Hypnotics and sedatives

Recommendations

Type	Recommendation
When to deprescribe
CBR	We suggest deprescribing be offered to older people taking long-term (beyond four weeks) benzodiazepines for insomnia as the risk of dependence and other potential harm (e.g. falls, fractures, impaired cognition) generally outweighs the potential benefits, except in special circumstances (e.g. palliative care).
Ongoing treatment
CBR	Given the harms of long-term use are likely to outweigh the benefits in most cases, we generally suggest against the use of long-term benzodiazepines for insomnia in older people and trial on-demand or intermittent use at the lowest effective dose in addition to appropriate investigation to identify and subsequently treat a cause. If symptoms are chronic and persistent, we suggest considering appropriate non-pharmacological therapies and/or safer alternatives for symptoms, provided this aligns with the individual's goals and preferences, following informed consent.
How to deprescribe
CBR	We suggest individualising the tapering schedule and adjusting it according to the individual's response. In general, we suggest reducing the dose by 25% to 50% every one to four weeks, ensuring the absence of withdrawal symptoms or reduced sleep quality before initiating further tapering. Once half the lowest standard dose formulation is reached, we suggest ceasing completely. However, smaller dose reductions may be appropriate or preferred by some individuals, particularly as lower doses are approached.
GPS	Healthcare providers should consider and offer adequate non-pharmacological management options (e.g. good sleep practices) as appropriate (ungraded good practice statement).
Monitoring
CBR	We suggest closely monitoring for withdrawal symptoms and/or symptoms indicative of relapse, including worsening sleep quality, changes in psychological or physical health status (e.g. anxiety, and agitation) and quality of life every one to two weeks following each dose adjustment until at least four weeks after the medicine is fully ceased if practical. After this initial period, we suggest monthly monitoring for at least three months, followed by monitoring every six months thereafter. However, this should be tailored based on individual factors such as their preferences, responses and tolerance to deprescribing. If in-person visits are impractical, we suggest advising people to report symptoms as needed.

CBR, consensus-based recommendation; GPS, good practice statement

Introduction

The optimal sleep requirement for older people is approximately the same as for younger people, which is seven to eight hours per night [650]. Sleep disturbances are prevalent in older people and may be exacerbated by common comorbidities such as dementia, depression, anxiety, and chronic pain [664]. When an older person presents with sleep complaints, taking a sleep history and assessing their sleep patterns are crucial, as some may experience a natural shift in sleep schedule, leading to earlier bedtimes and early morning awakenings [665].

Despite the first-line therapy being psychological and behavioural interventions, chronic use of sedative-hypnotics for sleep disruptions is common among older people. Community-dwelling older people are six times more likely than younger people to be prescribed long-term benzodiazepines, with women at higher risk than men [666]. Insomnia is the most frequent indication for benzodiazepines and other sedative-hypnotics in this population [667].

Sleep disruptions, including insomnia, can be distressing for a person and can have important consequences [668]. Longitudinal studies found an association between sleep complaints and depression, falls, cognitive impairment, and compromised physical performance in older people [669-671].

Justification of recommendations

Refer to the narrative evidence summary, the GRADE Summary of Findings table in the guidelines, and the Technical Report for a complete presentation of the deprescribing evidence based on the GRADE framework (including other factors considered in developing the recommendations).

Sedative-hypnotics, including benzodiazepines, are indicated only for short-term use up to two to four weeks at the lowest effective dose, yet long-term use remains a growing concern [672]. The prevalence of use continues to rise among older people, with the highest increase seen in those aged 85 years and older [673]. Chronic use has been associated with an increased risk of osteoporotic fractures [674], falls, and cognitive impairment [675].

Appropriate non-pharmacological therapies and/or safer alternatives for the management of sleeping complaints should be offered to individuals at all times as pharmacological treatment should not be the sole treatment for insomnia [650].

Given the risk of potential harm and dependence with long-term use (e.g. beyond four weeks) generally outweigh the benefits, deprescribing should generally be considered in older people [650]. However, there may be special circumstances where the benefits could potentially outweigh the risks (e.g. palliative care). If psychological and behavioural interventions are proven ineffective, trial on-demand or intermittent use of sedative hypnotics at the lowest effective dose may be considered if appropriate in addition to appropriate investigation to identify and subsequently treat a cause if symptoms are chronic and persistent. The limitations of sedative hypnotics, potential benefits, and harms should be thoroughly communicated to the patient so they can make an informed decision about their treatment choice.

When deprescribing sedative-hypnotics, the tapering plan should be individualised, ensuring withdrawal symptoms are managed before further dose reductions. A common approach involves reducing the dose by approximately 25% every 1 to 4 weeks [650]. A slower taper may be necessary for final dose reductions or if withdrawal symptoms occur [650]. The Maudsley deprescribing guidelines recommend a hyperbolic tapering strategy for psychotropic medications, including benzodiazepines, to minimise relapse and withdrawal effects [43]. This strategy involves progressively smaller reductions at lower doses, offering a gradual approach that may be preferred by some individuals.

Narrative summary of evidence on deprescribing

We identified 19 studies (six RCTs, two cohort studies and 11 before-and-after studies) related to sedative hypnotic deprescribing from the systematic review and meta-analysis [652, 663, 676-692].

Overall, the current evidence for sedative hypnotic deprescribing is derived from studies with low and very low certainty outcomes. While deprescribing was shown to lead to a significant reduction in the number of hospitalisations and some (not all) sleep measures in RCTs, these outcomes are low and very low in certainty. The evidence to date is from studies with small sample sizes and generally have serious methodological limitations. In many studies, targeted sedative hypnotics also include Z drugs, sedating antihistamines, and certain antidepressants (e.g. mirtazapine). It was challenging to interpret study results considering the substantial differences in efficacy, safety, and risk profiles unique to each drug class. The majority of the studies had a very short follow-up duration (< 12 months) and it may be possible that severe and persistent withdrawal syndromes, if occurred, were not captured. The evidence at this stage remains insufficient to inform evidence-based recommendations.

If sedative hypnotics are considered appropriate to deprescribe, closely monitoring for any withdrawal symptoms and/or symptoms indicative of relapse, including worsening sleep quality, changes in psychological or physical health status (e.g. anxiety, and agitation) and quality of life may be appropriate.

Key study characteristics and results

A narrative summary of each study is provided below, highlighting key characteristics and main findings.

Curran 2003 randomised 104 long-term benzodiazepine users (daily use for at least six months) to immediate tapering (n=45), or to commence tapering after 12 weeks (n=38) [677]. A third group (n=21) who did not wish to discontinue their benzodiazepine served as a control group. All participants were followed for 52 weeks. Participants were excluded if they had dementia, a history of seizures, or current major psychiatric disorders. Other daytime benzodiazepines and psychoactive medicines were permitted during the study. At 12 weeks, mortality was not significantly different between the discontinuation group and control group (OR 0.29, 95% CI 0.01 to 7.32; participants = 104). Similarly, there was no significant difference between the two groups at 12 weeks in terms of ADWEs measured using the Benzodiazepine Withdrawal Scale Questionnaire (BWSQ) (MD 1.50, 95% CI -6.09 to 9.09), depression measured using the Geriatric Depression Scale (MD 0.30, 95% CI -0.85 to 1.45), or quality of life measured using the Short-form 36 (MD 0.00, 95% CI -12.97 to 12.97). Medicine withdrawal was frequently successful (80%), without any adverse effects on sleep (raw data not provided).

Petrovic 2002 randomised 40 inpatients who had been taking benzodiazepines for at least three months to placebo (n=20) or 1 mg lormetazepam (n=20) [683]. After one week, all participants had either the placebo or lormetazepam discontinued. At 30 days, there was no significant difference in the number of participants who had their benzodiazepine completely discontinued between the group with abrupt discontinuation and the group with temporary substitution (OR 0.25, 95% CI 0.06, 1.02). There was also no significant difference between the two groups in the return of underlying symptoms, in this case worsening of sleep (OR 0.21, 95% CI 0.02, 2.08).

Kuntz 2019 randomised 150 people who had two or three Z-drugs prescriptions dispensed in 2016 to receive educational information only (n=50), educational information and pharmacist consultation (n=49), and usual care (n=50) [681]. Participants were excluded if they received less than seven doses of medicines, or were using antipsychotics, cholinesterase inhibitors or memantine. At six months, participants who were provided with evidence-based information about Z-drugs (with or without a follow-up pharmacist consultation) were significantly more likely to discontinue Z-drugs than those who received usual care (OR 0.28, 95% CI 0.13, 0.59). Additionally, intervention group participants also had a significantly lower number of Z-drugs dispensed (MD -0.90, 95% -1.44, -0.36) and a significantly lower number of hospitalisations following discontinuation of Z-drugs (MD -0.10, 95% CI -0.16, -0.04) but there was no change in the number of emergency department visits (MD 0.00, 95% CI -0.17, 0.17).

Tham 1989 randomised 36 inpatients who had been taking temazepam 10mg at night for at least a month to abrupt discontinuation (n=15) to gradual withdrawal (n=16) [687]. Participants who were randomised to abrupt discontinuation received a placebo for ten nights whereas the other group received temazepam 5mg for the first four nights, then 2 mg for the next four nights, then placebo for the last two nights. There was no significant difference in the mean hours of sleep between the two groups during the ten-night study period (MD 0.00, 95% CI -0.83, 0.83).

Tabloski 1998 randomised 20 women who had been using long-term sedative hypnotics (diphenhydramine, lorazepam, flurazepam, nortriptyline, triazolam) at least five nights a week for at least six months to gradual tapering then complete withdrawal (n=10) or continuation (n=10) [686]. Participants were excluded if they were taking other centrally acting drugs such as antidepressants, antihistamines for allergies, beta-blockers, narcotic drugs, and neuroleptic medicines. After two weeks of complete withdrawal, there was no significant difference between the two groups in the number of wakings (defined as the number of bouts of wakefulness per hour of sleep) (MD 0.30, 95% CI -0.54, 1.14). Control group participants reported significantly longer hours of sleep (MD 1.43, 95% CI 0.88, 1.97) and sleep duration (MD 28.00, 95% CI 14.90, 41.10). However, participants who had discontinued sedative hypnotics reported a significantly shorter sleep latency (MD -13.70 minutes, 95% CI -26.95, -0.45) and wakefulness after sleep onset (MD -28.50, 95% CI -45.60, -11.40).

Fung 2024 included participants who had used lorazepam, alprazolam, clonazepam, temazepam, and/or zolpidem for insomnia at least two nights per week for at least three months [691]. Participants were randomised to either masked taper plus cognitive behavioural therapy (CBT) (n=92) or standard CBT plus unmasked tapering (n=96). Participants in both arms received the deprescribing intervention to have their benzodiazepines discontinued using two different mechanisms for tapering to investigate potential placebo effects with a masked taper strategy. Benzodiazepines were gradually tapered by approximately 25% per week over nine weeks. In some cases, two benzodiazepines were tapered simultaneously. At six months, 116 out of 176 (66%) participants with complete follow-up data had their benzodiazepines discontinued, which translates to 64/87 (74%) in the masked taper group and 52/89 (58%) in the unmasked group. Both groups of participants had an improvement in the severity of insomnia symptoms, measured using the Insomnia Severity Index (lower scores indicate lower severity of insomnia). In the masked taper group, the difference in scores from baseline to 6 months was MD -6.41, 95% CI -7.87, -4.95. In the unmasked taper group, the score difference from baseline to 6 months was MD -6.57, 95% CI -8.00, -5.14. Three participants (2%) had falls that led to discontinuation of the intervention or hospitalisation/emergency department presentation (two participants in the masked taper group and one in the unmasked group).

Gardner 2024 compared two direct-to-patient education interventions (YAWNS-1 and YAWNS-2 packages) with usual care on the use of benzodiazepines and Z-drugs for insomnia in community-dwelling individuals [692]. Individuals were included if they took benzodiazepines and Z-drugs for at least three nights a week for three or more months of insomnia, and were excluded if they resided in a residential care, took other prescription sedatives for insomnia, a score of less than 10 on the mini-Montreal Cognitive Assessment or had any other indications for using benzodiazepines and Z-drugs (e.g. seizure, spasticity related to a spinal injury). We compared intervention group participants (YAWNS-1 or YAWNS-2 combined, n= 378) with usual care (n = 187). At six months, the proportion of participants discontinuing was higher in the intervention groups (YAWNS-1 or YAWNS-2) compared with the usual care group (OR 0.27, 95% CI 0.15, 0.48). However, there was no significant difference in dose reduction at six months between the intervention groups and the control group (OR 0.70, 95% CI 0.42, 1.15). Among participants who stopped their original sedative hypnotics (n=136), there was no significant difference between the intervention groups and the control group in the occurrence of adverse drug withdrawal events (OR 1.58, 95% CI 0.54, 4.66). The most common withdrawal symptoms were insomnia (40/136, 29%), anxiety (12/136, 9%), other mental health effects (4/136, 3%), and physical withdrawal symptoms (7/136, 5%). Withdrawal symptoms lasted for more than four weeks for 13 participants (10%). Withdrawal symptoms were rated as severe in 10 participants (7%).

Allary 2024 reported the long-term effects of benzodiazepine and Z-drugs discontinuation among 45 participants enrolled in a previous RCT with available follow-up data at 12 months [663]. Participants were included in the original RCT if they had taken a benzodiazepine or Z-drug for the past two years and were wanting to stop. Participants were excluded if they were experiencing a crisis (e.g. suicidal ideation), having an alcohol or drug dependence disorder, or other indications for use (e.g. epilepsy). At 12 months after discontinuation, depressive symptoms intensity measured using the Beck Depression Inventory–II reduced with reduced use of benzodiazepine or Z-drug (unstandardised regression coefficient, 0.879, p < .01). However, there was no statistically significant association between the change of benzodiazepine or Z-drug use and worry intensity measured using the Penn State Worry Questionnaire or sleep quality measured using the Pittsburgh Sleep Quality Index.

Kosto 2023 reported a cohort study that included inpatients who had been taking benzodiazepines or Z-drugs for insomnia for at least three months [680]. The study compared study quality between patients who received usual care (n=114) and patients who had received explanations and recommendations for deprescribing with (n=55) or without a tapering schedule (n=46). Participants were excluded if they were using the sedative hypnotic as part of the treatment for a psychiatric disorder, had dementia, took more than one sedative hypnotic, or were taking a narcotic drug. After three months, there was no significant change in sleep quality measured using the Pittsburgh Sleep Quality Index (PSQI) between the group who received usual care and the groups who received the educational intervention (MD -3.30, 95% CI -5.09, -1.51).

Puustinen 2014 conducted a cohort study that included 92 primary care patients who had been taking benzodiazepines (temazepam, zopiclone, zolpidem) for at least one month for a primary diagnosis of insomnia [684]. Participants were excluded if they were taking antipsychotics or antiepileptics concurrently, had a history of, active alcohol or drug abuse, severe psychiatric disorder, severe neurological disease, or smoked more than ten cigarettes daily. While their benzodiazepine dose was gradually tapered over four weeks, participants were provided with either melatonin 2 mg (n=46) or placebo (n=46) at night. At six months, among the 89 available for follow-up, 34 (38%) were no longer taking benzodiazepines and 44 (49%) were reduced to on-demand use.

Van der Linden 2023 conducted a before-and-after study that included 173 people admitted to the geriatric ward who were taking regular hypnotics (benzodiazepines and Z-drugs) for insomnia, anxiety or an undefined reason [689]. Participants were excluded if they were using multiple benzodiazepines and Z-drugs or had severe psychiatric or neurological disease. One month after hospital discharge, there was no significant difference between the pre-intervention and post-intervention period in terms of sleep quality measured using PSQI (MD -0.17, 95% CI -1.27, 0.93), delirium (OR 1.14, 95% CI 0.44, 2.96), or the number of participants who fell at least once (OR 0.86, 95% CI 0.31, 2.38).

Chae 2024 reported a before-and-after study that included a patient educational outreach program targeting primary care patients who were prescribed at least one long-term benzodiazepine [652]. Patients with a single prescription for less than 15 tablets were excluded. Among the 25 patients who initiated a deprescribing discussion with their primary care physician, seven (28%) had their benzodiazepine discontinued and nine (36%) had a dose reduction.

Bourgeois 2014 reported a before-and-after study that included 38 nursing home residents who had been using benzodiazepines or Z-drugs daily for at least three months for insomnia [676]. Participants who used benzodiazepine during the day for anxiety, or sedative antidepressants (trazodone, amitriptyline, mirtazapine) were excluded. After two months, 25 residents completely discontinued their benzodiazepines or Z-drugs, seven had a dose reduction, and six restarted the medicine. At month eight, one of the 25 residents restarted the medicine, one resident died, and one additional resident discontinued their benzodiazepines or Z-drugs. Compared to baseline, there was no significant difference in ADWEs measured using BWSQ (3.9 ± 2.8 to 4.1 ± 2.6, p = 0.865) or quality of life measured using EQ-5D (0.439 to 0.456, p = 0.879) after discontinuation.

Fixen 2022 conducted a before-and-after study that included 93 primary care patients who had filled a prescription for a sedative hypnotic (benzodiazepine or nonbenzodiazepine sedative-hypnotic such as zolpidem, zaleplon, eszopiclone) in the past 12 months [678]. Patients were excluded if they had dementia with behavioural symptoms or anxiety disorders but were not prescribed any other medicines for anxiety. All patients received an educational information packet about deprescribing, and their primary care providers received educational intervention from clinical pharmacists. Among the 37 participants who discontinued the medicine, 28 (76%) were prescribed the medicine for symptoms of insomnia. The other indications were anxiety, insomnia and anxiety, muscle spasms, essential tremors, and fear of flying. Anxiety was reported in 7/37 participants (19%) who discontinued the medicine.

Gemelli 2016 reported a before-and-after study that included 36 nursing home residents who had been taking sedative hypnotics (benzodiazepines, Z-drugs, sedating antihistamine, mirtazapine, melatonin, trazodone) for extended durations with a confirmed insomnia diagnosis [679]. Participants were excluded if they had a seizure diagnosis. In the study, pharmacist recommendations to discontinue or taper the sedative hypnotic were accepted by 19/36 (53%) residents.

Lui 2021 conducted a before-and-after study that included 111 primary care patients who had been referred by their physicians to the pharmacist to deprescribe benzodiazepines and/or Z-drugs [682]. In the study, 36 (32%) discontinued their sedative hypnotics and 36 (32%) had a dose reduction of 50% or more. Among the 36 patients who discontinued completely, 26 (72%) remained off the medicines at six months.

Wilson 2018 conducted a before-and-after study that included 50 inpatients who had been taking sedative use (benzodiazepines and Z-drugs) at least weekly in the past month and at least three doses in the week before the hospital admission [690]. All 50 patients received an educational brochure to encourage the conversation about deprescribing with the medical team. At 30 days after hospital discharge, 32/50 (64%) who received the intervention had their sedatives deprescribed.

Ragan 2021 conducted a before-and-after study to reduce sedative hypnotic prescribing in older veterans [685]. The study reported that academic detailing led to a significant reduction in the prescribing of benzodiazepines (-23%, p< 0.001) and benzodiazepine receptor agonists (-15%, p< 0.001) but an increase in the use of alternative medicines for insomnia (+23%, p< 0.001) which included melatonin, trazodone, and mirtazapine.

Tsunoda 2010 conducted a before-and-after study that included 30 nursing home residents who received at least one benzodiazepine hypnotic (brotizolam, flunitrazepam, etizolam, quazepam, estazolam, nitrazepam, flurazepam, diazepam) without a history of substance abuse within the past six months [688]. The included residents had a psychiatric diagnosis of schizophrenia (n=12), primary insomnia (n=9), dementia (n=7), and bipolar disorder (n=2). Four participants dropped out of the study due to insomnia. Among the 26 participants who discontinued their benzodiazepine, there was a significant improvement in the body stability (ability to maintain and control or resist changes in equilibrium) and a recovery in cognitive function during daytime. Body stability was assessed by measuring the range and the total length of the trunk motion by varying the resistance applied, both with eyes open and closed with feet together, with a shorter length indicating better stability.

Narrative evidence summary: withdrawal schedules

Various methods were used for deprescribing in the included studies and tapering was the most common method. While there was no direct evidence that any particular method was associated with the greatest benefits and harms, dose tapering is likely more acceptable than abrupt cessation and helpful in determining the lowest effective dose for some people requiring dose reduction rather than complete cessation.

In an RCT, Z-drug discontinuation was personalised based on a telephone consultation with a pharmacist (n=149, low certainty) [681]. In other RCTs, the dose titration regimen was individualised based on the original dose and benzodiazepine to minimise the risk of withdrawal (study=1, n=138, low certainty) [677], the dose for sedative hypnotics was halved every week until completely ceased (study=1, n=20, very low certainty) [686]. A study targeted benzodiazepines compared abrupt discontinuation to titration for a week using 1mg lormetazepam, which was less than half the average daily benzodiazepine dose (n=40, very low certainty) [683]. No significant difference in outcome was reported in the study. Similarly, in a study targeted temazepam, abrupt discontinuation was compared with gradual withdrawal over ten days and no significant difference in outcome was reported (study=1, n=36, very low certainty) [687]. The method was not described in one RCT (n=565, very low certainty) [692].

In the two cohort studies, the study that targeted both benzodiazepines and Z-drugs stated a drug-specific tapering schedule was followed (n=215, very low certainty) [680] whereas another study that targeted only benzodiazepines stated benzodiazepines were gradually withdrawn over four weeks and supplemented either melatonin 2 mg or placebo (n=89) [684]. The latter study did not report any critical or important outcomes. In another before-and-after study, tapering was either based on a standardised tapering regimen, abrupt discontinuation, or “any attempt” (study=1, n=173, very low certainty) [689].

In the single-arm before-and-after studies, withdrawal schedules were summarised below, with the method not described in three studies (n=357, n unstated in one study) [678, 685, 690]:

25% reduction either every one or two weeks (study=1, n=38, very low certainty) [676]
Weekly reduction of 25% of the regular daily dose from baseline each week for three weeks (study=1, n=30, very low certainty) [688]
Gradual dose reduction (studies=2, n=249, very low certainty) [663, 691]

The following single-arm before-and-after studies did not report any critical or important outcomes:

Individualised (study=1, n=111) [682]
Gradual dose reductions or abrupt discontinuation (study=1, n=36) [679]
Tapering plan based on previously published clinical guidelines (study=1, n=25) [652]