No studies directly addressed any of the research questions on the safety, comparative efficacy, and cost-effectiveness of using benzodiazepines in older adults to manage disruptive behaviour or treat anxiety. The literature search identified a total of 596 citations, and 15 publications were included in this report (Appendix 3). Six relevant systematic reviews addressed the association between benzodiazepine use in older adults and falls,7,8 cognitive decline,9 fractures,10 cognitive and psychomotor adverse events,11 and mortality.12 Nine evidence-based guidelines and recommendations addressed the use of benzodiazepines by older adults for dementia,13–16 mood and behavioural disorders,17 panic disorder,18 and sleep disorders.19–21
4.1. Systematic reviews and meta-analyses
4.1.1. Risk of falls
In 2009, Woolcott et al.7 conducted a systematic review and meta-analysis to evaluate the impact of nine classes of medication on the risk of falls in older adults. The drugs that were investigated included benzodiazepines, antidepressants, antihypertensives, diuretics, beta-blockers, sedatives/hypnotics, neuroleptics/antipsychotics, narcotic analgesics, and non-steroidal antiinflammatory agents. The report was an update of three previously published reviews: two meta-analyses by Leipzig et al. (1999)22,23 and a systematic review by Hartikainen et al. (2007).8 References were obtained from the three reviews and a literature update was done to obtain additional studies. The updated search involved multiple databases. Many of the studies that were identified by Hartikainen et al.8 could not be included in the meta-analysis because the data were presented in a format that did not allow for pooling. Patient and trial characteristics were documented in the report. A risk of bias assessment was done in duplicate using an appropriate method for randomized and non-randomized studies (Downs and Black checklist24). The results of the risk of bias assessment were not presented, and it was not stated to what extent these results were considered in the review’s analysis and conclusions. Additional limitations with this review were the exclusion of non-English language publications, uncertainty about the use of duplicate reviewers for study selection and data extraction, and a lack of formal assessment of statistical heterogeneity. The population of interest for this review was patients older than 60 years. The authors did not report the inclusion criteria that were used in each primary study. However, the mean age of participants was older than 65 years in all studies.
The meta-analysis included 11 non-randomized studies reporting the effects of benzodiazepines (four cross-sectional, two case-control, and five cohort). The pooled estimate of effect for these studies showed a statistically significant increase in the risk of falls among patients taking benzodiazepines (odds ratio [OR] 1.57; 95% credible interval [CrI] 1.43 to 1.72). The review included nine non-randomized studies reporting the effects of antidepressants (four cross-sectional, three case-control, and two cohort) and the pooled estimate of effect was statistically significant (OR 1.68; 95% CrI 1.47 to 1.91). The risk of falls after the use of benzodiazepines and antidepressants was similar to that reported for antipsychotics and sedatives/hypnotics (). The results from the Bayesian meta-analysis were compared with those obtained using a frequentist approach and were similar in direction and magnitude. Overall, the authors concluded that the use of psychotropic medications is associated with a statistically significant increase in falls among older adults.
Risk of Falls in Older Adults Using Nine Drug Classes, from Woolcott et al.
In 2007, Hartikainen et al.8 conducted a systematic review to assess the association between medication use and the risk of falls in older adults. Limitations with the reporting of this systematic review include a failure to provide the eligibility criteria for including studies, methods for study selection and data extraction, and instruments and methods for risk of bias assessment. The authors specified that the population of interest was adults older than 60 years. Four of 29 included studies had a population younger than 65 years.
Seventeen studies reported an association between benzodiazepines and fall or fall-related fractures, and three studies reported no association. Fourteen of these studies included patients who were 65 years or older, and three included patients who were at least 60 years of age. The authors noted that the risk of falls was increased after a new prescription and with long-term use, and was independent of the product’s half-life (for example, long-acting versus short-acting). Twelve studies reported an association between the use of antidepressants and the risk of falling, and five studies found no association. The authors concluded that the use of psychotropic drugs seems to be associated with an increased risk of falling. However, they noted that the available evidence was often low quality and inconsistent.
4.1.2. Risk of cognitive decline
In 2005, Verdoux et al.9 conducted a systematic review to investigate if an exposure to benzodiazepines is associated with an increased risk of cognitive decline. Primary studies were selected for inclusion if they met the following criteria: published in English in a peer-reviewed journal; involved participants who were recruited from the general population; the length of follow-up was at least one year in duration; documented benzodiazepine use at baseline and follow-up; standardized cognitive assessment at baseline and follow-up; and the association between benzodiazepine use and change in cognitive performance explored. There are several methodological limitations with this review: the literature review was conducted in one database (MEDLINE from 1966 up to April 2004), there was a failure to fully report the methods that were used for study selection and data extraction, there was no mention of a formal risk of bias assessment, and non-English language publications were excluded.
The review included six prospective cohort studies involving patients older than 60 years. All studies compared current or former users of benzodiazepines with non-users. Overall, the results of the individual studies were inconsistent with two studies reporting that benzodiazepines were associated with less risk of cognitive decline, two studies reporting no statistically significant association, and three studies reporting an increased risk of cognitive decline. Furthermore, the studies that reported a statistically significant increase in cognitive decline differed in the at-risk populations (new, chronic, or former users). The authors stated that the discrepancies in these findings prevent conclusions from being made about the direction and magnitude of a potential association between benzodiazepine use and cognitive decline.
The inconsistent findings may be due to the heterogeneity in study design. The studies varied in sample size (range of 242 patients to 3,309 patients), length of follow-up (range of two years to six years), adjustment of confounding factors, and methods for outcome assessment (for example, Short Portable Mental Status Questionnaire or categorical diagnosis of dementia by a physician). The authors noted that selection bias may be present in the included studies because all failed to provide details about the proportion of individuals who refused to participate. This could favour the inclusion of patients who had less exposure to benzodiazepine use and less cognitive decline, or less exposure to benzodiazepine use or less cognitive decline, thereby underestimating the strength of association between the drug and the primary outcome. Well-designed studies are needed to accurately assess the impact of benzodiazepine use on cognitive decline in older adults.
4.1.3. Risk of fractures
In 2007, Takkouche et al.10 conducted a systematic review and meta-analysis to assess the risk of fractures among users of psychotropic drugs. The drug classes that were included in the review were benzodiazepines, antidepressants, non-barbiturate antiepileptic drugs, barbiturate antiepileptic drugs, antipsychotics, hypnotics, and opioids. The literature search involved many databases and was conducted without language restriction. Published studies were included if they met the following criteria: original data were presented from case-control or cohort studies, the outcome of interest was defined as fracture (falls not followed by fractures were not included), the exposure of interest was a psychotropic medication, and relative risk (RR) estimates with confidence intervals (CIs) were provided or enough data were provided to calculate them. Studies that did not provide adjusted or crude data of RR and cross-sectional studies were excluded. The study selection was performed independently by two reviewers. An appropriate instrument was used to assess the risk of bias of each included study; however, it was unclear if this was done by multiple independent reviewers. Publication bias was assessed using appropriate methods (examination of funnel plots and Egger’s et al.’s regression test25).
This review did not specify an age threshold in the eligibility criteria and included studies regardless of patient age. There were 23 studies that investigated the association between benzodiazepine use and the risk of fractures, including 16 case-control studies and seven cohort studies. Fourteen studies specified that patients had to be older than 65 years, two involved patients at least 60 years of age, one included patients at least 55 years, and six did not provide an age threshold for inclusion. In comparison with non-users, the RR for fractures in benzodiazepine users was 1.34 (95% CI 1.24 to 1.45) based on random effects meta-analysis. These results were similar when the analysis was stratified according to the risk of bias assessment, study design, setting, and use of short-acting or long-acting agents. Drugs from the benzodiazepine class in the reference case meta-analyses included those with different pharmacokinetic properties. This was reflected in the high statistical heterogeneity (Ri 0.57, P = 0.00001) for the reference case, which was reduced when the trials were stratified according to short-acting (Ri 0.3, P = 0.23) and long-acting agents (Ri 0.3, P = 0.19).
Fifteen studies examined the association between the use of antidepressants and the risk of fracture. Pooling showed a statistically significant increase in fracture risk with antidepressant use relative to non-use (RR 1.60; 95% CI 1.38 to 1.86]. This finding was consistent when the data were stratified by study design, setting (hospital or general population), or type of antidepressant. The authors reported that there was no evidence of publication bias for the benzodiazepines and for the antidepressants. The results of these meta-analyses appear in Appendix 4. The authors concluded that the use of psychotropic medications may be associated with the development of fractures and that larger prospective studies could be designed to provide a more accurate assessment of this clinically important outcome.
4.1.4. Risk of Mortality
In 2009, Charlson et al.12 conducted a systematic review examining the association between benzodiazepine use and mortality. The literature search involved many databases. A formal risk of bias assessment was conducted using a comprehensive checklist. There are several methodological limitations with this review, including the failure to provide a clear description of eligibility criteria, failure to fully report the methods that were used for study selection and data extraction, and the exclusion of non-English language publications.
The authors identified one population-based survey and three cohort studies that assessed the risk of mortality in older adults (). The three cohort studies reported that there was no statistically significant association between the use of benzodiazepines and an increase in all-cause mortality among older adults. One study also assessed the risk of fracture-related mortality and reported a RR of 2.71 (95% CI 0.37 to 19.76) for benzodiazepine use compared with no use. The population-based study provided a comparison of the risk of mortality related to benzodiazepine poisoning in patients at least 60 years of age with those younger than 60 years. The authors reported a statistically significant increased risk of mortality in patients who are older than 60 years (RR 7.1 [95% CI 3.2 to 15.5]). The number of events in each study was not reported. Therefore, the non-statistically significant findings may be due to inadequate statistical power. Given the observational study designs and the large number of unadjusted confounders, the authors concluded that there are insufficient data to accurately assess the relationship between benzodiazepine use and mortality.
Risk of Mortality Among Patients Taking Benzodiazepines, Charlson et al.
4.1.5. Risk of adverse events in the treatment of insomnia
In 2005, Glass et al.11 conducted a systematic review and meta-analysis to evaluate the risks and benefits of using sedative hypnotics for the treatment of insomnia in older adults. Active and placebo-controlled randomized studies were eligible for inclusion if they met the following criteria: published in English; mean age of participants was at least 60 years; investigated the use of a sedative hypnotic given for at least five consecutive nights; included a washout period after previous drug treatments; and excluded patients with psychiatric disorders, concurrent use of drugs affecting the central nervous system, and severe or acute physical illnesses that could disrupt sleep. Studies of barbiturates and chloral hydrate were excluded. The literature search involved many databases. Study-level risk of bias was assessed using an appropriate method (Jadad scale26). However, it is unclear if these findings were used in the meta-analysis. Study selection, data extraction, and risk of bias assessment were conducted independently in duplicate or triplicate. Publication bias was assessed using examination of funnel plots and Egger’s regression test.25 The methods that were used to pool data were appropriate and well reported. However, studies were pooled at the drug class level. Therefore, the benzodiazepine group consisted of short-acting agents (triazolam, midazolam, brotizolam, lormetazepam, loprazolam), intermediate-acting agents (nitrazepam, temazepam, flunitrazepam), and long-acting agents (flurazepam and quazepam).
Overall, this is a well-conducted meta-analysis with a low risk of bias.27 However, this study has limited applicability to the research questions because the trials focused on the short-term treatment of insomnia as opposed to the treatment of anxiety, the primary outcomes for clinical efficacy in this review are related to sleep (quality of sleep, amount of sleep, and number of awakenings), and the primary meta-analyses were done by pooling data for all sedative hypnotics (benzodiazepines, benzodiazepine-receptor agonists, and an antihistamine). Despite this apparent indirectness,28 there is potentially useful information about the adverse events observed with the short-term use of benzodiazepines and similar sedative hypnotics ().
Adverse Events for Benzodiazepines in the Treatment of Insomnia.
Of the 24 randomized controlled trials that were included in the review, 21 included a benzodiazepine treatment group. Sedative hypnotics were associated with more cognitive, psychomotor, and overall adverse events compared with placebo. There were no statistically significant differences between benzodiazepines and benzodiazepine-receptor agonists (zopiclone, zolpidem, and zaleplon). The authors concluded that the benefits associated with sedative use are marginal and outweighed by the risks, especially if patients are considered to be at a higher risk for falls or cognitive impairment.
4.2. Guidelines and Recommendations
Nine evidence-based guidelines and recommendations addressed the use of benzodiazepines by older adults for dementia,13–16 mood and behavioural disorders,17 panic disorder,18 and sleep disorders19–21 (Appendix 5). The recommendations in each guideline focus on patient safety and the potential adverse effects of these drugs in older adults. Two guidelines19,20 caution that the adverse effects of benzodiazepines generally seem to be worse in older adults. One of these guidelines recommends that dosages be reduced for older adults and that short- or intermediate-acting agents are preferable. This is consistent with the product monographs for these agents.2
Dementia, particularly Alzheimer disease and vascular dementia, can be associated with behavioural and psychological symptoms, including depression, agitation, psychosis, wandering, aggression, and incontinence. The Social Care Institute for Excellence and the National Institute for Health and Clinical Excellence have published a guideline on the treatment and diagnosis of dementia that addresses the use of benzodiazepines by these patients.14 The guideline was developed with an objective of providing advice on supporting people with dementia and their care givers in health and social care. This guideline was developed using a rigorous methodology in accordance with the Appraisal of Guidelines for Research and Evaluation (AGREE) criteria.29 The guideline offers recommendations on training and competencies for health care professionals prescribing medications for the management of disruptive behaviour in people with dementia. The guideline noted that for patients with Alzheimer disease or vascular dementia and clinically significant agitation, there is moderate-quality evidence suggesting that benzodiazepines administered through intramuscular injection may have beneficial effects in reducing psychotic symptoms and aggression or agitation that outweigh the risk of adverse events. The Scottish Intercollegiate Guidelines Network’s (SIGN) guideline for the management of patients with dementia15 was developed using rigorous methods. The guideline states that no systematic reviews or randomized controlled trials examined the use of benzodiazepines in the management of dementia symptoms.
The search identified two guidelines from Canadian organizations that were developed with rigorous methodology. The Canadian Coalition for Seniors’ Mental Health published guidelines addressing the assessment and treatment of mental health issues in long-term care with a focus on managing mood and behavioural symptoms.17 The guideline recommends that health care professionals weigh the potential benefits of pharmacologic intervention against the potential for harm. Short- or intermediate-acting benzodiazepines are noted as a therapeutic option for the treatment of long-term care residents with severe behavioural symptoms. The Guidelines and Protocols Advisory Committee in British Columbia prepared guidance on the recognition, diagnosis, and management of cognitive impairment in the elderly.13 The guideline addresses interventions for the behavioural and psychological symptoms of dementia and states that benzodiazepines are not recommended because of their high potential for adverse events such as confusion and falls.
Two relevant guidelines were published by the American Psychiatric Association (APA).16,18 Both were developed using a rigorous methodology, based on the criteria of the Appraisal of Guidelines for Research and Evaluation instrument.29 One guideline focused on the treatment of patients with dementias.16 In the section about the treatment of psychosis and agitation, the following advice is provided:
“Data demonstrating benefit from BZDs are modest, but BZDs occasionally have a role in treating patients with prominent anxiety [III] or on an as-needed basis for patients with infrequent episodes of agitation or for those who require sedation for a procedure such as a tooth extraction or a diagnostic examination [II]. Adverse effects of BZDs include sedation, worsening cognition, delirium, increased risk of falls, and worsening of breathing disorders. Lorazepam and oxazepam, which have no active metabolites, are preferable to agents with a longer half-life such as diazepam or clonazepam [III].” (p 13)16
The second guideline focuses on the treatment of patients with panic disorder. This guideline advises that an antidepressant is the recommended first-line pharmacotherapy for geriatric patients with panic disorder, citing co-occurring mood disorders as the primary reason. Furthermore, the APA recommends that benzodiazepine use be avoided whenever possible, indicating that the use of long half-life benzodiazepines and long-term use in general can be problematic in older adults. If a benzodiazepine is used, the APA notes that these patients may be at a higher risk for falls and fractures and that monitoring is needed.
The National Institutes of Health in the United States published a State-of-the-Science Conference Statement on the manifestations and management of chronic insomnia in adults.20 This statement was prepared using a rigorous method that involved a systematic review. However, this document offers little information about the use of benzodiazepines by older adults other than noting that adverse effects seem to be worse in the elderly. The American Medical Directors Association19 released a guideline addressing the treatment of sleep disorders in long-term care. The guideline recommends that the continuous use of benzodiazepines be discouraged because of the risk of side effects, physiological tolerance, and adverse effects upon discontinuation. Neither the age range to which the recommendation applies nor the level of evidence was provided in support of the statement. In 2007, Montgomery and Lilly21 prepared a systematic review and offer clinical guidance about the treatment of insomnia in patients 65 years age and older. The review was prepared in accordance with the BMJ’s Clinical Evidence method.30 The authors concluded that the use of benzodiazepines may improve sleep outcomes, but are likely to cause adverse events in patients 65 years and older.
4.3. Limitations
Six systematic reviews and meta-analyses addressed the association between benzodiazepine use in older adults and falls,7,8 cognitive decline,9 fractures,10 cognitive and psychomotor adverse events,11 and mortality.12 However, no studies addressed the research questions about the safety, comparative efficacy, and cost-effectiveness of using benzodiazepines in older adults to manage disruptive behaviour or treat anxiety.
A limitation of the available evidence is the lack of clarity and consistency in defining older adults. Hartikainen et al.,8 Woolcott et al.,7 Verdoux et al. ,9 and Glass et al.11 all specified that the population of interest was patients older than 60 years of age. The World Health Organization notes that most developed countries considered a cut-off age of 65 years when defining an elderly or older person. The United Nations’ cut-off age is 60 years.31 The studies in this review did not provide justifications for selecting 60 years of age as an appropriate threshold. Takkouche et al.10 did not specify an age threshold in their eligibility criteria and included studies regardless of patient age. Most of the studies (70%) that were included in the review by Takkouche et al.10 involved patients older than 60 years. However, there is uncertainty about the population in the remaining studies. This could limit the generalizability of the findings to a population entirely comprised of older adults. No studies addressed the potential for withdrawal, abuse, and addiction in older adults using benzodiazepines.
The included reviews did not provide the indication for benzodiazepine use in the patient populations. Therefore, the quality of evidence in this rapid review is limited by a lack of direct relevance to the population of interest for this review (i.e., older adults with anxiety, behavioural problems, and/or depression). In addition, there were no economic evaluations of the use of benzodiazpines in older adults. In 2005, Mamdani et al.6 speculated that, although benzodiazpines are inexpensive, they may carry an additional hidden cost in the increased health care resources that are used to treat adverse events. Given the increased adverse events reported in the systematic reviews and the high prevalence of benzodiazepine use in older adults, formal economic evaluations of each indication would be beneficial.
The evidence base for the systematic reviews consisted almost entirely of observational studies. Most studies made statistical adjustments for a range of potential confounding factors. However, the dosage of benzodiazepines was not addressed in the meta-analyses because of a lack of reporting in the primary studies.7,10 Confounding by indication may have occurred because patients who were prescribed benzodiazepines may have underlying symptoms (for example, anxiety or agitation) that carry an increased risk of falls.10 Benzodiazepines in the meta-analyses included those with different pharmacokinetic characteristics (short-, intermediate-, and long-acting). This was noted as a source of statistical heterogeneity in one report. However, a subgroup analysis that stratified results according to benzodiazepine properties showed that the risk of fractures was similar in each group.10 The lack of consistency about the definitions of benzodiazepine users and non-users could affect the events rates for either comparator.12
