Abstract
Evaluation of: Martin JLR, Sainz-Pardo M, Furukawa TA et al. Benzodiazepines in generalized anxiety disorder: heterogeneity of outcomes based on a systematic review and meta-analysis of clinical trials. J. Psychopharmacol. 21, 774–782 (2007)Citation. The benzodiazepines have been extensively prescribed for decades for vague indications such as anxiety, sleeplessness and muscle tension. Despite increasing knowledge of their adverse effects, such as sedation, psychomotor and cognitive impairment, and dependence on long-term use, and the recent advent of better alternatives, their use continues largely unabated. The paper under review assesses the sparse high-quality data related to efficacy (denoted by the dropout rate for failure to respond), effectiveness (dropout rate for any reason) and dropout for adverse effects. The conclusion is that efficacy was significantly higher for the drugs as compared with placebo; by contrast, no convincing evidence was found of any short-term effectiveness: and adverse effects were 1.5-times more frequent in the drug-treated patients. Various reasons for these results are discussed. I point out the changes in diagnostic criteria over the years and the lack of accepted methods of assessing estimates of effectiveness in clinical practice. Excessive prescribing of these controversial drugs is likely to continue.
Keywords::
Generalized anxiety disorder (GAD) is defined in various successive versions of the American Psychiatric Association’s diagnostic schema. It is characterized by marked worry and anxiety sustained over 6 months or more, and is accompanied by severe impairments of personal, occupational and social life Citation[2]. Comorbidity is common, indeed the rule, particularly with depressive disorders Citation[3]. The prevalence of GAD is generally estimated as approximately 4–6% Citation[4].
The benzodiazepines were developed in the 1930s, but not actually introduced until the late 1950s. Very quickly, they became widely used for a variety of common but ill-defined indications, such as anxiety, insomnia and muscle tension. Because of this early introduction, they were inadequately evaluated by modern standards. The indications were not stringently defined, doses were often wide or unspecified, and measures of outcome vague or ad hoc. Risks included sedation, cognitive and psychomotor impairment, and paradoxical responses. In particular, long-term efficacy and safety were not assessed in any meaningful way. The clinical impression arose that these drugs were prone to tolerance and their effects quickly waned. Nevertheless, long-term use was common. Gradually it became clear that such use, say beyond 3–6 months, was associated in 20–30% of users with a clear withdrawal syndrome when dosage reduction or discontinuation was attempted Citation[5]. Dependence without dosage escalation had supervened. Despite these warning signals, long-term efficacy was hardly addressed.
The doubts about the risk:benefit ratio of long-term benzodiazepine therapy have now been generalized back to questions concerning their short-term use, for less than 4 weeks, the duration of usage for which they are widely licensed.
Furthermore, the advent of newer treatments for GAD, such as the selective serotonin reuptake inhibitors (SSRIs), serotonin–norepinephrine reuptake inhibitors (SNRIs) and, recently, pregabalin, has re-awakened the need to quantify the efficacy of this large class of drugs, and to set their practical clinical effectiveness against their risks, particularly when treatment becomes prolonged Citation[6,7].
The study under review sought to use systematic review and meta-analyses to rectify this omission and to show the ‘real’ effectiveness of these drugs and not merely their efficacy under formal but artificial clinical trial parameters.
Methods & results
The authors searched Medline, Embase and the Cochrane Controlled Trials Register for papers in all languages describing double-blind randomized, placebo-controlled trials concerned with the benzodiazepines in the treatment of GAD. The primary outcome for ‘effectiveness’ was withdrawal for any reason; for ‘efficacy’, withdrawal for lack of efficacy; and for ‘side effects’, withdrawal owing to adverse events. Of over 1000 studies screened, only 23 met their strict criteria for inclusion. A total of 14 were 4-week studies; 13 involved diazepam, seven lorazepam and four alprazolam.
Pooled analysis indicated less risk of treatment discontinuation owing to lack of efficacy for active medication, compared with placebo (relative risk [RR]: 0.29; 95% confidence interval [CI]: 0.18–0.45; p < 0.00001). Nevertheless, risk of discontinuation for all causes (effectiveness) showed no clear advantage for the tranquillizers over placebo (RR: 0.78; CI: 0.62–1.00; p = 0.05). The more recent the study, the smaller was the positive effect of the drugs. Finally, the relative risk of dropout for adverse effects was 1.54, an additional risk of 50% for the benzodiazepines over the placebo.
The investigators concluded that their systematic review did not find convincing evidence of the short-term effectiveness of the benzodiazepines in the treatment of GAD. The waning efficacy over time was attributed to possible publication bias, to improving quality of design and reporting, or to increased placebo response.
Discussion & significance
The poor quality of the bulk of the trials evaluated reflects the length of time that the benzodiazepines have been available. Initially only a cursory appraisal of a relatively few patients was adequate to obtain marketing permission in most jurisdictions. The first benzodiazepine, chlordiazepoxide, was introduced within months of its first administration to a human. Nevertheless, if the drugs were really effective, the later more rigorous trials should still have shown clear effectiveness as well as formal efficacy. One clue lies in some data presented by the authors but not discussed. They show that effectiveness seems to wane as the Diagnostic and Statistical Manual of Mental Disorders (DSM) diagnostic criteria changed. From DSM-III in 1980 to DSM-IV-Text Revision in 2000, the duration of symptoms necessary to establish the diagnosis was prolonged from 1 to 6 months. Also, the earlier criteria established GAD at the bottom of a hierarchy, whereas the later criteria sanctioned comorbidity. Both of these changes would militate against the effectiveness of the benzodiazepines, as their effects are known to tolerate over a few weeks, and comorbidity typically involves depression, against which they are ineffective.
The question arises as to how valid dropout rates are as surrogate markers for efficacy, effectiveness and safety. Anyone who has been involved in a clinical trial is aware of the complexity of the issues. Withdrawal for any reason encompasses lack of efficacy and adverse effects, but also administrative reasons and withdrawal of consent. These will not be the same for active medication and for placebo. Often the reason for dropout is compound, with the investigator throwing in his hand when a patient who is failing to respond adequately develops an adverse effect that by itself would not justify discontinuation. Despite these reservations a general impression of clinical usefulness can be gleaned. But other outcome variables may be more informative. The usual one is effect size Citation[8]. Also, the rate of remission to a predefined criterion is increasingly favored by regulatory bodies, such as the European Medicines Agency (EMEA). Withdrawal for any reason is not usually accepted as the best estimate.
A second methodological issue concerns the validity of pooled and meta-analyses Citation[9]. This approach has become increasingly popular, and has been espoused by statisticians and clinicians alike. The problem is achieving consensus on the criteria for including the relevant studies. As the present example shows, often only a tiny proportion of the large number of studies identified as meriting consideration actually makes it over the final hurdle. This is a particular problem when the treatment under consideration has been extant for decades. The search strategy, which is a critical factor in valid meta-analyses, was quite appropriate in this study. Although the technique has been criticized as combining apples and pears, the final analysis of fruit is better than losing the ability to discern major trends in the individuals in the bowl!
Expert commentary & five-year view
The increasing availability of remedies for GAD, including drugs coming off patent and the use of nonpharmacological methods of treatment such as cognitive–behavioral therapy Citation[10] or even exercise Citation[11] raises the question of the current relevance of reviewing the effectiveness of the benzodiazepines. Surely, it can be argued that they are obsolescent. This is not so – far from it. A general practitioner prescribing survey showed that new prescriptions for anxiolytic benzodiazepines were given to three-quarters of a million patients in the UK in 2005 [Donoghue and Lader, in preparation]. Up to 15% of the over 70-year-olds received continuous treatment for more than 3 months. Up to 6% of the over 80-year olds were given medication continuously for more than 1 year: these are the people particularly at risk of adverse events such as cognitive and psychomotor impairment, even to the point of being misdiagnosed as dementing Citation[12]. Over all age groups, the mean duration of benzodiazepine treatment was approximately 25–30 days, but it was 70 days if a codiagnosis of depression was made. Similarly, in the USA, benzodiazepine use has hardly altered over the past 12 years, and they have not been superseded by the SSRIs or the SNRIs Citation[13]. This continuing high usage has been justified as having a rational basis, although the effectiveness of these drugs was glossed over Citation[14].
Will the benzodiazepines start to lose their popularity? Although this and other papers have failed to establish their effectiveness, they are found by many patients to have a rapid and gratifying therapeutic effect, and many patients are reluctant to stop them. Papers like this one under review are important in that they challenge the whole basis for acceptance of a therapeutic agent. Together with continuing concern about the risks of the benzodiazepines, both short and long term, this will ensure that the debate will continue about their place, if any, in the modern armamentarium of anti-anxiety treatments. Although the newer drugs such as the SSRIs and the SNRIs have adverse effects, these are better documented than those with the benzodiazepines. In addition, their efficacy and effectiveness have been established using modern, valid and sensitive trial techniques. Nevertheless, I predict that the present prescribing pattern will persist for some years to come.
Financial & competing interests disclosure
The author consults with Lundbeck, Pfizer and Servier. He also advises various lawyers on drug matters.
The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in the production of this key paper evaluation manuscript.
References
- Martin JLR, Sainz-Pardo M, Furukawa TA et al. Benzodiazepines in generalized anxiety disorder: heterogeneity of outcomes based on a systematic review and meta-analysis of clinical trials. J. Psychopharmacol.21, 774–782 (2007).
- Kessler RC, DuPont RL, Berglund HU et al. Impairment in pure and comorbid generalized anxiety disorder and major depression at 12 months in two national surveys. Am. J. Psychiatry156, 1915–1923 (1999).
- Nutt D, Argyropoulos S, Hood S et al. Generalized anxiety disorder; a comorbid disease. Eur. Neuropsychopharmacol.16(Suppl. 2), 109–118 (2006).
- Wittchen HU, Zhao S, Kessler RC et al. DSM-III-R generalized anxiety disorder in the National Comorbidity Survey. Arch. Gen. Psychiatry51, 355–364 (1994).
- Schweizer E, Rickels K. Benzodiazepine dependence and withdrawal: a review of the syndrome and its clinical management. Acta Psychiatrica. Scand. Suppl.393, 95–101 (1998).
- Rynn MA Brawman-Mintzer O. Generalized anxiety disorder: acute and chronic treatment. CNS Spectr.9, 716–723 (2004).
- Struzik L, Vermani M, Coonerty-Femiano A et al. Treatments for generalized anxiety disorder. Expert Rev. Neurother.4, 285–294 (2004).
- Mitte K, Noack P, Steil R et al. A meta-analytic review of the efficacy of drug treatment in generalized anxiety disorder. J. Clin. Psychopharmacology25, 141–150 (2005).
- Everitt BS, Wessely S. Clinical Trials in Psychiatry (Second Edition), John Wiley & Sons, Chichester, UK (2008).
- Hofmann SG, Smits JAJ. Cognitive–behavioral therapy for adult anxiety disorders: a meta-analysis of randomized placebo-controlled trials. J. Clin. Psychiatry69, 621–432 (2008).
- National Institute for Clinical Excellence (NICE). Clinical Guidelines for the Management of Anxiety (Panic Disorder, With or Without Agoraphobia and Generalised Anxiety) in Adults in Primary, Secondary and Community Care. NICE, London, UK (2004).
- Barker MJ, Greenwood KM, Jackson M et al. Cognitive effects of long-term benzodiazepine use. CNS Drugs18, 37–48 (2004).
- Vasile RG, Bruce SE, Goisman RM et al. Results of a naturalistic longitudinal study of benzodiazepine and SSRI use in the treatment of generalized anxiety disorder and social phobia. Depression Anxiety22, 59–67 (2005).
- Stahl SM. Don’t ask, don’t tell, but benzodiazepines are still the leading treatments for anxiety disorder. J. Clin. Psychiatry63, 756–757 (2002).