placebo (28), cognitive and behavior therapy (CBT) (84) or a self-help program (SH) (42). Treatments were given in flexible doses for six weeks, then withdrawn over four weeks. Cognitive and behavior therapy was given for one hour at weeks 0, 1, 2, 4 and 6. The self-help program consisted of relaxation tapes and a list of self-help organizations. Additional medications were allowed if required and the placebo group received significantly more additional medication (TCAs, benzodiazepine and promethazine) than the other groups. Twenty-eight of 210 patients received some additional psychotropic medication. The placebo group received significantly more additional medication than the other groups (p<0.01). In all diagnostic groups, including GAD, diazepam (mean = 9.9 mg/d) was the least effective treatment; however, no statistical differences in response were seen between the five treatment groups. It seems clear from this and the preceding studies that benzodiazepine trials need to include groups that receive placebo, tricyclic antidepressants, and/or non-drug therapies.
GENERALIZED ANXIETY DISORDER EFFICACY STUDIES
This review is limited to 20 controlled studies that have examined the effectiveness of the benzodiazepines in the treatment of generalized anxiety disorder diagnosed by specific criteria (Ansseau et al 1985, Fontaine et al 1983, Rickels et al 1983, Tyrer and Owen 1984, Castillo et al 1987, Cohn and Wilcox 1986, Ross and Matas 1987, Olajide and Lader 1987, Hohn-Saric et al 1988, Feighner et al 1982, Lapierre et al 1982, Rickels et al 1982, Elie and Lamontagne 1984, Zung et al 1986, Zung 1987, Fabre 1987, Cohn et al 1986, Petracca et al 1990, Strand et al 1990, McLeod et al 1990). The diagnostic criteria utilized in these studies included the Research Diagnostic Criteria, DSM-III, and DSM-III-R. All but two were designated as double-blind in nature. A placebo control group was utilized in nine studies (Ansseau et al 1985, Fontaine et al 1983, Rickels et al 1983, Tyrer and Owen 1984, Castillo et al 1987, Cohn and Wilcox 1986, Ross and Matas 1987, Olajide and Lader 1987, Rickels et al 1982, Zung et al 1986, Zung 1987). Only three of the 20 studies utilized a crossover design (Ansseau et al 1985, Tyrer and Owen 1984, Olajide and Lader 1987) while the remainder were of a parallel design. Diazepam was utilized in 11 (Fontaine et al 1983, Rickels et al 1983, Tyrer and Own 1984, Ross and Matas 1987, Olajide and Lader 1987, Feighner et al 1982, Lapierre et al 1982, Rickels et al 1982, Elie and Lamontagne 1984, Zung et al 1986, Fabre 1987), and alprazolam in 5 (Castillo et al 1987, Cohn and Wilco 1986, Hoehn-Saric et al 1988, Elie and Lamontagen 1984, McLeod et al 1990). Of the non-benzodiazepine drugs, buspirone was utilized in 10 studies (Tyrer and Owen 1984, Cohn and Wilcox 1986, Ross and Matas 1987, Olajide and Lader 1987, Hoehn-Saric et al 1988, Rickels et al 1982, Cohn et al 1986, Petracca et al 1990, Strand et al 1990), while imipramine was used in two (Hoehn-Saric et al 1988, McLeod et al 1990).
The parallel-design studies generated a number of interesting findings that were not entirely consistent or satisfying. Of the seven parallel-design studies that utilized a placebo, two (Kahn et al 1986, Fontaine et al 1983, Rickels et al 1982) found the benzodiazepine more effective than placebo while three (Castillo et al 1987, Ross and Matas 1987, Zung et al 1986) studies found the benzodiazepine equal to the placebo. Additionally, Zung (1987) found clorazepate only more effective than placebo in dysphoric GAD patients. These efficacy findings underscore the importance of including a placebo in the research design of future GAD drug efficacy studies. In seven non-placebo parallel-design studies buspirone, a non-benzodiazepine anxiolytic, was found to be equal to the benzodiazepines (Cohn and Wilcox 1986, Ross and Matas 1987, Feighner et al 1982, Fabre 1987, Cohn et al 1986, Strand et al 1990, Petraka et al 1990). However, in the 2 placebo controlled studies, one found the drug equal to placebo (Ross and Matas 1987) while the other found the drug more effective than placebo (Rickels et al 1982). Two studies have found the tricyclic antidepressant, imipramine, as effective as alprazolam (Hoehn-Saric et al 1988, McLeod et al 1990) while a third study found imipramine more effective than chlordiazepoxide (Kahn et al 1986).. Neither study had a placebo control group. Two studies found one benzodiazepine more effective than another (Fontaine et al 1983), Elie and Lamontagne 1984). Of the 14 studies using parallel design, only six included a placebo control and of those six < or = 50% showed the benzodiazepines and buspirone superior to placebo.
The three crossover-designs produced inconsistent results. First, of all, Tyrer and Owen (1984) found that one week of diazepam was no more effective than one week of treatment with either buspirone or placebo. From this they concluded that the short-term use of anxiolytic drugs is a highly suspect prescribing practice. Ansseau et al (1985) found that the benzodiazepines, methylclonazepam and lorazepam, were more effective than placebo but the former benzodiazepine was actually found to be more effective than the latter. Olajide and Lader (1987) found a three-week course of diazepam to be more effective than either buspirone or placebo. All three of the crossover design studies included placebo control but only two out of the three showed benzodiazepines superior to placebo. However, none of these three studies scheduled a washout period between treatments. Due to the possible carryover effects in these relatively short treatment trials (4-21 days), it is difficult to reach conclusions regarding efficacy.
Small sample sizes contributing to the possibility of type II errors, generated significantly to uncertainty about the conclusions. Of the parallel-design studies (Kahn et al 1986, Zung 1987, Cohn et al 1986, Feighner et al 1982, Rickels et al 1982) only five followed Solomon and Hart's (1978) recommended 40 patients per treatment group. Although the crossover-design studies overall require fewer patients, none had as many as 40 in a treatment group. Since only two (33%) of the placebo-controlled benzodiazepine studies clearly showed the drugs more effective than placebo, we have to seriously question the role of inadequate sample size confounding this finding. Inappropriately brief treatment periods, also make the data difficult to interpret. Rickels et al concluded from their data that a six-week treatment trial is as predictive of treatment outcome as longer treatment periods of either 14 or 22-weeks, thus indirectly suggesting that six weeks is a reasonable duration of a therapeutic trial in individual patients (Rickels et al 1983). Of the other parallel-design investigations, five, treated patients for six or more weeks (Castillo et al 1987, Hoehn-Saric et al 1988, .Petracca et al 1990, Strand et al 1990, McLeod et al 1990). Interestingly, of the only two studies (Castillo et al 1987, Strand et al 1990).to achieve adequate sample size and treatment duration, one found that alprazolam and clobazam were no more effective than placebo in the treatment of generalized anxiety disorder patients (Castillo et al 1987).
Conclusions. Few well designed studies have been conducted examining the efficacy of benzodiazepines in GAD diagnosed according to DSM-III criteria. Numerous problems exist within these studies that make assessment difficult. First, the DSM-III criteria used in most studies may not be specific enough to separate GAD from other anxiety disorders. Diagnosed GAD patients could possible have an unrecognized panic disorder of reduced severity. Secondly, within the GAD population, severity of illness may be a factor related to treatment response. There is some evidence that benzodiazepines are of greater benefit when used to treat patients with moderate to high levels of anxiety (Shapiro et al 1983). Additionally, Zung's (1987) data suggests that benzodiazepines are effective in dysphoric GAD patients. Third, long-term treatment beyond six weeks with benzodiazepines may not be necessary in all GAD patients. It is estimated that only half of GAD patients have a return of symptoms after the discontinuation of diazepam after either 6, 14, or 22 weeks of treatment (Rickels et al 1983). However, for the GAD population as a whole, it is still unclear whether active medication is of significantly greater benefit than placebo. Finally, benzodiazepines are not the sole treatment for GAD. For example, tricyclic antidepressants have been shown to be as efficacious as benzodiazepines in at least three studies (Kahn et al 1986, Hoehn-Saric et al 1988, McLeod et al 1990). Tricyclics appear to decrease psychological symptoms including anger, hostility, obsessions and compulsions to a greater degree than do the benzodiazepines. Therefore, the nature of symptomatology should perhaps be considered when choosing drug therapy for GAD.
Benzodiazepines are widely prescribed and that an extensive literature supports their use in patients with anxiety disorder makes it clear that these drugs are effective. What may not be so clear is whether they are the treatment of choice for patients with generalized anxiety disorder, a mild disturbance seen almost exclusively by family practitioners. Design problems, including the recruitment of subject from psychiatric populations, continue to make interpretation of the data difficult. Severity may be a factor influencing treatment response. There is, in fact, evidence that benzodiazepines are of greater benefit when used to treat either patients with moderate to high levels of anxiety or dysphoria. Recent studies of benzodiazepine treatment for generalized anxiety disorder suggests that the drugs may be effective at least in the short-term. However, tricyclic antidepressants also appear to be of value and could eventually prove superior to the benzodiazepines for GAD. Left unanswered is the question of whether drug therapy is superior to non-drug therapy for anxious family practice patients.
The ideal GAD efficacy study would be a placebo-controlled parallel design study with four treatment groups of 40-50 patients each of which would include a placebo, a benzodiazepine, a tricyclic antidepressant, and some form of psychological therapy. Patients would be recruited from family practice settings and diagnosed according to DSM-III-R criteria including the absence of panic attacks (lifelong). The active treatment should be preceded by a three week drug washout period and followed by a treatment phase lasting approximately six weeks. The HAM-A rating scale and a validated patient self-rating scale should be the basic tools for quantifying GAD symptoms.
MANAGEMENT OF THE GAD PATIENT
Do benzodiazepines have a role in the treatment of the generalized anxiety disorder patient? The above studies have shown that benzodiazepine administration alone is by no means the only or even the most effective treatment for generalized anxiety disorder. Thus their role in the treatment of the disorder needs to be placed in proper perspective. Prior to the initiation of any psychological or pharmacological treatment, organic causes for the condition need to be excluded. Hoehn-Saric and McLeod (in press) recommend a through medical work-up for most patients. Additionally, physicians should strongly encourage patients to eliminate caffeine and alcohol from their diets (Butler et al 1987).
Mild forms of generalized anxiety disorder may respond to psychological interventions alone the only pharmacotherapeutic intervention being prn use of a rapid-onset benzodiazepine for sleep. However, controlled studies using psychological therapies suggest that this treatment strategy should not be restricted to single therapy such as relaxation or cognitive therapy. Instead, Butler et al have concluded that effective psychological therapy of the generalized anxiety disorder patient should combine relaxation therapy to reduce arousal; cognitive therapy to replace anxious thoughts with positive thoughts; techniques for improving coping skills; and systematic desensitization to reverse avoidant behavior (Butler et al 1987). This finding seems ironic since family physicians treat most patients with generalized anxiety disorder and normally, they treat them with benzodiazepines rather than the psychological techniques in which they have not been trained (Beardsley et al 1988).
Severe forms of generalized anxiety disorder may require a combination of psychological and pharmacotherapies. As we have said, pharmacotherapy alone does not appear to produce a clearly superior effect on the disorder. The Hoehn-Saric study suggests an explanation why the benzodiazepines may not be the "pharmacologic panacea" for generalized anxiety disorder. They studied 60 outpatients suffering from GAD according to DSM III-R criteria (Hoehn-Saric et al 1988). Patients were randomly assigned to receive either alprazolam or imipramine and during the first two weeks of treatment, changes in symptomatology resulting from alprazolam consisted almost exclusively of reductions in somatic symptoms. Conversely, imipramine failed to improve somatic symptom ratings during the first 2 weeks of treatment but was more effective than alprazolam in improving psychic symptoms. With few exceptions, imipramine remained significantly more effective relative to alprazolam throughout the study. Therefore it appeared that alprazolam was primarily effective in lowering the somatic component of GAD, at least early on in treatment, whereas imipramine predominantly affects psychic component. Although, imipramine showed a latency in its effect, the two drugs approached comparable effectiveness overall. This study suggests that a rational pharmacotherapeutic prescribing practice may involve starting a benzodiazepine and tricyclic antidepressant simultaneously. Because the efficacy of the two drugs appears to intersect after two weeks, it probably would be reasonable at this point in time to discontinue the benzodiazepine. This treatment algorithm seems desirable since it circumvents the potential problems of benzodiazepine habituation and withdrawal.
Finally, patients should be cognizant of the fact that medication is only a temporary adjunct in the treatment of generalized anxiety disorder during which time they should be acquiring adequate coping skills. They should be encouraged to believe once effective coping mechanisms are firmly established they should be able to get along without medication.
GENERALIZED ANXIETY DISORDER AND ALCOHOLISM
It has been hypothesized that alcoholism in some patients may be the precipitated as a form of self-medication of GAD. Thus anxiolytic pharmacotherapy would be logical in these patients. However, the benzodiazepines have a high rate of addiction in the alcoholic patient population. However, since buspirone is a non-addicting anxiolytic agent effective in the treatment of GAD, it is an ideal agent to test the above hypothesis.
Two placebo-controlled double-blind studies have tested the hypothesis and achieved conflicting findings (Malcolm et al 1992, Tollefson et al 1992). Tollefson et al (1992) treated a group of 51 dually diagnosed patients with buspirone (42 ± 14 mg/d) or placebo for 24 weeks. Subjects were required to be alcohol for at least 30 days prior to entering the study. Buspirone was superior to placebo as an anxiolytic and was associated with both a reduction in the number of days alcohol craving and overall clinical improvement. Malcolm et al (1992) treated a group of 67 dually diagnosed patients with buspirone 45 or 60 mg/d (mean = 52 mg/d) or placebo for 6 months. Subjects differed from the Tollefson study in that they had just completed an inpatient detoxification program and were regarded as more severely ill. Hamilton anxiety scores declined significantly for both treatments but did not differ between treatments. Additionally, there were no differences between the treatments on all drinking parameters that included: study drop outs, time to first drink, time to 5 consecutive drinking days, time to first intoxication, and number drinks per drinking day for nonabstainers. Based on the severity of the alcoholics treated in the Malcolm study, most clinicians would conclude that it is unlikely that buspirone has much of a practical effect on drinking behavior.
REFERENCES
Ansseau M, Doumont A, Thiry D, et al (1985). Initial study of methylclonazepam in generalized anxiety disorder. Evidence for greater power in the cross-over design. Psychopharmacology 87,130.
Barlow D.H., Blanchard E.B., Vermilyea J.A., Vermilyea B.B., DiNardo, P.A. (1986). Generalized anxiety and generalized anxiety disorder: description and reconceptualization. Am J Psychiatry 143,40.
Beardsley RS, Gardocki GJ, Larson DB, et al (1988). Prescribing of psychotropic medication by primary care physicians and psychiatrists. Arch Gen Psychiatry 45:1117-9.
Butler G, Gelder M, Hibbert G, et al (1987). Anxiety management - developing effective strategies. Behav Res Therap 25:517-22.
Castillo A, Sotillo C, Mariateegui J (1987). Alprazolam compared to clobazam and placebo in anxious outpatients. Neuropsychobiology 18,189.
Cohn JB, Bowden CL, Fisher JG, et al. (1986). Double-blind comparison of buspirone and clorazepate in anxious outpatients. Am J Med 30(suppl 3B):10.
Cohn JB, Wilcox SC (1986). Low sedation potential of buspirone compared with alprazolam and lorazepam in the treatment of anxious patients: a double- blind study. J Clin Psychiatry 47,409.
Elie R, Lamontagne Y. (1984). Alprazolam and diazepam in the treatment of generalized anxiety. J Clin Psychopharmacol 4:125.
Fabre LF (1987). Double-blind comparison of buspirone with diazepam in anxious patients. Curr Ther Res 41:751.
Feighner JP, Merideth CH, Hendrickson GA. (1982). A double-blind comparison of buspirone and diazepam in outpatients with generalized anxiety disorder. J Clin Psychiatry 43(Sec 2):103.
Fontaine R, Annable L, Chouinard G, et al (1983). Bromazepam and diazepam in generalized anxiety: A placebo-controlled study with measurement of drug plasma concentrations. J Clin Psychopharmacol 3,80.
Hoehn-Saric R, McLeod DR, Zimmerli WD (1988). Differential effects of alprazolam and imipramine in generalized anxiety disorder: somatic versus psychic symptoms. J Clin Psychiatry 49,293.
Hoehn-Saric R, McLeod DR (in press). Management of generalized anxiety disorder. In Coryell W, Winokur G (eds), The Clinical Management of Anxiety Disorders. Oxford University Press, New York.
Kahn RJ, McNair DM, Lipman RS, et al (1986). Imipramine and chlordiazepoxide in depressive and anxiety disorders. II. Efficacy in anxious outpatients. Arch Gen Psychiatry 43,79.
Lapierre YD, Tremblay A, Gagnon A, et al. (1982). A therapeutic and discontinuation study of clobazam and diazepam in anxiety neurosis. J Clin Psychiatry 43:372.
Malcolm R, Anton RF, Randall CL, et al (1992). A placebo-controlled trial of buspirone in anxious inpatient alcoholics. Alcohol Clin Exp Res 16:1007-13.
McLeod DR, Hoehn Saric R, Zimmerli WD, et al (1990). Treatment effects of alprazolam and imipramine; physiological versus subjective changes in patients with generalized anxiety disorder. Biol Psychiatry 28:849-61.
Noyes R., (1988). Proposed revision of the DSM-III classification of anxiety disorders. In Noyes R, Roth M, Burrows G (eds) Handbook of Anxiety, Volume 2; . Elsevier, Amsterdam, 81-107.
Olajide D, Lader M (1987). A comparison of buspirone, diazepam, and placebo in patients with chronic anxiety states. J Clin Psychopharmacol 7,148.
Petracca A, Nisita C, McNair D, et al. (1990). Treatment of generalized anxiety disorder. Preliminary clinical experience with buspirone. J Clin Psychiatry 51
Rickels K, Weisman K, Norstad N, et al. (1982). Buspirone and diazepam in anxiety: a controlled study. J Clin Psychiatry 43 (sec 2):81.
Rickels K, Case G, Downing RW, et al (1983). Long-term diazepam therapy and clinical outcome. JAMA 250,767.
Ross CA, Matas M (1987). A clinical trial of buspirone and diazepam in the treatment of generalized anxiety disorder. Can J Psychiatry 32,351.
Shapiro AK, Stuening EL, Shapiro E, et al (1983). Diazepam: How much better than placebo? J Psychiat Res 17,51.
Solomon K., Hart R (1978). Pitfalls and prospects in clinical research on antianxiety drugs: benzodiazepines and placebo - a review. J Clin Psych 39,823.
Strand M, Hetta J, Rosen A, et al (1990). A double-blind controlled trial in primary care patients with generalized anxiety: a comparison between buspirone and oxazepam. J Clin Psychiatry 51(suppl 9)40-5.
Tansella CZ (1980). The long-term treatment with benzodiazepines: suggestions for further research. Int Pharmacopsychiatry 15,99.
Tollefson GD, Montague-Clouse J, Tollefson S (1992). Treatment of comorbid generalized anxiety in a recently detoxified alcoholic population with a selective serotonergic drug (buspirone). J Clin Psychopharmacol 12:19-26.
Tyrer P, Owen R (1984). Anxiety in primary care: is short-term drug treatment appropriate? J Psychiat Res 18,73.
Tyrer P. (1986). Classification of anxiety disorders: a critique of DSM-III. J Affective Disord 11,99.
Tyrer P Murphy S, Kingdon, et al (1988). The Nottingham study of neurotic disorder: comparison of drug and psychological treatments. Lancet 2:235-40.
Zung WWK, Mendels J, Tillman S, et al. (1986). A comparison of the incidence of sedation in anxious outpatients treated with diazepam and prazepam. Curr Ther Res 39:480.
Zung WWK. (1987). Effect of clorazepate on depressed mood in anxious patients. J Clin Psychiatry 48:13.