INTRODUCTION
The prevalence of insomnia is 21% in the general adult population [1], increasing with age [2]. Benzodiazepines (BZDs) are the most common pharmacological treatment for insomnia. Moreover, they have been developed as safer alternatives to barbiturates and are widely used [3,4]. However, recently, the long-term use of BZDs has become associated with the risk of cognitive decline [5] and muscle relaxant effects [4]. In particular, the muscle relaxant effect through γ-aminobutyric acid (GABA) receptor stimulation increases the risk of falls, highlighting the need for caution in the use thereof to ensure medical safety [6].
Alternatives to BZDs include drugs that do not act on GABA receptors, such as the melatonin receptor stimulator ramelteon and the orexin receptor antagonists (ORAs) suvorexant and lemborexant. ORAs are particularly notable for their efficacy in preventing delirium [7,8] and the low risk of falling associated with their administration [9]. Consequently, they are increasingly used as alternatives to BZDs [10,11].
Falls are a clinical problem, and quality care and patient safety are priorities for healthcare organizations [12]. Falls potentially result in several negative outcomes, which include a reduced quality of life due to fractures [13], prolonged hospital stays [14], and increased healthcare costs [15]. Falls can be caused by internal factors [16], such as age, sex, body mass index (BMI), a previous history of falls or stroke [17], depression [18], and diabetes mellitus [19]. External factors include structures of buildings, mobility aids [20], and unfamiliar environments [21]. Furthermore, medications can considerably affect the risk of a fall. Sedative hypnotics, antidepressants, antipsychotics, antihypertensive drugs, diuretics, β-blockers, antiparkinsonian drugs, and non-steroidal anti-inflammatory drugs (NSAIDs) are known risk factors for falls [4,22,23]. Sedative hypnotics are exceptionally modifiable, prompting fall prevention measures, such as switching from BZDs to ORAs [24]. This shift has contributed to the widespread use of ORAs.
Despite the widespread use of ORAs, falls continue to occur in patients who have been administered these medications [9]. Preventing falls in patients who are administered ORAs is essential, as 10% of patients who experience a fall have been administered ORAs [25]. Elucidating the risk factors for falls in these patients can contribute to safer pharmacotherapy and fall prevention. Therefore, in this study, we aimed to determine the risk factors correlated with falls in patients who have been administered ORAs.
METHODS
Ethical Considerations
This study was approved by the Institutional Review Board of the Ogaki Municipal Hospital on May 31, 2024 (approval number: 20230530-18). Moreover, this study was conducted in accordance with the principles of the Code of Ethics of the World Medical Association (Declaration of Helsinki). Patient consent was obtained online by using the opt-out method.
Study Design
The data of hospitalized patients in Ogaki Municipal Hospital who had been administered ORAs between January 1, 2022 and December 31, 2023 were retrospectively analyzed. The ORAs that were evaluated included suvorexant and lemborexant. The exclusion criteria were as follows: 1) patients who took medicines they brought from home while in hospital and 2) those with unknown height or weight. Exclusion criterion 1 was set in this way to avoid missing data, as it is difficult to investigate the medication status of drugs brought from home in our electronic health record. Falls were defined as those reported by healthcare professionals through incident and accident reports during hospitalization. Patients were categorized into two cohorts based on whether they had experienced a fall: the fall and non-fall cohorts. Propensity-score matching was performed by age, sex, BMI, fall risk score, and patient background (cerebrovascular disease, history of delirium, cognitive decline). Underlying diseases were diagnosed by physicians based on standard diagnostic criteria. A diagnosis of depression was made based on the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition criteria, requiring the persistence of five or more of the nine symptoms for a minimum of 2 weeks. After propensity score matching, statistical analysis was performed to compare patient backgrounds between the two cohorts. Logistic regression analysis was then used to identify factors associated with falls among the matched cohort, including variables with p<0.2 in the univariate analysis to capture as many relevant factors as possible given the small sample size. Age, BMI, and fall risk score were already included as covariates in the propensity score model; therefore, no separate sensitivity analysis was conducted.
Endpoints
The endpoint of the study was the odds ratio (OR) of the patient demographic and clinical factors for falls.
Data Collection
The variables analyzed in this study included age; sex; BMI; comorbidities, such as delirium, cerebrovascular disease, hypertension, diabetes mellitus, heart failure, depression, schizophrenia, Parkinson’s disease, and dementia; the fall risk score; and concomitant medications, such as antidementia drugs, antidepressants, antipsychotics, BZDs, non-BZDs, antiparkinsonian drugs, NSAIDs, antihypertensive drugs, antidiabetic therapy, and the type of ORAs. Antihypertensive medications included calcium-channel blockers, angiotensin-receptor blockers, angiotensin-converting enzyme inhibitors, diuretics, and β-blockers. Non-BZDs included zolpidem, zopiclone, and eszopiclone.
Cognitive and gait function was assessed by nurses. The fall risk score assesses cognitive and gait function. A decline in either or both cognitive [26] and gait function [27] are known factors that increase the fall risk. These two factors are independent risk factors, and combining them is thought to enable clearer identification of high-risk groups. Based on analysis of past fall cases and multidisciplinary discussions, our hospital has adopted this classification method. Thus, the risk score was assessed on a scale of A–D. Patients without a decline in cognitive or gait function had a risk score of “A.” Those without a cognitive decline but with a decline in their gait function had a risk score of “B.” Those with a cognitive decline but without a decline in their gait function had a risk score of “C.” Those with a decline in both their cognitive and gait function had a risk score of “D.” Electronic medical records were retrospectively reviewed to obtain this data. This assessment was conducted once a week in hospitalized patients, and the lowest A–D value was taken as the patient’s fall risk score. Information on falls was collected from incident reports. Other data were examined retrospectively using electronic medical records.
Statistical Analysis
The Mann–Whitney U and χ-squared tests were utilized to compare the fall and non-fall cohorts. The caliper value for propensity-score matching was set at 0.2. Logistic regression analysis was conducted on the demographic and clinical factors of the patients to calculate the ORs and confidence intervals (CIs) for falls. All statistical analyses were performed with EZR (Saitama Medical Center), which is a graphical user interface for R (The R Foundation for Statistical Computing) [28]. Statistical significance was set at p<0.05.
RESULTS
Of the 2,348 patients, 1,324 were eligible. Sixty-eight patients in the falls group and 1,256 in the non-falls group were propensity score-matched by age, sex, BMI, and fall risk score. After matching, 66 patients were in the falls group and 132 in the non-falls group (Fig. 1). The fall rate in this study was 5.1% (68/1,324).
Table 1 presents the demographic and clinical characteristics of the patients in the fall and non-fall cohorts. Propensity-score matching demonstrated similar trends in age, sex, BMI, and patient background characteristics. The use of antidepressants was significantly more common in the falls group than in the non-falls group (7 [10.6%] in the falls group vs. 3 [2.3%] in the non-falls group; p=0.017). Other variables with p-values <0.2 included lemborexant use (46 [69.7%] vs. 74 [56.1%]; p=0.067), concomitant BZD use (11 [16.7%] vs. 11 [8.3%]; p=0.095), heart failure (31 [47.0%] vs. 79 [59.8%]; p=0.096), depression (1 [1.5%] vs. 12 [9.1%]; p=0.064), and schizophrenia (16 [24.2%] vs. 20 [15.2%]; p=0.123).
To identify risk factors for falls, logistic regression analysis was performed on the parameters listed in Table 1 for items with p< 0.2. The use of antidepressants was correlated with an increased fall risk, with an OR of 14.40 (95% CI: 1.65–125.00; p=0.016). Conversely, comorbid depression was identified as an inhibitory factor, with an OR of 0.04 (95% CI: 0.00–0.63; p=0.022) (Fig. 2). No other significant associations were found for lemborexant and BZD use, comorbid heart failure, or schizophrenia.
DISCUSSION
In this study, the fall rate in patients taking ORAs was 5%. Falls in patients who were administered ORAs with concomitant antidepressant medication were identified as a risk factor for falls, whereas depression was indicated as an inhibitory factor.
Of the concomitant medications, the use of antidepressants was identified as a risk factor for falls. Antidepressants may contribute to falls through several pharmacological mechanisms. These include sedation and reduced attention, which can delay reaction times; orthostatic hypotension, which may cause transient balance impairment; impaired motor coordination and muscle weakness; electrolyte disturbances such as hyponatremia; and cardiovascular effects leading to dizziness or syncope [29,30]. The degree of risk may vary by antidepressant class. Tricyclic antidepressants have strong anticholinergic and sedative properties, conferring a higher fall risk. Selective serotonin reuptake inhibitors and serotonin–noradrenaline reuptake inhibitors are generally less sedating but have been linked to hyponatremia and mild dizziness. Noradrenergic and specific serotonergic antidepressants, such as mirtazapine, can produce marked sedation, although this varies between agents. Antidepressants have various effects, so it is natural to investigate them based on their pharmacological effects. Logistic regression analysis for each mechanism of action was attempted, but the number of cases was small and no significant differences were found for any of the mechanisms of action.
The lack of a significant association between BZD use and falls in the present study was unexpected, as both BZDs and Z-drugs are well-established fall risk factors in older adults [3,4,6]. One possible explanation is that BZD use is widely recognized as a risk for falls, leading to more proactive preventive measures in clinical practice. Moreover, in some patients, the concomitant use of ORAs may have improved sleep quality without promoting excessive daytime muscle relaxation, thereby reducing activity levels during periods of residual sedation. These factors could have mitigated the observed impact of BZDs on fall risk in our population.
By contrast, the risk of falls decreased with depression as a comorbidity. ORAs have been reported to ameliorate depressive symptoms [31,32]. However, our cross-sectional design does not allow us to determine whether ORAs directly improved depression and thereby reduced falls. The observed association may instead reflect differences such as behavioral status in patient characteristics among ORA users.
Furthermore, a divergence in the results was observed between the symptoms and treatment of depression, with antidepressants and depression being risk and inhibitory factors, respectively. Diagnosing depression in older adults is challenging, resulting in low diagnostic rates [33]. Conversely, antidepressant prescriptions remain high among older individuals [34], suggesting a mismatch between the diagnosis rate of depression and the frequency of antidepressant prescriptions. In addition to treating depression, antidepressants are used for pharmacologically managing peripheral neuropathy [35], nocturia [36], and dementia-related apathy [37]. Therefore, antidepressants are frequently used to treat conditions other than depression. In fact, of the 198 eligible patients, 44 had a diagnosis of depression and were treated with antidepressants. On the other hand, 10 depressed patients were not being treated with medication for depression and 7 patients were using antidepressants but had not been diagnosed with depression.
This study had four limitations. First, as a single-center study at an acute-care hospital, the findings may not be generalizable to facilities with different environments or patient populations. In particular, most patients had cerebrovascular disease, delirium, or cognitive decline, reflecting the hospital’s high proportion of central nervous system (CNS)-involved admissions rather than explicit inclusion criteria. Patients without CNS conditions were not excluded but were less frequently admitted, which may limit generalizability. Second, the observed protective association of comorbid depression should be interpreted with caution, as only one patient in the fall group had a depression diagnosis, making the estimate unstable. Further studies with larger samples are needed to confirm this finding. Third, reporting bias could have occurred, as falls were defined based on reported incidents, potentially resulting in an underestimation of minor falls. However, owing to the retrospective nature of this study, unreported falls could not be investigated. A fourth limitation is that falls are subject to several influences that cannot be investigated. It is influenced by staffing on wards and fall prevention measures, but these factors have not been fully investigated. Therefore, identical results may not be obtained in different settings, such as other hospitals or institutions. However, reports investigating the risk of falls in ORAs are scarce, and we believe that this will help to put the risk of falls into perspective.
In conclusion, this study demonstrated that the concomitant use of antidepressants is a risk factor for falls in patients using ORAs, particularly in those with impaired gait and cognitive function. Conversely, comorbid depression appeared to be a factor in preventing falls.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation 
Print 
