Hospital initiation of benzodiazepines and Z-drugs in older adults and discontinuation in primary care

Introduction

Benzodiazepines and z-drugs (i.e. benzodiazepine receptor agonists or BZRAs) are frequently prescribed medicines, most often being used in the management of insomnia and anxiety. As with any medicine, the benefits for these clinical indications must be balanced against the risk of harm for the individual. Risks associated with BZRAs are of particular concern when prescribed to older people and/or on a long-term basis as often these risks outweigh the benefits.¹ Despite this, the prevalence of use of these agents is high internationally, in particular long-term use in older adults.² In the United States, trends indicate that outpatient use of these medicines has increased significantly in recent decades.³ Although Ireland has among the highest levels of long-term use among older adults among OECD countries,² prescribing rates of long-acting benzodiazepines have been decreasing slowly since 2005.⁴

Older people may be especially prone to adverse effects from long-term BZRA use as pharmacokinetic and pharmacodynamic changes in ageing can lead to increased retention time resulting in prolonged, potentiated effects.⁵ These factors can result in hangover effects, and contribute to increased risk of falls, fractures and road traffic accidents among others.⁶ This issue is highlighted in the American Geriatrics Society’s Beer’s criteria which states that they should generally be avoided due to increased risks of falls and delirium, but that they are suitable in some specific conditions.⁷ The European STOPP (Screening Tool for Older Persons’ Prescriptions) criteria deem use in older people for a period of greater than a month to be potentially inappropriate prescribing.⁶ These aformentioned sequelae of long term benzodiazepine and z-drugs prescribing can also increase healthcare costs consequently burdening others in the health system.⁸

In addition, long-term use can result in the development of tolerance and dependence, which presents an example of a barrier to reduce (i.e. deprescribe) these agents. A systematic review of barriers to deprescribing found that fear was the most common patient barrier for BZRAs, including withdrawal symptoms and condition deterioration.⁹ Prescriber barriers are also a significant factor to consider, for example inertia, fear of the unknown and incomplete information on the rationale for prescribing have been shown to prevent change.¹⁰ Older adults may receive care from several prescribers, where they have multiple conditions, and may have frequent hospital admissions. This can further impede deprescribing when a general practitioner (GP) is trying to optimise treatment in the context of transitions of care and medications initiated by multiple other prescribers. Therefore, improving communication of information from initiating prescribers across the secondary care-primary care interface may address some of these barriers to deprescribing. Studies have shown that providing relevant information in hospital discharge summaries encouraged faster discontinuation of proton pump inhibitors where long-term use is not indicated.^11,12 Although previous research has found hospital admission are associated with increased initiation of benzodiazepines,^13–16 the impact of hospital discharge summary information on deprescribing has not been considered to date.

Therefore this study aims to examine factors associated with continuation of hospital-initiated BZRAs among among adults aged 65 years and over, and specifically whether instructions on hospital discharge summaries are associated with their discontinuation.

Methods

Results

Among 38,229 patient, there were 7,876 BZRA initiations during the study, and 7,458 of these were initiated in primary care and 418 were hospital-initiated. This represents 6.7 initiations per 1,000 prescriptions to BZRA naïve patients in primary care, and 25.8 initiations per 1,000 hospitalisations involving BZRA naïve patients. In terms of time exposed, there were 0.29 initiations per 1,000 days naïve in primary care, and 1.90 per 1,000 days naïve during hospital stays. Hospital initiations were more often to older patients, men, and to those with GMS or no health cover (Table 1). For hospital-initiated BZRAs, 59.1% of these had some level of instructions on the discharge message. Most often this was specifying use as required, a duration of prescription, or instructions to taper or review.

In 39.7% of cases of hospital inititation (n=166/418) there was a BZRA prescription in primary care within 90 days of discharge. Factors associated with continuation following discharge included lower age, being initiated on a Z-drug (compared to a benzodiazepine) or being on a more regular medicines (Table 2). Of these 166 cases, for 98 (59.6%) the BZRA was deprescribed during follow-up (mean time to discontinuation was 184 days (SD 230), Figure 1). Presence of instructions on the hospital discharge summary was independently associated with likelihood of discontinuation (adjusted hazard ratio 1.67, 95% CI 1.09 to 2.55, Table 3).

• Download figure
• Open in new tab

Figure 1.

Kaplan Meier curve of time to benzodiazepine receptor agonist discontinuation among hospital initiations with and without instructions

Discussion

Our study shows that inclusion of instructions on hospital discharge summaries is associated with shorter time to discontinuation of BZRAs amongst older people in primary care. Previous studies on BZRAs being prescribed in hospital have identified specific cases of potentially inappropriate prescribing¹³, found that the period 3 months before and after hospitalisation is associated with increased rates of initiation of benzodiazepines¹⁴, and assessed the influence of specific in-patient departments on use.¹⁵ Another study found increased rates of BZRA initiation and continuation in older adults specifically.¹⁶ The positive association of number of medicines or polypharmacy with continuation is consistent with previous research on benzodiazepines initated after hospitalisation.^14,15,16 Those started on a z-drug alone rather than a benzodiazepine were more likely to be continued following discharge, possibly reflecting a different perception of these medicines in the eyes of prescribers. Although health cover did not show a significant association in our study, previous evidence suggests an inverse relationship between socioeconomic status and BZRA prescribing rates.¹⁸

A strength of this study is the use of actual hospital prescribing at discharge, compared to previous studies using GP prescribing after discharge only. Our study is limited by a lack of information on use of non-prescription or non-pharmacological treatments, such as cognitive behavioural therapy (CBT), which may influence rates of prescribing. We also were unable to account for prescribing indication for BZRAs, including scenarios such as palliative care where long-term prescription may be appropriate. Although information on co-morbidities was lacking, we adjusted for prescribed medicines as a proxy for illness burden. For our primary analysis, we would not expect co-morbidities to differ systematically between discharge messages with or without instructions.

However our study examines a novel question and suggests that when additional instructions relating to hospital-initiated BZRAs are conveyed to GPs, earlier discontinuation of these medications can be facilitated. This sharing of information has been identified as an essential component of clinical handovers¹⁹ and hospital discharge summaries are a conduit for this information at secondary-primary care interfaces. The literature has identified that amongst hospital-based junior doctors and community-based GPs particularly, there can be a disconnect in intraprofessional perceptions of what to include on a hospital discharge summary²⁰, and that more training and inclusion in medical curriculums may improve continuity of care.^20,21

Our findings are consistent with systematic review evidence that identifies lack of information on discharge summaries as a patient safety risk.²² Without instructions from the original prescriber, GPs may be reluctant to reduce or stop a BZRA and thus short-term prescriptions can develop into long-term use. Systems for communication between prescribers at transitions of care, such as shared electronic medical records or electronic discharge summaries, should incorporate structured fields to record important information to support medication review and deprescribing, including indication, intended duration of use and review timeframe. IT-based summaries in particular have been shown to be preferable to both patients and prescribers, because of their structured and accessible nature.²³

Long-term BZRA use may also be avoided by considering non-pharmacological treatments where available, e.g. CBT, good sleep hygiene and psychotherapy, especially in those not responding to pharmacotherapy.^24,25 A meta-analysis of BZRAs in older adults for insomnia suggests that the magnitude of adverse effects may outweigh the clinical benefit received.²⁶ There is also evidence that BZRAs could increase all-cause mortality risk,²⁷ and dementia, especially when used for longer than three years.²⁸ A number of interventions, involving patient education, tapering, pharmacological substitution, and psychological support, have been shown to be effective and feasible for deprescribing long-term BZRAs in older adults.²⁹

The WHO’s Third Global Patient Safety Challenge, Medication without Harm, identifies transitions of care, polypharmacy, and vulnerable populations as priorities.³⁰ Ensuring complete information is communicated when iniatiating medications in hospital is a means of targeting all of these priorities, and would change healthcare systems to support deprescribing, medication optimisation, and patient safety.