Interventions to Reduce Benzodiazepine and Sedative-Hypnotic Drug Use in Acute Care Hospitals: A Scoping Review

Background Benzodiazepines and sedative-hypnotics (BZD/SHD) are commonly utilized in the acute care setting for insomnia and anxiety and are associated with cognitive impairment, falls, fractures, and increased mortality. Interventions to reduce use of BZD/SHD in hospitals are not well characterized. The objective of the scoping review was to identify and characterize interventions to reduce the use of BZD/SHD by adults for anxiety and sedation in hospitals. Methods We included studies and abstracts published in English that described an intervention to reduce BZD/SHD in adult hospital patients. Six databases (PubMed, EMBASE, CINAHL, PsycINFO, Scopus, and Web of Science) and the grey literature (Opengrey, Grey Matters, Google Advanced) were searched up to July 2018. Titles and abstracts were screened and full-text articles were reviewed for potential inclusion by three independent reviewers. Data on each eligible study was charted in a Microsoft Excel® database. Stakeholder consultation occurred before and after the scoping review was completed. Results There were 9480 records identied from all sources and 35 studies were included in the scoping review. Included studies were divided into two categories that emerged from stakeholder feedback: sedatives prescribed in hospital or home medications. The most common study designs were pre-/post-test (24, 68.6%) and randomized controlled trials (6, 17.1%). The majority of studies tested a single intervention (28, 80%) and these were most commonly education, relaxation training and sleep protocols. Patients were frequently the target of relaxation training and behavior change interventions, while sleep protocols, multifaceted interventions and education were usually directed at healthcare providers, either alone or in combination with patients. Most studies reported positive outcomes in decreasing BZD/SHD use (23, 65.7%), including some that were statistically signicant (13, 37.1%). Conclusions This scoping review found a variety of interventions aimed at decreasing the utilization of BZD/SHD in the acute care setting, where previously little was known. Current literature addressed the initiation of BZD/SHD in hospital, rather than chronic medications that had been prescribed in the community. Stakeholder consultation supported these ndings and pointed out important factors to consider when designing an intervention for

There were no exclusions placed on publication date, type of publication (e.g. abstracts, journal articles, reports), although it had to be a completed study and not a protocol; or study design. Only English language studies would be considered for inclusion. The PICO framework (22) was used to determine inclusion and exclusion criteria: Population: Patients ≥ 18 years old in acute care hospitals who were prescribed benzodiazepines, zopiclone, zaleplon, zolpidem, eszopiclone, trazodone, chloral hydrate, melatonin, ramelteon, suvorexant, doxepin (up to 6 mg daily), or l-tryptophan; for sedation/insomnia or anxiety/stress. Excluded were studies where other drug classes were prescribed for sedation or anxiety (e.g. antipsychotics, antihistamines, antidepressants (excluding trazodone) and barbiturates); or where BZD/SHD were used for other indications such as epilepsy, drug/alcohol withdrawal, pre-or peri-procedural use, or palliative care.
Intervention: Frameworks, guidelines, policies, models that have been evaluated to reduce, stop or prevent the use of benzodiazepines or sedative-hypnotic drugs were included. Studies with no intervention (e.g. utilization patterns only) were excluded.
Comparison: there were no limits placed on the presence or type of comparison.
Outcomes: Drug utilization data, proportion of patients using benzodiazepines or sedative-hypnotic drugs, or proportion of appropriate use were included outcomes. Studies with a description of an intervention without an evaluation or outcomes assessment were excluded.

Information Sources and Search Strategy
The database literature searches were conducted from inception until July 2018. A preliminary search was conducted in PubMed, by an experienced medical librarian, and the results were reviewed by the research team to identify relevant articles and to re ne the search strategy. Final searches were then conducted in the following six databases: PubMed, EMBASE, CINAHL, PsycINFO, Scopus, and Web of Science (see Additional File 1). The nal search strategy was prepared by the medical librarian and peer-reviewed by an information specialist using the Peer Review of Electronic Search Strategies checklist. (24) Search results from bibliographic databases were imported into Covidence web-based software platform (25) to screen titles and abstracts, and for full text review.
The Canadian Agency for Drugs and Technologies in Health's (CADTH) "Grey Matters: a practical tool for evidence-based medicine", OpenGrey, and Google Advanced were used to search the grey literature. With the assistance of a CADTH information specialist, the Grey Matters checklist was reviewed and irrelevant sources were omitted. Seventy relevant sources, international databases and agency websites, were searched independently by two research assistants using various combinations of search terms (see Additional File 2).
Individual relevant studies within systematic reviews were also retrieved for full text screening and the systematic reviews were excluded. Reference lists of all included articles were searched by two research assistants to identify any studies that would meet inclusion criteria. Authors were contacted directly by email if the full text of the article was not published or could not be obtained through the university or health authority's library. If no full text or additional information was received on a study within one month, the study was included or excluded based on the information available.

Selection of Sources of Evidence
Two research assistants were trained by the principal investigator on title and abstract screening processes and application of inclusion and exclusion criteria.
Title and abstract screening were conducted independently by two research assistants (C.G. and P.A.). Discrepancies were resolved by the principal investigator (H.N.) using consensus. Full texts of included articles were then retrieved for further review. While there were no language limits in the literature search, at the full text review, only English language studies were included. The principal investigator and the two research assistants independently screened all full text articles. Discrepancies during full text review were resolved through consensus by another research team member (J.B.).

Data Charting Process
A standardized data charting template was drafted in Microsoft Excel® by two research assistants and was further revised and developed throughout meetings with the research team. A training session took place where the principal investigator performed the extraction of two eligible articles with the research assistants. Two research assistants then charted the data of the remaining articles (see Additional File 3). Data charting was veri ed and revised independently by two members of the research team (H.N. and J.B.). Discrepancies were resolved by consensus.

Synthesis
The included articles were described by intervention type (e.g. education, sleep protocol) and target (patient and/or healthcare provider). Each intervention was attributed to one of two categories that emerged from the rst stakeholder consultations: whether interventions addressed sedatives prescribed to patients in hospital, or patients' home therapy that was continued in hospital.

Consultation
Two series of stakeholder meetings were conducted; feedback was obtained at the beginning to guide the scoping review and at the end to discuss the scoping review results. Stakeholders were healthcare providers and patients from the Queen Elizabeth II Health Sciences Centre of NSHA who volunteered to participate and signed an informed consent form. In the rst series, stakeholders were invited to answer open-ended questions surrounding their experience with BZD/SHD use in the hospital, how to change BZD/SHD use, other options for insomnia or anxiety, and resources available to decrease BZD/SHD use in the hospital. In the second series, participants were provided with the results of the rst stakeholder meetings and were asked to comment on the feasibility of interventions identi ed by the scoping review. Meetings were recorded and transcribed, or detailed notes were taken.

Results
A total of 9480 records were identi ed from all sources and 35 articles were included in the scoping review ( Figure 1).

Study Characteristics
Study characteristics are summarized in Table 1, and individual study details are in Table 2. Most studies were available as full text (30,85.7%) with the remainder in abstract form. The majority were a pre-/post-test design (24, 68.6%) followed by randomized controlled trials (6, 17.1%). The most frequent comparators were no intervention or usual care (29, 82.9%), or a single intervention (28,80.0%). Single interventions could have multiple targets and/or formats, e.g. education could be provided verbally and in writing, to various healthcare providers and/or patients, but was categorized as a single intervention.

Computerized alerts
Automated alerts in computerized provider order entry systems were introduced to improve BZD/SHD use in two pre-/post-test studies. (48,49) In both studies, a statistically signi cant improvement in the use of recommended sedatives over non-recommended sedatives was found, as well as increased adoption of dosing recommendations. (48,49)

Miscellaneous Interventions
There were a total of three studies that evaluated miscellaneous interventions that included pharmacy practice,(50) therapeutic body wrap,(51) and a simulated nature environment.(52) The pharmacy practice initiative was a pre-/post-test design and did not result in decreased sedative use.(50) Therapeutic body wrap was a psycho-body technique employed to calm patients experiencing anxiety during an acute psychosis and reported mixed results.(51) Finally, a randomized controlled trial studying a simulated nature environment (lighted ceiling panels that resembled the sky) did not demonstrate any changes in sedative use.(52)

Description of interventions for home medications
There were seven studies that speci cally addressed patient's chronic BZD/SHD therapy on admission to hospital, and these utilized behavior change strategies, withdrawal protocols, and deprescribing.(53-59) Behavior change interventions were described in three studies, two were pre-/post-test designs and one was a randomized controlled trial. (53)(54)(55) Two studies yielded a positive outcome,(53,54) but only one was statistically signi cant.(54) One report had mixed results.(55) A withdrawal protocol for patients taking chronic therapy was investigated in a pre-/post-test study and a randomized controlled trial by the same investigator. (56,57) One study had a positive outcome (57) and in the other study there was no difference.(56) A deprescribing intervention with followup after discharge was described in two pre-/post-test study designs. (58,59) One study reported a positive outcome without conducting a statistical analysis, (58) and one study had a positive result that was statistically signi cant. (59) In the rst series, 21 participants consented and attended three group meetings and two individual meetings to provide feedback in May 2018.(60) Two themes emerged naturally from the stakeholder discussion; that of the issue of sedatives newly prescribed in hospital and sedatives that were home medications and continued in hospital. Transcribed comments were categorized into solutions, barriers, risks, risk modi ers, and bene ts.
For the sedatives prescribed in hospital, top barriers to reducing use were (1) easy to prescribe, (2) healthcare providers are busy, (3) staff are not aware of risks, and (4) the hospital environment. Solutions provided were related to policy changes, medication reviews by pharmacists, education, and alterations to the environment. For sedatives as home medications continued in hospital, top barriers to reducing use of BZD/SHD were (1) di cult to make changes to home medications, (2) patient unwilling to make changes, (3) withdrawal takes a long time, and (4) sedatives are not causing any problems. Suggested solutions were deprescribing and withdrawal regimens.
The second series of stakeholder meetings were held in May 2019 to review the preliminary results of the scoping review and obtain feedback on interventions identi ed. Thirty-four participants attended four group meetings and ve individual meetings. A graphic was developed to facilitate discussion and included top barriers and solutions from the rst series of stakeholder meetings mapped to the most common interventions found in the scoping review. Stakeholders discussed the advantages and disadvantages of each type of intervention (Tables 3 and 4). Overall, stakeholders were passionate about improving patient care and reducing risk associated with the use of sedatives for sleeping disturbances or anxiety. They felt that feasible interventions should take into account individual patient preferences and risks, staff workload, and the in exible aspects of the hospital environment. Strategies that increased awareness through education of staff, physicians, patients and families and a focus on best practices were a common theme throughout the discussions.

Discussion
This scoping review identi ed 35 studies that reported a wide variety of interventions to reduce the use of BZD/SHD in hospitalized patients. Single interventions were most commonly used (80%) compared to multifaceted interventions. Education was the most frequently employed strategy, followed by sleep protocols and relaxation training, either alone or in combination with other strategies. Most studies (66%) reported a positive outcome for the primary endpoint with just over half of these reporting a statistically signi cant change in BZD/SHD utilization. The intent of a scoping review is not to assess the quality of the evidence found, however, the majority of studies (69%) were pre-/post-test single group designs which are generally considered of higher risk of bias.
There was little overlap in the type of interventions used when considering the target populations of patients and healthcare providers. Patients were frequently the target of relaxation training and behavior change interventions, and these resulted in mainly positive outcomes. (43)(44)(45)(46)(47)(53)(54)(55) Sleep protocols, multifaceted interventions and education were usually directed at healthcare providers, either alone or in combination with patients. (3,(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42) Of these, sleep protocols and multifaceted interventions produced the most positive outcomes. (26)(27)(28)(29)(30)(31)(32)(38)(39)(40)(41)(42) Education on its own, in the form of written materials, lectures and/or audit and feedback, was not as successful. (3,(33)(34)(35)(36)(37) These results are further supported by systematic reviews of educational interventions to change professional behavior which found small or no effects on improving performance or patient outcomes.(61-64) Computerized alerts (e.g. reminders) directed at healthcare providers did lead to positive outcomes in two studies (48,49) included in the scoping review and as an intervention may be more effective. (65) This scoping review demonstrated there is a relative lack of primary literature available to guide clinicians on how to reduce BZD/SHD use in hospitals, compared to the community setting. A recent scoping review focused only on deprescribing strategies and found 74 original research articles that aimed to reduce BZD/SHD use in the community.(17) Pharmacologic interventions were the primary method employed (42 studies, 57%), whereas in the hospital setting only four studies (11%) used pharmacologic substitution and deprescribing.(17) This is likely due to the time required to monitor the addition of pharmacotherapy and/or support the gradual dose reduction of BZD/SHD,(66) which was con rmed during stakeholder meetings. As well, stakeholders provided very clear feedback that changing BZD/SHD prescriptions that had been initiated in the community setting was very challenging. A higher rate of positive outcomes (66%) was found in this scoping review for the hospital setting, compared to deprescribing interventions in the community (47%). (17) There may be numerous reasons for this, including broader inclusion criteria for the types of interventions and a lower proportion of randomized controlled trials in our scoping review.
A strength of this scoping review was the extensive stakeholder consultation which highlighted important factors to consider when implementing interventions to decrease the use of BZD/SHD in hospitals and served to integrate knowledge translation throughout the process. Engaging nurses and pharmacists in interventions were common themes during the stakeholder consultations, which has been echoed in recent published guides. (19,32) Interventions should empower nurses to create a safe environment for patients, rather than generate extra work, and it was found that nurses were active participants in most of the studies. On the other hand, while pharmacists are ideally positioned to identify inappropriate use of BZD/SHD and provide education on alternatives, the scoping review found fewer examples of pharmacists lling this role. Another critical issue to address were the barriers to appropriate prescribing that physicians experienced such as being called late at night when patients cannot sleep or having easy access to sedatives on pre-printed orders. Stakeholders emphasized preventative measures such as changing the hospital environment, providing non-pharmacologic strategies, and developing policies around the appropriate use of BZD/SHD. (19,32) Interestingly, patient-focused interventions such as relaxation training and behavior change methods were successful in the scoping review but were not well-described in the guides. An additional strength was the conduct of the review, which adhered to the published methods (20,21) with searches of multiple medical databases and the grey literature with no limits placed on publication date, study design, or publication type.
This review was limited to studies that evaluated changes in utilization of BZD/SHD and did not include an assessment of other outcomes such as sleep quality, patient satisfaction, or cost-effectiveness. However, the objective of the scoping review was to identify and characterize interventions and not to describe the outcomes. Although some potential interventions may have been missed due to the outcome eligibility criteria, interventions not previously described in the guidelines such as relaxation therapy were found, indicating an expansion of previously limited knowledge around interventions in the hospital setting was achieved. Stakeholder feedback was obtained from healthcare providers who volunteered for the meetings, therefore selection bias may be present and their comments may represent local institutional culture and resources and may not be generalizable to the broader healthcare system. Nevertheless, stakeholders identi ed many issues that were previously reported in the literature. Stakeholder consultation was also utilized to develop the initial research question and provide context for the results in order to make these ndings directly applicable and relevant to the hospital setting.
The scoping review found a variety of interventions aimed at decreasing the utilization of BZD/SHD in the acute care setting, where previously little was known. While most strategies identi ed were single interventions and/or included an educational component, the most successful strategy may need to consider the targeted population. For patients, behavior change and relaxation therapy interventions warrant further research. Approaches that are solely aimed at healthcare providers, such as computerized alerts, also require additional study. Finally, both patients and healthcare providers may respond to multifaceted interventions and sleep protocols. The majority of research addressed the initiation of BZD/SHD in hospital, therefore more evidence is required to determine the best strategies to reduce chronic medications that had been prescribed in the community. Stakeholder consultation supported these ndings and pointed out important factors to consider, such as the hospital environment and staff workload, when designing an intervention for hospital patients.

Consent for publication
Not applicable

Availability of data and materials
Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.

Competing interests
The authors declare that they have no competing interests.

Funding
This work was funded by the Dalhousie University Pharmacy Endowment Fund. The funding body had no role in the study design; collection, analysis, and interpretation of data; or in writing the manuscript.
Authors' contributions HLN designed the study, obtained the funding, conducted the stakeholder meetings, screened the studies, charted the data, analyzed the data and drafted the manuscript. CG and PA searched the literature, screened the studies, charted the data and revised the manuscript. JB, JEI and SKB revised the study design, analyzed the data, and revised the manuscript. All authors read and approved the nal manuscript.  Tables  Table 1 Study characteristics, n = 35

Characteristic
No. of studies (%) Article type     (3) -Two week inpatient program of cognitive behavioral therapy -Motivational interviewing and education with follow-up after discharge -Educational brochure to promote tapering after discharge Withdrawal protocol (3) -Substitution with another benzodiazepine then complete withdrawal -Tapering protocol in combination with multifaceted interventions Deprescribing (2) -Pharmacist intervention in collaboration with physician with follow-up with family physician and community pharmacist after discharge